Today's Best CPUs Compared... To a Pentium 4

Dr. Damage writes "How do current $74 CPUs compare to the $133 ones? To exclusive $1K Extreme Editions? Interesting questions, but what if you took a five-year-old Pentium 4 at 3.8GHz and pitted it against today's CPUs in a slew of games and other applications? The results are eye-opening." Note that this voluminous comparison is presented over 18 pages with no single-page view in sight.

Comments

P4 pride

[dushkin]

I'm at work, where I have a P4 winXP machine.

AND I'M PROUD OF IT.

Re:P4 pride

[MrNaz]

I bought 5 surplus P4 machines with 512mb ram and 40gb HDDs for my community center's library. They have *CRT* monitors. Beat that!

Re:P4 pride

[Shadow of Eternity]

My gaming setup used to be two computers (pentium4 and a q9450) both hooked up to dual-input FW9012 and P260 trinitron CRTs. That was two computers both running at 3500x1200 and putting a combined weight of about 300lbs on my desk.

We almost didn't need to heat the apartment in winter.

Re:P4 pride

[h00manist]

Almost all our boxes at work are p4. But now I would like to move some of them into some sort of virtual infrastructure rather than upgrade them all.

Re:P4 pride

[ZeRu]

You have my compassion.

Re:P4 pride

[old_kennyp]

Currently typing this on a P4 3GHz machine with 1.5Gb ram. running Ubuntu. I regularly use a P3 Mobile laptop (Dell C400) with XP, recently upgraded it to 512Mb ram & 20Gb hdd. Also have a PII 266 under my desk which runs my Mail / Web server. I think it has 384Mb ram but a 200Mb Hdd. runs Debian Ken

Re:P4 pride

[sznupi]

AthlonXP (of the slower kind - 1700+ / 1.46 GHz) is fine too...as long as one chooses properly written software and keeps the machine clean; having few times more RAM and faster HDD than was common back then also helps greatly. I rarely see a typical, home machine which is more snappy, even though they have few times more processing power - but are almost universally held down by bloat, until quite recently by small RAM and, still, by slow HDDs in case of ever more popular laptops.

Too bad the test didn't include, say, P3 Tualatin 1.4 GHz. Even in times of P4 they were very competitive with much higher clocked Netburst processors. And now...Netburst is long forgotten, software is optimised for C2D-like architectures; which are descendants of Tualatin - I'm curious if it has become even more competitive. ;)

Re:P4 pride

[Anonymous Coward]

How much money did they pay you? Does your library get free electricity?

Re:P4 pride

[muckracer]

> I'm at work, where I have a P4 winXP machine.
> AND I'M PROUD OF IT.

Well, there is no need to be ashamed of the P4 part...

Re:P4 pride

[dushkin]

Our ERP program doesn't run on anything else :(

Re:P4 pride

[Anonymous Coward]

I just upgraded my home computer from a 3ghz p4 northwood to a 3ghz p4 prescott on a slightly-newer motherboard. I have to say, the difference is more than I expected. beforehand, I got fairly serious lag whenever a page opened, say, 5 youtube videos or other flash objects at once. now it handles that much with relative ease.

in a few years time, maybe I'll even upgrade to one of those pentium-D's I've heard so much about.

Re:P4 pride

[Gadget_Guy]

My sister still used the original 386 running Windows 3.1 that I set up for her many years ago. I have thrown out countless computers since then that would blow that system out of the water, but she has no interest in upgrading it.

She just runs a few games and Word 6. I really have no argument to use to convince her to upgrade, because it still does what she wants. There isn't anything that she can't do now. (Obviously she doesn't access the Internet)

Re:P4 pride

[Jimmy King]

I've got a 750Mhz Duron machine with 512MB ram and a 20GB hdd at home right now hooked up to a 15" CRT that I bought in 1995. It's due to be gotten rid of at my next upgrade cycle, though.

I've also still got my dual p3 motherboard with 2x450Mhz CPUs and 256MB ram in it sitting in a box, but it's not in a case or in use in any way. I had a couple of 200Mhz machines around until 2-3 years ago, too, and I managed to keep my family's first computer, a 33Mhz 486DX that we upgraded to a whopping 8MB of RAM and a 100MB hdd up and running until around 2002 or so.

Man, this thread is bringing back some memories of a time long in the past.

Re:P4 pride

[realityimpaired]

I still have a working grey box XT clone in the garage. 10MB hard drive, Hercules Monochrome graphics, 540k of RAM, and an amber scale CRT. I plan on keeping it to show the grandkids... "this was grandma's first computer. they were in the stone age back then!"

Re:P4 pride

[Lumpy]

Problem is my old 3.8ghz P4 with a nvidia 7800 AGP card plays all the games that the quad core monster next to it with a current Nvidia card can. It also rips BluRay and HDDVD disks just as fast as the quad core monster.

Why? most software today is STILL NOT written for multiple cores. the incredibly old P4 I have runs as fast as the new hardware because software still does not take advantage of the extra cores.

DUH, most of us that have been running SMP machines for a long time (My first was a dual processor P-II) have known that the multi-core trend will actually do nothing for computing speeds until software makers pull their heads out of their butts and start using them.

Re:P4 pride

[silent_artichoke]

(Obviously she doesn't access the Internet)

Reading that made my arms itch like a junkie without a fix.

Re:P4 pride

[jridley]

Me, too. In fact, they just bought me a brand new Core2Duo based machine about 3 months ago - I used it for 2 months then gave up and put the old P4 back on the desk. The reason? The new machine had Office 2007 on it, the old, Office 2003. The speed differences between the machines are utterly negligible for what I do (I need Firefox, Outlook, and about a dozen SSH windows - I could be using a 300 MHz Pentium if I wanted to), but having to put up with Office 2007 was costing me real productivity, even after I "got used to it" the workflow for the types of things I was doing was significantly slower.

I'm significantly more productive on the old P4 based machine and 2003 than I was on the Core2 machine with 2007. Both are running XP; I tried 7 for about 6 weeks and again, even after getting used to it, it was still a net loss in productivity for me, so I went back.

Re:P4 pride

[jridley]

I'm using a CRT monitor, and in fact I have turned down offered "upgrades" to LCD several times.

The reason? My CRT is very clear and is 1600x1200. The "upgrades" are 1280x1024. No, sorry.

If they come by with a 1680x1050, I'll probably do it, but thus far they're dropping in whatever the cheapest crap from Dell is that they can get.

Re:P4 pride

[Anonymous Coward]

I have an Atari ST, with a Motorola 6800, 512mb, @7Mhz with a monochrome monitor. Almost identical specs to the original Macintosh.

Re:P4 pride

[tomhudson]

Why "obviously"? Win3.1 + chameleon netsock worked fine for connecting to the net on any 386 with 2 megs of ram, (though it worked better with 4 or 8 megs).

Re:P4 pride

[Fireye]

Wouldn't it make more sense just to load Office 2003 on the C2D machine? It'll be more energy efficient, faster, and has less chance of taking a random nose-dive due to old age.

Re:P4 pride

[Anonymous Coward]

Stick with your normal screen (4x3) as long as you can. I'm running 1600x1200 too - although mine is an LCD panel. I really hate these silly "short screens" (they try to say "wide screen", but looking at word processing documents quickly shows you they are "short screen"). At work I have a 1680x1050 and also a 1600x1200 and at home just the 1600x1200. I'm planning to swap my "primary" (left most) screen at work with the 1600x1200 one. (Initially I had the 1680x1050 in this setup and the 1600x1200 was added later as a second screen - but short screen is so annoying that I want to swap them).

Re:P4 pride

[ElSupreme]

Don't go for anything less than 1920x1200. It isn't worth it. My 19" 1600x1200 died and I was misserable until I could pony up 400$ for a GOOD 24" monitor. I still wish I had that CRT. It was georgeous.

I have a couple of inexpensive (not even the cheap ones) LCD monitors I found for cheap, and they aren't comparable to my good 400$ one.

Re:P4 pride

[Anonymous Coward]

Yeah, but Mosaic would have quite a lot of trouble rendering modern websites. Perhaps she could try elinks...

Re:P4 pride

[michael_cain]

I don't drag it out very often, but still have this one up on the closet shelf. 20 MHz 386SX, separate floating point coprocessor, pre-1.0 Linux, MGR windowing software. Passive matrix 640x480 LCD display, MGR drove it in one-bit mode. I recall doing a large amount of technical work on it -- simulations, 100-page technical reports, etc. Four D-sized nicad batteries lasted about two hours, but it was easy to carry a spare set.

Re:P4 pride

[Vr6dub]

Why not just install 03 on the C2D box?

Re:P4 pride

[Yadyn]

Is yours a Gateway too? Prescott FTW

Re:P4 pride

[BrokenHalo]

they were in the stone age back then!

No they weren't. I still have an XT (I believe the original 8086 came with 512K of RAM, my 8088 has 640), but that machine was neat stuff by comparison with the Burroughs B3700 mainframe I worked with back in the '70s and early '80s. That had only about 320K of core memory, which was surprisingly fast, but the whole machine and its associated air-conditioning used to dim the streetlights in the neighbourhood.

Re:P4 pride

[FreonTrip]

Ripping, I can believe, but encoding should be dramatically faster on the new system. GPU-assisted solutions like Badaboom should grind through high-definition video with some facility, and modern h.264 software encoders (such as Handbrake) are very well-optimized for arbitrary numbers of CPU cores.

That said, I'm in the same camp of keeping multiple computers around. My Athlon X2 5050e with a 9600GT's done fine for all the games I play, and my media center PC with a Radeon 9800 Pro went from bearable to peppy when I replaced its 3.2 GHz Pentium 4 with a Celeron e1500. :)

Re:P4 pride

[hairyfeet]

I've actually been selling quite a few of those lately. The older Athlons and P4s frankly are fine for the work the majority of folks are doing-surfing, watching videos, burning DVDs, and with the economy in the toilet a good 2.6GHz P4 seems to give folks a nice netbox at a very affordable price. The simple fact is most folks don't need the latest Core or Phenom II CPU when all they are doing is going to Facebook and checking their webmail.

Re:P4 pride

[ultranova]

I bought 5 surplus P4 machines with 512mb ram and 40gb HDDs for my community center's library. They have *CRT* monitors. Beat that!

My previous computer, before the current i5-750/Radeon 7550, was a 1GHz AMD Duron with Geforce 2MX. And I still use a CRT monitor, which is fine now that I've finally forced Windows to accept that refresh rate below 85Hz is not acceptable, no matter how badly some games want 60Hz.

Re:P4 pride

[ultranova]

The reason? My CRT is very clear and is 1600x1200. The "upgrades" are 1280x1024. No, sorry.

My 19" CRT is also capable of 1600x1200@85Hz, but only use that resolution for games. My desktop runs at 1154x864, since that's the smallest size text stays readable. Windows's DPI scaling doesn't seem to work well in a single program...

Re:P4 pride

[Anonymous Coward]

The original IBM PC (5150) came with 16K RAM, this minimum config was changed to 48K a bit later on, then to 64K for the last 5150's.

The XT (5160, the one with a 10MB (later 20MB) hard disk) originally could handle 256K, then this changed to 512K, and finally 640K.

You could not install a hard drive in a 5150 without installing an XT power supply (From 110 to 130 watts, IIRC).

Re:P4 pride

[MobileTatsu-NJG]

Did the math co-processor do anything useful for gaming or anything like that? I had a 286 when I was 12 and always wondered if that would have bought me anything.

Re:P4 pride

[Cederic]

My 19" CRT runs 1600x1200 very readably (and has done for almost a decade).

My 17" laptop runs 1920x1200 very readably and I am continually bemused that desktop monitors just aren't as good. I don't want a 24" monitor, I want a sensibly sized one..

Re:P4 pride

[mr_da3m0n]

I have gotten quite nostalgic lately, of my adolescence, the games I used to play, the older, tamer internet back then, the multiple game server services, the inanity of Windows 9x and its song and dance to install drivers, DOS boot disks that could save your life and more.

So what I did is, I pulled out this old IBM machine with a Pentium 200 MMX processor and 64 megabytes of memory I had adopted, saving it from a trash bin in a weird alley behind a bar three years ago, installed Windows 98SE on it, and all my old games which don't run nearly as well in dosbox or NTVDM. I also hunted the backstore of local computer stores to find a 3Dfx Voodoo 3 PCI since I used to be such a 3Dfx fanboy. It took a while but I found one, the exact model I wanted.

A friend of mine built a similar machine, brought it to my place, and we're having old-school, nostalgia-filled LAN parties. Over IPX/SPX.

It really is a ton of fun. Doom, Duke Nukem 3D, Blood, Diablo, Warcraft 2, Take no prisoners, Quake and assorted mods... I'm discovering Diablo 2 ten years too late. And a bunch of games I had always wanted to play over network, which wasn't necessarily possible back then, most of my friends didn't even have modems at that point.

A bit off topic, but if you are to save old machines, try using them for that purpose, it's really fun and makes you feel 14 again.

Re:P4 pride

[commodore64_love]

My *current* computer is a Pentium 4 at 3000 megahertz.

It has Windows XP-SP2 and has worked great for 7+ years, and I don't see any reason to go faster. My brother had one of those new dual-processor AMD X2's and yes it's faster (about 2.5 times), but it's weighed down by Windows/Vista 7 so it ends up running no faster than my machine!

Re:P4 pride

[sznupi]

At least SMP machines can last longer; accidentally, my first SMP one (still lying around somewhere) is also a dual P2. Last time I checked, still quite usable with proper software (in this case - win2k, Opera 9.27, fb2k, miranda, irfanview, mplayer)

Plus now typical new machine doesn't seem to be so ram-starved, which might really enable that longevity.

Re:P4 pride

[sznupi]

Thing with encoding video is, no matter which CPU you use (talking about reasonable choices now, like between cheapest Celeron available currently (which is a 2.5 GHz C2D with 1 MiB L2 and 800 MHz FSB) and Core i7)...it's still too slow. Too slow to be interactive in regards to your workflow. In every case it's eesentially a long batch job around which you must plan your work...unless of course you're doing something very short.

(and GPU solutions are lacking in quality)

Re:P4 pride

[Dogtanian]

My *current* computer is a Pentium 4 at 3000 megahertz.

Pfft... until 5 months ago, my "current" computer was a Pentium 4 at 1.8 MHz. (Actually, I don't know how my year-old Aspire One compared with that, because I never used it for serious work).

Re:P4 pride

[michael_cain]

For gaming, probably not much. I did several modeling and simulation things that needed double-precision floating point, and the co-processor was much faster than the Linux emulation at that time. Many applications that needed just a little floating point faked it with fixed-point calculations, swapping accuracy for speed. As I recall, the big impact on rendering for games came with the introduction of the MMX and more especially the SSE instructions.

Re:P4 pride

[DiLLeMaN]

My *current* computer has a G4 @ 1.25MHz, TOP THA-- oh, wait, wrong architecture.

Re:P4 pride

[ducomputergeek]

I turned the heat down at my old apartment to 60 and my Quad-core Powermac G5 kept it at a comfortable 68 so long as the outside temp wasn't less than 20 degrees.

Re:P4 pride

[trapnest]

My two littlest sisters use a Powermac G3. 400MHz cpu, 512MB SDRAM, OS 9.2. It runs really well for what they do (simple games, typing, etc.)

My older sister uses the hp netbook I got her. Dual core atom, 2GB ram, win 7...

Anyone know a decent browser for PPC OS 9?

Re:P4 pride

[gcountach]

Pfft... until 5 months ago, my "current" computer was a Pentium 4 at 1.8 MHz. (Actually, I don't know how my year-old Aspire One compared with that, because I never used it for serious work).

Only 1.8 MHz? Yes, I think you've won the slowest computer of the century award.

Re:P4 pride

[MrKaos]

My *current* computer has a G4 @ 1.25MHz, TOP THA-- oh, wait, wrong architecture.

After a while I run out of places to store old computer, most of them still work. I threw my first PC away not so long ago. A original IBM PC 8088, with cassette interface and BASIC ROMS, everything still worked, it seems like a waste it ended up on a tip - but who was going to use it? I try to redeploy them in my house or elsewhere like friends, sell what I can or donate assembled systems. A P4 still makes a decent media centre so I guess it all come down to the application.

Games don't use multiple cores?

[damburger]

From the article:

For about the same price as the Core i3-530, the Athlon II X4 635 offers four cores that perform better in applications that rely heavily on multiple threads, such as video encoding, 3D rendering, and Folding@Home. In other uses, such as video games and image processing, these two CPUs perform almost identically. The Athlon II X4 635 leads slightly in overall performance and, as we established on the previous page, in terms of performance value. If that's all you care about when choosing a processor, then your decision has been made.

How can game engines not take advantage of multiple cores? I had no idea this was the case, and find it very surprising given that the PS3 has 7 cores to work with. Are games so lazily programmed that they don't take advantage of that either?

Re:Games don't use multiple cores?

[seifried]

How can game engines not take advantage of multiple cores?

Because not everyone has multiple cores so PC games have to go for some version of the lowest common denominator whereas a console game has a known platform to work with that is standardized.

Re:Games don't use multiple cores?

[zaibazu]

The so called 7 Cores are pretty specialized sub units. With the lack of good middleware and development kits at the PS3 release, the platform is just now after years starting to get somewhat used

Re:Games don't use multiple cores?

[Rhaban]

advantages of multiple cores are not so evident when dealing with real-time physics/rendering/etc.
If all your processes must communicate with each other constantly, you lose the benefits of having each process processed by a different core.

Re:Games don't use multiple cores?

[h00manist]

Are games so lazily programmed that they don't take advantage of that either?

Obviously it's not exactly easy to make programs that can run either on multiple cpu's or a single one just as well.

Re:Games don't use multiple cores?

[Anonymous Coward]

Graphics and game-play come first, then if there's time, optimisation. The average person won't play a game if it doesn't look pretty enough, and just about no one will play a game with abhorrent game-play. Now, there are a few developers who know their shit and can manage to put out something that plays like a dream, is pretty, and isn't resource hog. There aren't enough of them to fulfil people's gamelust so people tend to let those who can't perfect the last one off the hook provided they nail the first two.
Sometimes they have the skill to optimise but not the time; a publisher wants to rush something out the door before its ready in the vain hope of being the hit at some big-buying time in the year. ...and then there are those developers who are just plain lazy. Doing things that I, as nothing more than a bumbling script kiddy, could do more elegantly and efficiently. Developers like that should be shot, but what can you do?

Re:Games don't use multiple cores?

[Anonymous Coward]

Dwarf fortress?

Re:Games don't use multiple cores?

[jamesswift]

Most games these days do in fact use all available cores.
Game developers have been talking openly for a few years now about how they use lockless data structures in multi-threaded engines.

Re:Games don't use multiple cores?

[SharpFang]

Actually, you just make a multi-threaded program and then set specified threads affinity to available CPUs or run them all on one if only one is available. No real difference here. Setting a thread affinity is one library call. Examining the number of cores and rudimentary thread distribution algorithm would be maybe 200 lines of code.

Obviously, it's very difficult to distribute the load *equally* between cores. You can split AI thread from, physics, data preloaders, networking, input handling, audio, CPU-side gfx, scripting engine etc. But each of them has a different load and you can only roughly estimate it beforehand, so using multiple cores optimally is quite hard. Still, using them to gain the upper hand over single-core is quite easy.

Re:Games don't use multiple cores?

[SharpFang]

Goddamnit Dwarf Fortress. It could really, really use multiple cores to handle physics. A good enemy flooding system based on a dam and an artificial lake will hog the fastest CPU.

Re:Games don't use multiple cores?

[dwater]

OpenGL Performer managed to enable applications to run on different platforms, from single CPU, single GPU, all the way up to hundreds of CPUs and upto (IIRC) 16 GPUs, without any changes.

OK, so the developers of OpenGL Performer were clever and motivated, but it certainly proves that it isn't a technical limitation and (IMO) invalidates your assertion that they "have to go for some version of the lowest common denominator".

http://en.wikipedia.org/wiki/OpenGL_Performer

Re:Games don't use multiple cores?

[Anonymous Coward]

Because game simulations are heavily linear (linear in time from frame to frame, and linear in execution in the parts of the engine that make up a frame).

Re:Games don't use multiple cores?

[Anonymous Coward]

At least he's planning on using them in the future. He just has other things to get done first, to get the game into a state worthy of multiple cores.

You can try telling him to do it first, but interrupting Moods often has undesirable consequences, and Toady One is in one hell of a Mood, of one type or another.

Re:Games don't use multiple cores?

[Verunks]

How can game engines not take advantage of multiple cores? I had no idea this was the case, and find it very surprising given that the PS3 has 7 cores to work with. Are games so lazily programmed that they don't take advantage of that either?

this was the case a couple of years ago, nowadays all major games(dragon age, mass effect 2, battlefield bad company 2, etc..) uses my dual core at 100%

the frostbite engine(used in bfbc2 and bf1943) is even designed to use up to 16 threads http://repi.blogspot.com/2009/11/parallel-futures-of-game-engine.html

Re:Games don't use multiple cores?

[Joce640k]

Um, yes ... except Performer was developed on SGI machines which were always multi-CPU (for all of Performer's history anyway). It's got nothing to do with PC gaming.

Re:Games don't use multiple cores?

[blacke4dawn]

You don't seem to get it. The communication between the different threads can easily overshadow the gain of parallel processing in realtime interactive environments compared to just using a single "thread".

Re:Games don't use multiple cores?

[SharpFang]

That doesn't increase difficulty of writing a game that can run either on one or on multiple cores. It's the same difficulty as running it on multiple cores only, which is obviously higher than running it on one core only.

Re:Games don't use multiple cores?

[redstar427]

First, they tested only a few games.
Plus, the video card is far more important with intensive 3D games.

However, some games do take advantage of multiple cores.
I ran Unreal Tournament 3 on a dual quad-core computer, and it used all 8 cores.

Re:Games don't use multiple cores?

[Dunbal]

Because not everyone has multiple cores so PC games have to go for some version of the lowest common denominator

      Which is honestly quite strange, because most games I know require you have the latest uber-$500 graphics card to run properly. I would argue that there is something else involved (eye candy important, multi-core not) in the design process.

Re:Games don't use multiple cores?

[Lord Byron Eee PC]

I don't play a lot of games, but I know for a fact that GTAIV was multi-threaded. And as to why all games aren't multi-threaded, it's because it's hard to do and it's even harder to do right. Video processing and ray-tracing are two areas where multi-threading is a natural choice, but in a game where you've got multiple input and output streams, interacting with several different pieces of hardware, and no tolerance for lags and delays, it is much more difficult.

Re:Games don't use multiple cores?

[Fred_A]

Um, yes ... except Performer was developed on SGI machines which were always multi-CPU (for all of Performer's history anyway). It's got nothing to do with PC gaming.

Performer was early 90s, the day of the MIPS R4000 SGI machines which were single CPU (Crimson, Indigo, O2, Indy even...)

Re:Games don't use multiple cores?

[Anonymous Coward]

and as any pc os can handle multiple threads even on ONE core, it would not matter, as the games would still work.

and the lowest common denominator for even casual gaming systems is dualcore by now anyways.

Re:Games don't use multiple cores?

[TheThiefMaster]

Speaking as a PS3 dev, the SPUs are very different to program for than a normal multi-core cpu (and you only get to use five and a half of them anyway, not 7).

On the flip side, everything based on UE3 (which is most big cpu-hungry multi-platform titles these days) is multithreaded to two or three significant threads: Game, rendering, and possibly physics (depending on physics engine used). None of them are SPU threads (though they may use the SPUs for some tasks), so PS3 performance isn't generally as good as the 360's, but in most games it's a non-issue as both platforms go over the 30 fps cap.

On PC, most UE3 games will run best on two cores, with anything above that being unnecessary.

Re:Games don't use multiple cores?

[ThePhilips]

Are games so lazily programmed that they don't take advantage of that either?

Some games now are multi-threaded.

The problem from perspective of game software is that it has to be near real-time. Synchronizing multiple threads in the time available to render a single frame (e.g. at 25fps that 40ms, or at 40fps - 25ms) is a very tricky task. It is more rewarding to invest into optimizing single-threaded engine, while optimizing multi-threaded variant is quite risky, often with bugs showing up only after the game reaches wide masses.

P.S. Same applied btw to video playback software.

Re:Games don't use multiple cores?

[Rogerborg]

nowadays all major games(dragon age, mass effect 2, battlefield bad company 2, etc..) uses my dual core at 100%

ORLY? 100%, you say? Presumably the games are running SETI@home or similar to eat up cycles while waiting for the GPU or the other core to finish. That's very kind of them.

Re:Games don't use multiple cores?

[dwater]

You beat me too it...of course it was not 'always multi-cpu'.

Furthermore, it is also not true that it has nothing to do with 'PC gaming', since there's nothing to stop the same technology/techniques from being applied to 'PC gaming'.

Re:Games don't use multiple cores?

[kripkenstein]

How can game engines not take advantage of multiple cores?

Several reasons.

• If your game will run on both single-core and multi-core machines, it might not make sense to optimize for the latter, which would likely make the former slower - and the former is already slower, so you care more about it.
• Many games care more about responsiveness than throughput. While you can run N threads on N cores, getting full utilization of your resources, games usually have a loop in which input feeds into the logic system, which feeds into the physics system, which feeds into the rendering system, and the loop starts again. To get from input to viewable output in the same frame - responsiveness - you can't split up the tasks into multiple threads and let them complete whenever they can. Multithreading might make sense within tasks (say, physics of physically separate regions), but not between them - rendering needs for physics to completely finish before it starts.

That said, console games usually are multithreaded. Since they know the target hardware, they can plan how to use the cores exactly. The tests in TFA, however, were not run on console games but on PC games, which as mentioned in the first point, tend to be optimized for the single-core case.

Re:Games don't use multiple cores?

[Anonymous Coward]

Yeah, you're right, it's all really about having multiple threads in your soft. All these deadlocks, stravations and races blahs are just there to frighten kiddies!

Re:Games don't use multiple cores?

[hvm2hvm]

Physics is very friendly to multithreading since most computations are done in parallel anyway. N objects interacting with each other would be simulated in a series of steps, and for each step you need to calculate the next attributes taking into account the previous ones of all the objects. Then, you would save this instance and start again. During each step, threads can more or less operate independent to each other.

A very good example of this would be NVidia PhysX.

Re:Games don't use multiple cores?

[Anonymous Coward]

No, they are using one core to submit commands/jobs to the GPU, while the GPU twiddles its thumbs. Srsly.

Re:Games don't use multiple cores?

[Zironic]

Shouldn't load balancing be the operating systems job?

Re:Games don't use multiple cores?

[Joce640k]

Jeez, the pedants are out in force today...

Let me try again: SGI always had multi CPU machines available for running Performer, and Performer was designed to use those CPUs (it was actually designed for 3 CPUs because that was the average CPU to graphics engine ratio in their machines).

Not many image generators would have used their single CPU desktop machines (which were mostly used for coding/CAD).

PC games not using multiple CPUs is completely separate from SGI machines running Performer.

Re:Games don't use multiple cores?

[Anonymous Coward]

Physics benefits massively from multiple cores - why else do you think people use compute clusters for physical simulations etc? For simple parallelisation techniques in molecular dynamics (replicated data strategies etc) the "processes" are communicating after every timestep.

Honestly, who marked this insightful?

Re:Games don't use multiple cores?

[iainl]

it very surprising given that the PS3 has 7 cores to work with.

Well, firstly you'll remember that just about every discussion of 360 vs. PS3 performance descends into an argument about whether developers are just "lazy" because they don't push half those cores at full power. But secondly, the PS3 really is a weird architecture. Those 7 cores are in addition to a single main CPU core that does most of the work.

Actually, the real question, is why PC programmers aren't making more use of 2-4 cores when the 360 (which a depressingly large number of PC games are ported from) has three symmetric cores doing all the CPU legwork.

Re:Games don't use multiple cores?

[AuMatar]

I'd also question how much is really gained by multithreading here. Gaming is not easily broken down into independent problems- AI, rendering, physics all touch the same data structures. My guess is that multi-threading does some speedup, but not a huge amount due to waiting for data locks. The real advantage of multiple cores is being able to run a browser on a 2nd monitor (or alt tab to it) so you can do more than just game on the thing.

Re:Games don't use multiple cores?

[aaaaaaargh!]

The main reasons:

• Many problems cannot be parallelized at all. If a problem is sequential in nature, multiple cores cannot solve it faster.
• Even when a task can be parallelized, this is at times complicated. Many developers lack the skills to implement or even invent efficient parallel algorithms. It's not just about spawing a few additional threads, there are usually complicated interprocess communication problems involved.
• Since mainstream machines currently may contain everything from 1 to 8 cores (including the virtual ones created by hyperthreading), developing for n cores is always going to involve tradeoffs. The program should still run well on a single core machine.
• Many game engines in use by studios are not yet updated to take full advantage of multiple cores and it is completely non-trivial or too expensive to change them accordingly.

Re:Games don't use multiple cores?

[MrNemesis]

As noted, the PS3 is more of a single core PPC processor plus 6 SSE-on-steroids units. Whilst it's true that parallelism needs to be incorporated into the engine design, the tasks you'd farm out to the SPE's or whatever they're called are very different from what you'd ask core3 to do on your x86 processor.

The CPU in the 360, however, is a genuine triple-core PPC processor.

Re:Games don't use multiple cores?

[Joce640k]

IRIS Performer was designed for multiple CPUs because multiple CPUs were *always* available to it (and perfectly normal in the sort of machines where Performer was used).

Re:Games don't use multiple cores?

[tepples]

Rapid switching among threads on one core can thrash the L1 cache.

Re:Games don't use multiple cores?

[GigaplexNZ]

Na, it's just running an inefficient spinlock while waiting.

Re:Games don't use multiple cores?

[tepples]

So what if you double-buffer large parts of the state of the game world? Have AI, rendering, and physics all look at one read-only copy of the state, and then have physics produce the next frame's state. Coming up with a consistent way to freeze state in this way also helps with the quicksave code.

Re:Games don't use multiple cores?

[Pojut]

Note: This is only my opinion. I have no evidence to back this up.

For the same reason why every channel on TV isn't in HD yet...developers are waiting for the public to catch up with the industry. While most PC games still aren't built from the ground up for multiple cores, they are taking increasing advantage of them. As single core CPUs are phased out (which, except for mobile and extremely-low-budget set ups, they have been), newer games are more and more optimized for it. It's only a matter of time before they are built with multi-core systems in mind. I think 2010 is going to be a big year for improvement in this area, but we won't really see ground-up multi-core designed PC games throughout the industry until 2011.

Re:Games don't use multiple cores?

[dwater]

That's not correct. Multiple CPUs were not always available to it...even if it were 'normal' (which I would question too, since, in plain numbers, there were many more uses on the single-CPU desktops than on the multi-CPU systems).

It was designed to make use of multiple CPUs (and graphics pipes) - that's what we're talking about - but it also allowed applications to work just fine on single CPUs. Applications using it ran on any of the IRIX single CPU systems "just fine".

Re:Games don't use multiple cores?

[Anonymous Coward]

"OpenGL Performer is available for IRIX, Linux, and several versions of Microsoft Windows. Both ANSI C and C++ bindings are available."

So, why can't this be used for PC gaming?

Re:Games don't use multiple cores?

[dcam]

Writing multi-threaded code is hard. Writing high performance, non-buggy multi-threaded code is very hard. Of the order of a magnitude harder.

In addition some things are easily parallelised, eg web servers. They have multiple users and each page that is hit represents a single isolated request, so even just one user accessing a site is easily parallelised.

With games you have one user and things need to happen in a sequence. You code to handle the physics of a bullet trajectory must syncronise with the code handling the AI of the enemies and so forth. This makes it rather hard to do.

Re:Games don't use multiple cores?

[dwater]

> Jeez, the pedants are out in force today...

Well, if you're wrong, you're wrong. Don't blame us since the bit you got wrong is pivotal to the discussion.

> Let me try again: SGI always had multi CPU machines available for running Performer

Well, I'm sure they[1] did since they made the things. Customers, on the other hand, did not. Heck, there were several games for SGI that ran on single CPU systems just fine.

> PC games not using multiple CPUs is completely separate from SGI machines running Performer

Well, again, that's irrelevant. My point is that, if SGI can do it, then "PC games" can too - ie there's no necessity to always code for the lowest common denomitor.

[1] I say 'they', when I should say 'we' - I worked there and my specialty was Onyx2 and OpenGL Performer and I used it mostly on single CPU systems since otherwise I would have to go into a lab.

Re:Games don't use multiple cores?

[Anonymous Coward]

Game engines such as Unreal Engine 3 are basically a huge pile of legacy code. Think Windows, but without the excuse of being 20 years old. UE3 hardly uses the SPUs at all (except insofar as PhysX can use them, and certain other special-case accelerators). UE3 can use 1-2 cores of main processor quite effectively; beyond that it's simply not architected the right way.

Lazy? Nah, EPIC are just too busy with GoW to care about making their game engine good for the developers who are paying $1.5M+ to use it.

OTOH, PhysX and Havok can both use any number of cores to accelerate their calculations. And lesser-known engines such as Gamebryo have much better support for multicores (heterogeneous and homogeneous).

Re:Games don't use multiple cores?

[smallfries]

I'd never heard of it until this thread intrigued me. OMG what kind of insanity is that? Rogue meets simcity. I'm itching to download that and play with it a bit, but I have a deadline bearing down on me that would be destroyed. That developer is the most ambitious game design genius I've seen, and I love the concentration on gameplay over graphics.

Re:Games don't use multiple cores?

[Bakkster]

Yup, multi-core programming is hard, particularly when you don't know how many cores you have. Also, if you make a game with 4 threads instead of 1, there will be overhead with the task switching that will slow it down on a single-core (or It may come as a surprise, but multi-threaded programming isn't as common or easy as the state of the art would imply.

Re:Games don't use multiple cores?

[Rogerborg]

Wait - maybe they're processing future frames. Using one core to predict your inputs...

Re:Games don't use multiple cores?

[Brian Feldman]

Because not everyone has multiple cores so PC games have to go for some version of the lowest common denominator

      Which is honestly quite strange, because most games I know require you have the latest uber-$500 graphics card to run properly. I would argue that there is something else involved (eye candy important, multi-core not) in the design process.

You need your hyperbole license revoked until you can use some semblance of realism. BioShock 2 literally came out less than a week ago and it runs at a full 60fps at 1920x1080, all graphics settings at their highest, on a Radeon 4870 -- a $200 graphics card when I got it a YEAR ago.

Re:Games don't use multiple cores?

[Anonymous Coward]

GTA IV was the first game that really just wouldn't run on a single CPU. Ok well I could get in game but once I was there, the game just wasn't playable. I can handle Crysis/Bioshock 1/2, Any Valve Game, BF2142, BC2, NFS:shift... all run very smooth on low settings at either a 720p or 1024x768 resolution.

The only game that has choked up my single core system almost as much as GTA is Supreme Commander is also super-CPU intensive and it DOES take advantage of multiple cores, but late in the game when the unit counts get high, my system chokes nearly to death, but the game is still playable.

Re:Games don't use multiple cores?

[Eskarel]

Primarily because parallel programming(taking advantage of multiple cores) is substantially more complex and difficult than the more traditional single core variety. It's not a matter of just flipping a switch.

Add to that the fact that multicore systems have only really become common over the last few years and most game engines in current games were begun well before this was the case. At that time dedicating serious resources to multicore was probably considered a waste.

Re:Games don't use multiple cores?

[Tridus]

Which is honestly quite strange, because most games I know require you have the latest uber-$500 graphics card to run properly. I would argue that there is something else involved (eye candy important, multi-core not) in the design process.

This is completely false and has been for years. Most games released today run perfectly well on $200 cards from two years ago. The trend to cross platform games has made it a requirement that things can run on the 360, which has largely frozen PC system requirements. Hell, my wife runs with no problem on a 9800GT, which is a relic by the standard you're trying to use, and it works just fine. There's less thread support because threads are complicated to use correctly in a game, and game developers didn't all stampede into it. Some games have used threads for a while, World of Warcraft is capable of using all four cores on an i7 (whereas Sins of a Solar Empire is effectively a single threaded game, that's just how it was designed). It's an increasing trend, more and more games are using multithreading effectively all the time.

Re:Games don't use multiple cores?

[CppDeveloper]

Writing multi-threaded code will likely result in a significant performance hit on single core cpus - the ones that would need it the most. Thread switching comes at a cost and thread safe coding is usually slower too.

Re:Games don't use multiple cores?

[Threni]

Also, games need to be fast enough on a single core system, and if it's fast enough for that, it'll be fast enough on a multicore system.

Re:Games don't use multiple cores?

[Anonymous Coward]

Er, real-time rendering is embarrassingly parallel. Real-time physics has collision and constraint solving, for which good parallel algorithms exist.

Game AI is a tougher nut to crack, but only because developers are used to having messages received instantly. If you allow time for messages to arrive, game AI is also embarrassingly parallel.

Simulating a parallel world is a highly parallelizable task.

Re:Games don't use multiple cores?

[awall222]

The OS can't load balance if you don't break your code up into enough threads. For example, If you have 6 threads that each take ~10 seconds, sure the OS should handle it. If you instead only break that code up into 2 threads, one taking 50 seconds and one taking 10 seconds, there's not much the OS can do.

Re:Games don't use multiple cores?

[Gr8Apes]

How can game engines not take advantage of multiple cores?

Because not everyone has multiple cores so PC games have to go for some version of the lowest common denominator whereas a console game has a known platform to work with that is standardized.

This is completely irrelevant to multi-threaded coding, the essential difference between a monolithic single threaded game and one that can take advantage of multiple cores.

The main reason games do not do multi-threaded coding historically was that the x86 did horrible context switching, causing a massive performance hit that was large enough to be noticeable. This is still true compared to other architectures, but the overhead is so minor in terms of a CPU's ability it no longer is noticeable. The second reason games are generally not multi-threaded is that parallel programming is not necessarily easy. Even windows still drives much of its IO through a single thread in the GDI component (not sure if this is finally fixed in W7). So it's not just game programmers that find this hard.
 

Re:Games don't use multiple cores?

[Quantumstate]

I would guess a fairly big factor is because generally the game logic which runs of the processor doesn't degrade well. With graphics you can lower resolutions, change texture sizes and add additional lighting effects which are optional so the game just looks a bit worse but plays the same. Trying to do the same with game logic is much harder, maybe some adaptive AI could be made to play better on faster hardware plus some extra graphical effects probably need some extra processor time but these changes would be much less.

Then the developers want the game to be playable on as many machines as is feasible so they head for as low a target as possible which is a single threaded machine. Making it work on multiple cores then turns into a pretty difficult task for very little gain because of the above.

Re:Games don't use multiple cores?

[Nadaka]

Indeed.

Re:Games don't use multiple cores?

[jittles]

I work on flight simulators and we DEPEND on multiple core processors to get everything done at once. What used to take multiple racks of computers can now be done on a single computer with dual quad-core CPUs.

You think IPC is slow on a single machine? Try using reflective memory across multiple computers. Of course we have to handle a bit more than your typical video game since we have to handle hundreds of buttons and switches from multiple crew member stations, night vision, FLIR and day TV cameras, as well as out the window displays.

Re:Games don't use multiple cores?

[asc99c]

I believe you're right - except about timescales. Because you're saying single core CPUs have now been almost phased out - and that's not quite right. They have been almost phased out of new PCs, but in terms of the overall install base of PCs, there are still a lot of single core machines out there. In fact I'd suspect the majority of actively used computers are still single core. I work at a software company and I'd guess about a quarter of the PCs here are still single core.

There are overheads involved in multi-threaded or multi-process programs so on a single core machine it will run slower. And realistically, it's those single core machines that you need it optimised the best for since they're the ones that will really struggle. Optimising games for multi-core processors could well rule out a big chunk of potential market. And so considering the engineering investment needed to do it right, I suspect no one wants to really take the plunge just yet. And even once the engine developers have, it will take a couple more years to filter through to the end game developers. I suspect we'll see a lot of people complaining how hard it is to code for this architecture, in a similar way as has happened for the PS3.

I suspect a majority of games won't be taking full advantage of multi-core processors until 2012 or 2013.

Re:Games don't use multiple cores?

[mdwh2]

Using SMP is far from trivial, including doing so for game engines. Some game engines do, some don't. But I'm not sure it's something to be shocked by, nor is it fair to write it off as "laziness".

The other factor you're missing is that the thing which can most importantly be made parallel - graphics rendering - is already done in parallel on the GPU. The issue here is making things run in parallel, that are still better done by the CPU, not the GPU.

Re:Games don't use multiple cores?

[realityimpaired]

Adding support for multi-cpu/multi-gpu configurations isn't impossible. It's not even that hard. What it is, however, is that this support does add a small amount of overhead. Nowhere near enough to counteract the performance improvement from operating in a multi-cpu environment, but they need to consider the game's performance on the low end systems that some people are still running. Namely: single cpu, single gpu environments.

The reason many games do not support multi-cpu/multi-gpu configurations is specifically because of that overhead. Unfortunately, the market segment of single-cpu/single-gpu configurations is still significant enough that many developpers do not feel that the can block out that group. When the market has matured enough that you simply don't see single-cpu systems any more, and it will happen within the next few years, you'll see games that are properly multi-threaded for these configurations.

Re:Games don't use multiple cores?

[Mashdar]

+1. Rhaban, physics/graphics is one of the MOST parallelizable operations we have. The "shared dataset" is the previous solved set, and no communication is needed so long as the previous set is in shared memory of some sort. The new data should be deterministically determined by the previous set. Graphics processors use this in a non-core-based system where specialized hardware modifies the data set in a pre-determined way massively in parallel.

Re:Games don't use multiple cores?

[Anonymous Coward]

So, why can't this be used for PC gaming?

Because it's not DirectX.

Re:Games don't use multiple cores?

[Andy Dodd]

It depends on the game. In general, newer games DO support multicore CPUs, but only within the past 2-3 years, as until the past 2-3 years, multicore systems weren't nearly as common as they are now.

If you have a single-core system, it is actually possible for the game to run slightly slower if designed for a multicore system. It's also a royal bitch to debug multithreaded applications, and plenty of new and wonderful ways to screw up.

Re:Games don't use multiple cores?

[Andy Dodd]

Similarly, an 8800GT (which I bought around two years ago for under $200) is more than enough for most games on the market. The 9800GT (identical chipset) is under $100 these days I think.

Re:Games don't use multiple cores?

[Anonymous Coward]

We have several PS3 systems at work where we do modeling and simulation prototypes. Sony doesn't provide development kits for linux users so we use good old Vi and spu-gcc. These systems run with all _6_ spus maxed out for days at a time. If you read the documentation this isn't a complicated system to understand. Trying to fit everything in 256kb of local store is challenging, though.

why only 6? b/c sony disables one to increase yields and another makes sure we can't get through the hypervisor.

Re:Games don't use multiple cores?

[Pojut]

Because you're saying single core CPUs have now been almost phased out - and that's not quite right. They have been almost phased out of new PCs, but in terms of the overall install base of PCs, there are still a lot of single core machines out there.

Sorry, should have been more specific...I meant they have been mostly phased out of new PCs. The reason I went with 2011 as a timeline is because of how inexpensive dual cores are becomming. Hell, AMD's top-of-the-line dual core CPU can be had for just over $100, with their quad core CPU's starting at under $100. I think it's a fairly safe assumption that by the end of 2011, just about anyone who plays modern games on their PC would have at least a dual core system. But who knows, with how powerful video cards are getting nowadays (and being able to handle physics processing in hardware), CPUs are becomming less important.

Re:Games don't use multiple cores?

[wintercolby]

Quake 3 had an SMP version that did perform significantly better on MP systems, before multiple core processors where available to retail users. It has been done, just hasn't become popular.

Re:Games don't use multiple cores?

[MikeBabcock]

That's hardly insightful being completely wrong. Physics is one of those tasks that lends itself very well to multi-threading actually.

Its just a completely different way of designing software. Its very hard to find good programmers. Its even harder to find good programmers who are skilled in threaded software design. Just guess how hard it is to find the ones who can debug it :-).

Re:Games don't use multiple cores?

[SQLGuru]

Except games have always had PC requirements printed on the box. Wasn't there a meme about "but can it run Crysis" that epitomized the fact that games push on. Eventually a game comes out that requires the horsepower of _something better_. It's usually the FPS games that push it first because of the need for realistic graphics, fast action, and intense AI (that can still suck, but getting better). RPG's, they aren't quite as resource hungry. RTS, somewhere in the middle. I know it's a catch-22. Game companies want to target the lowest possible machine to give the widest possible install base but they also want to push the envelope as far as possible to get the most PAZOW (which drives sales to a large extent). Finding that balance is hard, but I have to think that giving up on single-core machines at this point is worth more in sales than fighting the performance hurdles to support both modes. Without the PAZOW factor, games tend to get lumped into the "Game X clone" category which hurts sales....even if the clone is better than the original.

Re:Games don't use multiple cores?

[Anonymous Coward]

Many problems cannot be parallelized at all. If a problem is sequential in nature, multiple cores cannot solve it faster.

There are no such problems. They just require exponential amount of cores :)

Re:Games don't use multiple cores?

[Lord Ender]

Left 4 Dead and Team Fortress 2 both have options to use multiple cores. I believe that, when enabled, the other cores to "physics processing." My understanding is that "physics processing" is geek-speak for "making the bodies of your slain foes collapse into realistic piles of death as they hit the ground."

Re:Games don't use multiple cores?

[mdwh2]

Er, real-time rendering is embarrassingly parallel

Yes, and it's embarrassingly parallelised on a GPU. How do multiple cores on a CPU help there?

That's one reason why it's particularly tough - game programmers are already having to juggle a multicore GPU, together with a CPU, even before we worry about running multiple threads on the CPU too.

Re:Games don't use multiple cores?

[jandrese]

The only game you know is Crysis? Most PC games these days are console ports, and even low end GPUs are way more powerful than what you find in the latest generation consoles. I'm using a card from 4 years ago and I still max everything out at 1600x1200 in most games.

Re:Games don't use multiple cores?

[mikael]

They do. But most of the cores will be assigned to producing graphics - the AI / gameplay / physics / networking update will be given their own threads.

Re:Games don't use multiple cores?

[rgviza]

Crysis and lots of other games use multiple cores. The problem is GPUs are the bottleneck. It does no good to have multiple cores if the GPUs can't process the frames as fast as the CPUs can push them. The only game that actually will use 100% of a modern core right now is Crysis. The rest of them easily run on a single core at less than 80% utilization. Get a quad video card set up and you'd see an advantage to a game that used 2 or more cores.

I'm not sure where you got this information from, but it's wrong.

Here's a partial list:
Supreme Commander
Grand Theft Auto 4
Unreal Tournament 3
Flight Simulator X
10tacle: Elevon
Activision/Lionhead: The Movies
Lost Planet
The Last Remnant
Left 4 Dead
ArmA 2
World in Conflict
Crysis
Far Cry 2
Alan Wake
Kingsoft Mission Against Terror
Rybka
Anno 1404: dawn of Discovery
Red Faction: Guerrilla
Operation Flashpoint 2: Dragon Rising
Left For Dead 2
Prototype
Resident Evil 5
Tomb Raider Underworld
Prince of Persia 2008
Shattered Horizon
Left 4 Dead 2
Batman: Arkham Asylum
Aliens vs. Predator
Colin McRae: Dirt 2
Dragon Age

Only problem is when you see a game like this using your cores, it's still dropping frames and your CPUs are at 30-50% because the GPUs just aren't up to snuff yet to take advantage of all this new frame pushing power.

In 2007 the assertion that nothing is using multi core was true. That's all changed, and we've found out that now we need a quad video setup to actually see any advantage to it.

Re:Games don't use multiple cores?

[Anonymous Coward]

Tx clerick for making this list (from Rage3d.com)
This is a list of games that support multicore... forget about the others ...

-- Quad Core --
Alan Wake - Ground up quad core support.
Bioshock (Unreal Engine 3) - Quad core support.
Call Of Duty 4 - Ground up quad core support.
Company of Heroes - Ground up quad core support
Crysis - MP Beta Dual Core support, full game ground up Quad Core support.
DiRT - Ground up quad core support (up to 8 cores reported).
Flight Simulator X - Quad core support with patch.
Lost Planet - Ground up quad core support. (octa core support as well).
MOH: Airborn (Unreal Engine 3) - Ground up quad core support.
Supreme Commander - Ground up quad core support.
The Elder Scrolls: Oblivion - Quad core ground up, can use 5 threads.
World in Conflict - Ground up quad core support.
Unreal Tournament 3 (Unreal Engine 3) - Ground up quad core support.

-- Dual Core --

Age of Empires 3 - Dual core
support.
Call of Duty 2 with 1.01 smp patch - Dual Core support
Enemy Territory: Quake Wars - Native dual core support (possible quad, need confirmation).
EVE online - Dual core (possible quad core, need confirmation)
Falcon 4.0 - Some Support, extent unknown.
Galactic Civilizations II - Dual core support.
Gothic 3 - Dual core support.
HL2: Orangebox - Dual core support
Stalker - Dual core support with 1.0004 patch.
Quake 4 - Dual Core with patch.
Titan quest + Titan quest Immortal Throne - Dual core with patch.
World of Warcraft - Dual Core with patch.

Re:Games don't use multiple cores?

[ThatMegathronDude]

You'll still have performance hits locking the threads that read the read-only state. You can't safely update that information without making sure that its not being read.

Re:Games don't use multiple cores?

[Anonymous Coward]

"How do multiple cores on a CPU help there?"

Well, you can do things like software occlusion and various levels of geometry removal / reduction before the data is sent to the video card.

As far as I know, it's not good form to simply dump as much geometry as you can onto the graphics card. If you have some spare CPU, reducing the working set before it even gets to the video card seems like good idea.

Re:Games don't use multiple cores?

[mdwh2]

Oh sure, I know it's possible to take advantage. But these things no longer fall under the "embarrassingly parallel" algorithms that the OP (you?) claimed AFAIK (I admit I may be wrong - are there embarrasingly parallel algorithms for such things? And would these form a significant proportion of the processing time?)

Re:Games don't use multiple cores?

[AvitarX]

Or, in the case of TF2 at least, comical piles of death.

Re:Games don't use multiple cores?

[Anonymous Coward]

" ... the OP (you?) "

Nope, I'm a different AC.

" ... But these things no longer fall under the "embarrassingly parallel" "

I don't really know myself, but I can imagine situations where you can divide the geometry set up into roughly N chunks (N = number of spare cores) and each core has a rudimentary Z-buffer for geometry removal. You can then combine the Z-buffers from the separate cores (the depth flags etc could still be valid, even though they were generated separately) into a unified result to filter the geometry through before sending the working set to the GPU.

It's might not just processing time, either - memory/bus bandwidth may come into it depending on what you're using and what you're doing. If you have the CPU free, it might be worth the effort but it's obviously dependent on the tradeoff between complexity in the code and what you get for it.

Re:Games don't use multiple cores?

[ElSupreme]

But if you had Radeon 2x 5870s in crossfire I bet you could get upt to 78fps!

Re:Games don't use multiple cores?

[sproketboy]

Is this supported? The link from wikipedia points to the web archive.

http://web.archive.org/web/20071224141002/www.sgi.com/products/software/performer/

Re:Games don't use multiple cores?

[camperdave]

Most games these days do in fact use all available cores.

Yes. You wouldn't believe how much faster Minesweeper is on a quad core.

Re:Games don't use multiple cores?

[dwater]

I don't know, but I'd assume not since the Rackable dropped graphics like a stone fairly soon after they bought the real SGI, and I think OpenGL Performer was dropped quite nicely quite a bit before then.

Re:Games don't use multiple cores?

[91degrees]

The main reason games do not do multi-threaded coding historically was that the x86 did horrible context switching, causing a massive performance hit that was large enough to be noticeable.

Does it? What works better? I know the CPU on the XBox 360 is pretty crappy at this.

Re:Games don't use multiple cores?

[Anonymous Coward]

Actually the biggest thing is you want to have a way to reliably produce results before you go and introduce the unpredictable element stemming from multithreaded applications. Being able to run a single-threadded version next to a multithreaded one and compare input/random seed/output to each other could be quite beneficial in the debugging process.

Re:Games don't use multiple cores?

[bluefoxlucid]

People always say this, but I don't understand the difference between multi-threaded and single-threaded programming. In multi-threaded programming, I have to lock around non-lockless data structures and make sure my code is logically able to complete without waiting for any other code that's holding the same lock. Beyond that, it makes more sense when I have multiple shit going on at once; there's no juggling, I just write my physics engine to track exactly one moving object with the environment and I'm good. Wind? Wind is a force that finds objects in its way and applies a force to them, which in turn causes another thread (controlling their movement) to process different data than with no wind. Lighting is the same way, as is player input (this one's way easy).

The hard part is WRITING A GIANT GAME. It's hard. Writing individual small programs is easy.

Re:Games don't use multiple cores?

[The End Of Days]

You're banging your head against my favorite philosophy - anything someone can't actually do must be easy. It's a bizarre way of thinking that seems to be shared by 99+% of the human race.

Re:Games don't use multiple cores?

[cgenman]

I spent 2,000 dollars on my laptop 3 years ago, including the highest-end graphics card that it would take. BioShock 1 runs at 20 fps with no reflections, AA off, 640x480, medium textures.

I really wish PC developers would take laptop users into account when making games. I'd argue that one reason why WoW has been so successful, is because it runs on hardware that we actually F**ing have.

Re:Games don't use multiple cores?

[BrokenHalo]

The only game you know is Crysis?

I don't even know that. I can remember playing Hunt The Wumpus, and I used to like a mindless Windows 3.x game called Rodent's Revenge... ;-)

Re:Games don't use multiple cores?

[nobodylocalhost]

Multicore programming isn't really that hard due to the fact your OS is suppose to handle what thread goes onto which core for you. All you really need to do is to split a single thread game into 4 threads. And no, you are not running more overhead. Who told you that? Your OS doesn't simply run every task in the same thread. In most cases, you already have multiple background processes running; such as your antivirus, your print/mail spooler, your networking service(such as dhcp client), and your UI. This is the same for most of the modern OSes, your so called "overhead" has already incurred before you even start the game.

The real problem with multi threaded game is syncing. Let's say if you break your game into graphics engine, audio engine, i/o handler, and resource fetcher. Of course resource fetcher has to be the parent thread that spawns everything else as siblings since resource access has to be shared. The real problem is when an event is triggered. not all threads will respond to it immediately due to the fact they are all at different stages of execution. So you end up having to write interrupts for each and every thread.

Re:Games don't use multiple cores?

[RCSInfo]

I would agree that there is still a large base of single core computers, but I think gamers generally upgrade much more frequently than business users. If your curious about the installed base of gaming computers, the best picture I have found is the monthly hardware survey built into Valve's Steam service. Take a look here to check it out. If I was a game developer, this would be the first place I would look.

Re:Games don't use multiple cores?

[Anonymous Coward]

Many developers lack the skills to implement

or T I M E to do it... time pressure is relentless in the game industry..

Re:Games don't use multiple cores?

[GreyWolf3000]

What I don't understand (because I'm not an expert) is that it seems like you could not improve the quality or realism of the physics without shortening the time interval of the "set." Each transition from one step to the next is an inherent parallel bottleneck, since you need to build up the shared set for the next step to use. At some point, won't the simulation will return to being entirely serialized? Perhaps you can get "good enough" results by reasonably small time intervals which are still large enough to realistically solve in parallel?

To come at this from another angle, if a gun from a pistol hits your back, and almost instantly after impact a shotgun burst hits you from the front, there will be a brief period of time where the victim is hurled forward by the pistol blast before being thrust backward by the shotgun.

Although, perhaps for a 30hz frame rate you can always get away with ~30ms time intervals for the simulation and guarantee every 'event' which could be graphically rendered is also physically simulated.

OK, I just answered my own question, but I spent too much time thinking about this not to post it.

Re:Games don't use multiple cores?

[GasparGMSwordsman]

Its never* been a technical limitation. It has always been a time/resource limitation. To add some speed increase for multi-core CPU's you would have to have a developer write and test code. While the developer was doing that, he would not be writing some other code. The result is always a trade off, do you want XZY feature or do you want ABC feature?

My personal preference will always be to add more content or flush out some nifty thing to do in the game rather than make it run faster on some computers.

*By never, I am only referring to times when we have had multi-core hardware.

Re:Games don't use multiple cores?

[Bakkster]

Multicore programming isn't really that hard due to the fact your OS is suppose to handle what thread goes onto which core for you.

I can tell you've never written a multi-core program before...

All you really need to do is to split a single thread game into 4 threads. And no, you are not running more overhead. Who told you that? Your OS doesn't simply run every task in the same thread.

The overhead comes from keeping track of 4 separate threads. Between processor scheduling, task switching, memory/cache thrashing, and message passing alone, a 4-thread sequential program will never run faster than an equivalent single-thread program on a single core.

The only possible benefit is to allow the program to split out tasks with a long wait (such as reading from HDD/optical drive) from those that can be done while waiting, but this gains the exact same speed advantage on a single- or multi-core processor.

The real problem with multi threaded game is syncing. Let's say if you break your game into graphics engine, audio engine, i/o handler, and resource fetcher. Of course resource fetcher has to be the parent thread that spawns everything else as siblings since resource access has to be shared. The real problem is when an event is triggered. not all threads will respond to it immediately due to the fact they are all at different stages of execution. So you end up having to write interrupts for each and every thread.

Again, you've definitely never written a multi-core program. Synchronization is simple compared to the real issues. What are those? Deadlock and Race Conditions. Most developers not being capable of handling these issues (particularly on such large programs) is the reason we don't see more multi-thread programs.

Re:Games don't use multiple cores?

[Joce640k]

...and because after Microsoft screwed SGI over the Fahrenheit project, the EULA for Performer forbid it from ever being used on a Microsoft operating system.

Re:Games don't use multiple cores?

[Mashdar]

The other way to improve on "quality" is to increase the set size, or rather, the resolution of your state. Again, when it comes to displays, anything above your display rate + resolution is meaningless, but if you want something you can scale (time stretch, zoom in on) it could be worthwhile.

But at this point the biggest factor in "realism" is not so much the temporal/spatial resolution so much as the complexity of the governing rules themselves (and the efficient use of polygons and quality of textures). Making a grassy field blow in a breeze does not require incredible color depth, for instance, but an accurate and convincing model of how the grass moves in the real world. Which may require considering all sorts of variables down to grass's resistance/reactance (forgive the electrical terminology) to torsion forces, etc.

Re:Games don't use multiple cores?

[Anonymous Coward]

The main reasons:

• The program should still run well on a single core machine.

i wish thunderbird and firefox did.

Re:Games don't use multiple cores?

[dwater]

> Its never* been a technical limitation

Indeed. That's what I was saying.

Re:Games don't use multiple cores?

[GasparGMSwordsman]

oops, your right. I misread your post. =P

Re:Games don't use multiple cores?

[Fallingcow]

I haven't played it in a few months, but L4D used to have somewhat frequent seconds-long (deadly in a multiplayer game) freezes when both cores were used. Had to cut it down to one to make it playable.

Re:Games don't use multiple cores?

[Cederic]

It's the very best type of insanity. The most insane part is the learning curve - I hope you like killing dwarves, because it's going to happen in many nasty ways before you achieve any kind of stability... :)

Re:Games don't use multiple cores?

[Lord Ender]

Never had freezes on my Intel quad-core system, either in L4D or L4D2.

Re:Games don't use multiple cores?

[DuckDodgers]

Changing an application into a good multi-threaded application can take more work than writing the original single-threaded version. By reputation, most game development companies work at a fast pace with long hours and plenty of overtime just for fixing regular bugs and adding normal game features. Many companies might not have the time or resources to tackle multi-threading at the same time.

I'm not saying it's a bad idea, just that it's not easy and pure laziness does not explain it.

Re:Games don't use multiple cores?

[Fallingcow]

It may have been an AMD thing.

Re:Games don't use multiple cores?

[asc99c]

I've never seen that survey before. It's got some very interesting data. The methodology looks pretty sound for getting average data. I'd suspect Steam users on average are a bit more into the enthusiast camp than the entire market, and so the machines may be slightly more up to date, but it's been around for long enough that it shouldn't be a huge difference.

From this data it does really look as if by the end of this year, single core CPUs will not be that much of a market sector for gamers.

The other point that really surprised me in the Steam data was active gamers by game. Counterstrike is still right up there, so presumably a lot of the single core users are playing older games and therefore not really part of the new sales market.

Hopefully the engine developers are more up to date on hardware trends than me and already have got into gear building proper multi-core optimised engines!

Re:Games don't use multiple cores?

[martin-boundary]

Many problems cannot be parallelized at all. If a problem is sequential in nature, multiple cores cannot solve it faster.

This sounds like common sense that is often repeated but it is false.

Take a simple example of a typically sequential problem: you have a process A which computes a result (true or false, say) and then you have a process B which needs the result of A to do its part. With a single processor, you have to compute A first, then B, and your intuition is correct.

But say you have 3 processors available. Now use processor 1 to compute A. At the same time, use processor 2 to compute B with input true, and use processor 3 to compute B with input false. When all the processes are finished, look up the result of A and choose the result of B that corresponds to A.

So the idea that some problems are inherently sequential is wrong, it depends on how you look at it.

Re:Games don't use multiple cores?

[Anonymous Coward]

You have heard of Crysis right? You know, the one released in November of 2007? How well does your machine play that one? Point is that there are games which do need the latest and greatest hardware (and even occasionally hardware that doesn't yet exist) to run properly.

Re:Games don't use multiple cores?

[theArtificial]

I think this list would've been better sorted with the engines they use. Batman for example uses Unreal 3 etc.

Re:Games don't use multiple cores?

[Gr8Apes]

Pretty much anyone that doesn't drop the FP registers.

Re:Games don't use multiple cores?

[Ailure]

Keep in mind that if you buy a laptop, it should never be for gaming but for portability (yet I still see too many laptop users keeping their computer in one place). Personally I prefer to buy a separate mid-end desktop computer with a mid-end laptop than wasting the money on a high-end laptop. I spent something a bit over 1000 USD on my last computer 3-4 years ago and it was able to run Bioshock with full FPS at max settings for the most part (some slight but acceptable hitches at places, not enough to detract from the game). I would still be using that computer if I just wasn't a nerd itching for new parts. :)

Unless you're going for the lowend, majority of desktop GPU's are quite large and requires their own fan. A laptop needs to be more compact, so there's less space for cooling, and can't clock as high. The components being put tightly together only adds to the heat issue.

Re:Games don't use multiple cores?

[Anonymous Coward]

Try that with a crypto codec running in CBC mode. If you can parallelize that you have broken the crypto :)

Re:Games don't use multiple cores?

[nobodylocalhost]

If you bothered reading your own links, synchronization is DESIGNED to resolve race condition issues.

As for process deadlocking, That is the whole point in utilizing a resource fetcher so you DO NOT access shared resources directly and provide a proper queue for shared resource requests. The only alternative is to literally spam read or write to a shared resource with each and every process until it can verify the data, which also doesn't work very well and may actually create more race condition issues.

As for overhead, like what I have already said, your OS already incurred it. The task scheduler already is tracking multiple separate threads, switching tasks, thrashing your memory and cache. What I said was multi thread programs will not be slower due to these reasons.

Re:Games don't use multiple cores?

[kalirion]

nowadays all major games(dragon age, mass effect 2, battlefield bad company 2, etc..) uses my dual core at 100%

Hell, Mass Effect 1 uses my dual core (C2D E8400) at 100%. Which is annoying since it means I can't play the game at steady 60fps. The difference between 1920x1080 4xAA, all other settings maxed and 800x600 all settings lowest is maybe 5-10 fps.

Re:Games don't use multiple cores?

[martin-boundary]

As you no doubt realize, crypto is difficult because of the huge search space, not because some parts aren't inherently parallelizable. The real problem is Huge/2 = Huge.

Nice try :)

Re:Games don't use multiple cores?

[Anonymous Coward]

Think it through. Every block depends on result of previous and there's (typically) at least 128 bits of state. Beyond possible intra-block optimizations this is about as non-parallelizable as it gets. You can do it if you have 2^128 processors, but then such a cipher isn't of much use, is it ;)

Almost realistic example: No matter how many cores you have, you can't calculate a standard SHA of a gigabyte file under a second. With a tree hash it is possible, but that's not compatible with most current protocols and formats. (lesson: avoid such chaining when developing future standards)

Re:Games don't use multiple cores?

[martin-boundary]

Yes, I understood you the first time :) If a problem is inherently exponential, then parallelism is moot.

(lesson: avoid such chaining when developing future standards)

Well said.

Re:Games don't use multiple cores?

[petermgreen]

I have to lock around non-lockless data structures and make sure my code is logically able to complete without waiting for any other code that's holding the same lock.
The point is not just writing multi-threaded code, it's writing multi-threaded code that actually performs better on a given system (lets say a dual core for now since that's probably the most common configuration right now) than the single threaded code would.

Re:Games don't use multiple cores?

[petermgreen]

That works if you have a significant chunk of the chain that produces a true/false result. If your previous chunk can produce a few billion possible results than unless you have a few billion processors this technique doesn't help you much.

Re:Games don't use multiple cores?

[petermgreen]

I'm not convinced they do anymore. Do many people really have the high end graphics cards needed for modern PC games but only a single core CPU?

Unfortunately i'm not aware of any survey which lets you see how common a given combination of graphics power and processor core count is.

Re:Games don't use multiple cores?

[petermgreen]

I believe you're right - except about timescales. Because you're saying single core CPUs have now been almost phased out - and that's not quite right. They have been almost phased out of new PCs, but in terms of the overall install base of PCs, there are still a lot of single core machines out there. In fact I'd suspect the majority of actively used computers are still single core. I work at a software company and I'd guess about a quarter of the PCs here are still single core.
But afaict gamers tend to replace thier PCs a lot more often than buisness users do. Won't most single core machines be AGP based anyway and hence unable to take the graphics cards modern games need?

Re:Games don't use multiple cores?

[petermgreen]

IIRC it was before multiple core processors were available full-stop, the only way to get two cores in those days was two completely seperate processor chips.

Anyone remember the abit BP6 which was specifically designed for running dual celerons? (I read about it but never actually owned one)

Re:Games don't use multiple cores?

[asc99c]

My approximately 6 year old motherboard has PCI-Express and will still take brand new graphics cards as PCI-E v2 is backwards compatible. That's socket 939, which was primarily for single core chips, although the Athlon X2 was launched towards the end of it's lifetime (I upgraded to an X2 4200+).

But either way, the poster above already convinced me I was wrong with the Steam survey data. I guess it's mainly down to the difficulty of programming part rather than install base.

Re:Games don't use multiple cores?

[Anonymous Coward]

I realized that a bit later when I had already posted that. Real world vs mathematics. "Oh, we just need chained arrow notation here, no biggie" :)

Re:Games don't use multiple cores?

[bluefoxlucid]

Well, the thing is the structure of multi-threaded code makes the "problems" that are "difficult" the hardest part. Like, you have to create data structures that are concurrent but lockless; or lock around them. This to me is a rather simple problem, especially compared to one giant program with a bunch of governing code.

The performance issue you raise is valid. If I lock the world, I'm basically going single-threaded. I have to avoid deadlocks, but yeah. The problem of course is that locking the world for the most part functions just like a single threaded program, except executed out of order (let the OS decide which thread to give CPU time to) based on who calls for CPU time first (since everyone else is waiting). This can make things laggy and cause execution to be uneven.

With multi-threaded code, the hard part is minimizing the locks and coding the engine to avoid locks. That being said, if you do this even remotely correctly it'll always take better advantage of multiple cores; there has to be SOMETHING that doesn't need use of a given locked data structure right now, so you always have concurrency. Shorter locks or data structures that work lockless means better responsiveness though, which is more important than i.e. overall AI or physics throughput. Having half as many moving objects as a hard-limit is better than having objects randomly start and stop moving.

But to me, these are just additional programming problems. They're small, isolated problems that are easier to handle than tuning what is effectively an operating system scheduler.

Re:Games don't use multiple cores?

[petermgreen]

Afaict the idea is every thread does it's work for a given frame, signals the monitor thread that it's done and waits for the monitor thread to signal the next frame.

When all the worker threads signal the monitor thread that they are done the monitor thread swaps the references to the source and destination states and signals the worker threads to begin working on the next frame.

This way you only change the source data structure while the monitor thread is the only thread running. The rest of the time the threads can read it without the need for any locking.

Conclusion

[houghi]

http://techreport.com/articles.x/18448/18 is the page with the conclusion

Re:Conclusion

[Jazzbunny]

Just install AutoPager and you get the article in one long page. You find performance per dollar at page 17 and other interesting nuggets of information well before that last page conclusion.

Re:Conclusion

[C0vardeAn0nim0]

works great. +1 internets for you sir.

mods, do your job. thanks.

Re:Conclusion

[happy_place]

I wanted a chart. What's with all this writing stuff? (shudder!) IT people don't read. If we wanted to read, we'd have girlfriends...

Re:Conclusion

[ServerIrv]

Thanks for that. AutoPager works like a charm.

Re:Conclusion

[crossmr]

performance per dollar is kind of pointless...
unless a lower price unit is actually providing an overall better framerate than a higher priced one.

No one ever says "So what if you're doing better than me, I'm doing it more efficiently!"

Re:Conclusion

[flytown]

its funny to read the long report as if the comic book guy from the simpsons was narrating

Re:Conclusion

[Gaffod]

How is it pointless? It's one of the most important things to consider for me. How else am I supposed to estimate if what I pay for the processor is worth it?

Re:Conclusion

[Gaffod]

Also available for chrome, apparently. Thanks, this is super useful.

Re:Conclusion

[crossmr]

because there are very few situations in which you get any kind of bonus or kudos for having the best performance per $. Its all about top performance overall.

Re:Conclusion

[Gaffod]

No it's not. No matter what processor you buy, unless you buy a new one every 3 months it will not be bleeding edge most of the time. By buying hypothetical more efficient, less powerful processors more frequently than hypothetical inefficient yet more powerful ones, you end up getting similar responsiveness overall with the same budget. That's your bonus. You don't rip yourself off by paying too much.

If you have money to burn and an obsession with 3 digit framerates, I guess what you say is true. In any case, I certainly don't.

Eye-opening?

[spge]

I had a job keeping my eyes open at all, reading that over-long, poorly structured article with no useful conclusion.

Re:Eye-opening?

[Anonymous Coward]

The conclusion I made is that liberal arts majors have no business trying to convey technical information. (Jump to the conclusions page to verify that the author was a liberal arts major.)

Re:Eye-opening?

[LtGordon]

Way too verbose. We want the numbers and a valid, non-emotive conclusion based on said numbers.

Re:Eye-opening?

[Anonymous Coward]

It's not his fault. It's the fault of the original poster here on slashdot... he's the one that misrepresented the article, which is clearly your standard comparison of the top 2 "midrange" cpus' and have them duke it out whilst comparing to all the previously reviewed CPUs, and they threw a P4 in there for a little drama.

Toms hardware ( arguably the "inventor" of the detailed comparison of CPUs and video cards ) always used to do this with Pentium class CPUs back in the day.
It's a "wow" factor device... they take the CPU that most people looking to upgrade probably still have on their desktop. "Seeee?? See how much faster your computer would be?? Buy a new one!"

Readers like it for the wow factor, and advertisers like it because of the take away.

Re:Eye-opening?

[GasparGMSwordsman]

This is the only time on /. where I think tl;dr is an appropriate response...

Re:Eye-opening?

[CheeseTroll]

Which is sad, because if they'd learned anything at all in the liberal arts, it should have been learning to write clearly and concisely. (Unless they studied PoliSci. ;-) )

Re:Eye-opening?

[azgard]

At least, the submitter should have written an ending instead of a cliffhanger. Everybody knows that we don't read TFA here on Slashdot.

I wish there were more such scatter charts...

[SharpFang]

especially for gfx cards (the discrepancies between performance and price are enormous) and hard disks (3D, with price, speed and size)

And the answer is...

[Anonymous Coward]

Pentium 4 is still fast enough for 97% of the applications.

Re:And the answer is...

[SharpFang]

...and the fastest modern CPU is still not fast enough for another 2%.

Re:And the answer is...

[LordLimecat]

Why on earth would you ever consider a P4 over an Athalon?

Re:And the answer is...

[Lord Kano]

Pentium 4 is still fast enough for 97% of the applications.

How much horsepower do you really need to run a web browser and a word processor? In the "business" world, that's all a large portion of them ever do.

LK

Re:And the answer is...

[nxtw]

Pentium 4 is still fast enough for 97% of the applications.

and only uses 5 times as much energy to do the same tasks!

Re:And the answer is...

[HeckRuler]

Do you know how many kilowatts you can buy with $100?
It's about 1000.

The ROI of paying $100 for a new CPU vs using an essentially free P4 is way out there. It's economical to use the P4. It's also eco-friendly unless you consider factory waste in china to be cleaner then a coal plant in America.

Re:And the answer is...

[nxtw]

The ROI of paying $100 for a new CPU vs using an essentially free P4 is way out there.

Except you can buy a significantly faster CPU and one significantly cheaper to operate...

Re:And the answer is...

[elashish14]

And yet, it's far less efficient. It uses the same amount of power for the Cinebench test as an i7-780 (more in fact) and doesn't even finish the test in time! Plus its idle power draw is still more than all others except the i7-900 series (although none were really impressive in this category).

The chips haven't gotten much faster, yes, but they've gotten considerably more efficient. I guess a large part of the reason is smaller fabrication, but I'm sure there are other logical technologies that make it possible too.

Re:And the answer is...

[Datamonstar]

Because you already have one.

Hey!

[Nomaxxx]

I still have a P4 you insensitive clod!

Re:Hey!

[jonadab]

Yeah. In addition to the P4, they should have thrown in a couple of additional historically-significant data points, like the Pentium //, the 486 DX, and the 8088.

What? Why are you looking at me like that?

P4 and MythTV

[Yeechang Lee]

I've been using a Pentium 4 3.0GHz-powered box as a MythTV frontend/backend for more than four years. It often records four high-definition over-the-air or FireWire MPEG-2 streams while playing back another.

For the first three years I used an Nvidia video card with Xv output to play the recordings at very good quality with 50-70% CPU usage. A year ago I moved to VDPAU, which gives me even better playback with under 5% CPU usage, and will do the same with h.264 recordings (generated by the Hauppauge HD-PVR, for example). Thanks to VDPAU, there's every possibility I'll be able to use the Pentium 4 box for another four years.

Re:P4 and MythTV

[vlm]

I've been using a Pentium 4 3.0GHz-powered box as a MythTV frontend/backend for more than four years.

Yeechang fails to mention that is roughly the sweet spot for a mythtv frontend. I have plenty of experience trying to get slower boxes to do myth, which can be done at some difficulty. One GHz Via C7 or whatever its called, with a semi-supported openchrome driver, now that was a challenge, but it eventually worked.

A P4 roughly 3 Gigs with about a gig of ram is enough that its no effort to set up at all. Just set up a plain old linux box and it'll work even with the plain jane VESA driver. Now you can do all this binary NVIDIA driver and XVmc and VDPAU or whatever for even better performance, but it'll "just work" on a stock plain old linux install.

You can spend more money on an even faster system for myth. But its just money down the drain, unless you're doing something totally exotic with high def, or trying to do more than five things at once like Yeechang, or attempting to do dual simultaneous displays, or trying to run a backend on the frontend machine, etc.

Re:P4 and MythTV

[Yeechang Lee]

Yeechang fails to mention that [a P4 3GHz] is roughly the sweet spot for a mythtv frontend.

Yes, it was indeed the sweet spot when I bought it more than four years ago. I certainly wouldn't buy a new P4 today, even if it were possible. I'd get an ION-based Aspire Revo for $200-300; that's clearly today's sweet spot.

My larger point stands; most people wouldn't expect that a box that was state of the art five years ago would still be adequate for recording and playing 1080i and even 1080p high-definition video, but it is.

Just set up a plain old linux box and it'll work even with the plain jane VESA driver. Now you can do all this binary NVIDIA driver and XVmc and VDPAU or whatever for even better performance, but it'll "just work" on a stock plain old linux install.

I am not aware of a single case of successful high-definition video playback with MythTV using Xorg's stock VESA driver, and the folks at mythtv-users would certainly want to hear of one. For high-definition playback some type of hardware acceleration, whether partial (as with Xv) or full (as with VDPAU) is required. Some people are successful with OpenChrome/Unichrome, or Intel video, or even ATi, but the vast, vast, vast majority of MythTV users use Nvidia cards and thus its binary drivers.

You can spend more money on an even faster system for myth. But its just money down the drain

I am happy with my P4 frontend/backend because it meets my needs. Again, though, I'd not choose a P4 if I were building a new MythTV box today. In any case, prices have decreased, not increased; there simply was no equivalent back then to today's inexpensive ION boxes.

Re:P4 and MythTV

[idiotnot]

You can spend more money on an even faster system for myth. But its just money down the drain, unless you're doing something totally exotic with high def, or trying to do more than five things at once like Yeechang, or attempting to do dual simultaneous displays, or trying to run a backend on the frontend machine, etc.

Pretty much true. I have a frontend-only host that does fine with MythTV high-def (1080i OTA). Where it falls hard is playing flash. YouTube HD stuff is passable, but Hulu is an exercise in futility.

Was trying to RTFA, but it appears to be slashdotted. First page doesn't tell much. :-(

Was considering moving to a Conroe-based CPU replacement, hoping that would fix the problems. Might be worth the ~$50 they're running now. Probably would run cooler, at least.

Re:P4 and MythTV

[Anonymous Coward]

Just curious. At what point do the power savings offered by new hardware cover the cost of replacing your P4 rig?

Re:P4 and MythTV

[wvmarle]

I have the same idea. The slowest CPU on the market is way fast enough for almost anything, unless you have very specific needs. The CPU speed issue is solved and done with. I lost interest some 10 years ago, and started to get more interested in what we are actually doing with it: the software that runs on it, and user interfaces.

Re:P4 and MythTV

[Lumpy]

I have a Via C7 motherboard recording HD channels. recording them takes ZERO horsepower. playing them back or transcoding from the huge Mpeg2 stream to a reasonable Mpeg4 is what takes a crapload of horsepower.

Now the HD-PVR you mention is a better solution, BUT does not have a native tuner therefore needs a $10.00 a month rental of a HD cable box to record with. That part sucks.
I want a mpeg2 -> H264 encoding card that is linux compatable. I'd love to offload encoding to hardware that is designed to do it.

Re:P4 and MythTV

[Lumpy]

for playback it's worthless for transcoding jobs.

Honestly, MythTV as a frontend sucks. I use it as a backend recorder ONLY and use xbmc for the frontends elsewhere in the house. Far better media management and a UI that is actually spouse friendly.

I so wish the guys making mythtv would abandon their ugly and hard to use frontend and simply write a plugin for XBMC and call it done.

Re:P4 and MythTV

[Big_Breaker]

With 100 watts of power consumption at ~10 cents a kilowatt hour you would be spending about $88 a year to run your backend 24x7. That doesn't count the extra draw for air conditioning in summer months (the benefit in winter is pretty minor). Different costs per kwh or power consumption scale accordingly. Hopefully your P4 is a northwood and not a prescott! At some point the reduction in power costs will justify a switch to something like the Revo. My total power costs are about $0.30c a kwh (don't get me started!) so I could pay for the switch in a year.

There is a great product called the "Kill-a-watt" that will measure the power consumption of a device simply by plugging it in through the kill-a-watt box. My Q6600 rig draws 120-140 watts for a good fraction of the day as measured by my kill-a-watt. It's a non-trivial cost and a 45nm chip might pay for itself in a year and a half.

Re:P4 and MythTV

[mpapet]

FYI, I'm using a plain HP dual-core desktop with the stock Intel graphics chip as a frontend/backend.

http://www.mythtv.org/wiki/Compaq_DC7700_Small_Form_Factor

I can render and playback HD just fine.

Re:P4 and MythTV

[Yeechang Lee]

At some point the reduction in power costs will justify a switch to something like the Revo.

Correct. I do have a Kill-a-Watt and made those calculations a while ago. This is a key reason why I say I wouldn't start with a Pentium 4 today even if I could buy one new (which I can't).

What is holding me back? 1) Inertia, since my frontend/backend continues to work 24/7 without any issues. As the saying goes, why change what isn't broken? 2) More to the point, I am waiting until the ION platform supports the Advanced 2X deinterlacer. Once it does in a $200 Revo-like form factor--hopefully soon--I'll buy one, but until then I'll stick with the Pentium 4.

Re:P4 and MythTV

[gbjbaanb]

yes, but... I doubts its using 100W 24/7. Also the cost of aircon isn't going to count any extra compared to the benefit during the winter (why would it be). So although the idea of saving power is laudable, it isn't justified by the cost savings.

Re:P4 and MythTV

[tokul]

It often records four high-definition over-the-air or FireWire MPEG-2 streams while playing back another.

You should praise your hardware MPEG encoder card and not CPU.

Re:P4 and MythTV

[Kuruk]

I can spend $88 dollars in a week on beer.

Re:P4 and MythTV

[Big_Breaker]

Me too. I love spending on new gadgets too and if electricity cost savings helps justify the purchase, so be it.

Re:P4 and MythTV

[bingoUV]

My Q6600 rig draws 120-140 watts for a good fraction of the day as measured by my kill-a-watt

Does this include the monitor? Could you post a list all the other components (other than the processor) that this 120-140 watts includes?

thanks

Re:P4 and MythTV

[bingoUV]

Right, I have lost interest in it too. But I/O is the problem nowadays. An overwhelming majority of the lifetimes of my processors is spent waiting for data to come from network/hard disk/USB stuff etc. A long time is also spent waiting for data to come from RAM, I guess the processor goes through hundreds of cycles after requesting data from RAM until getting it.

So I have started to get more interested in I/O speed. I feel it is in general pointless for getting too much interested in software/user interface because I don't have time/expertise/energy to write all the software I use. You can choose/configure software to be efficient, but that's where it stops for me.

Re:P4 and MythTV

[wvmarle]

Why no interest in software as you can not write it all yourself, but interest in hardware that you most likely can not build yourself either? Plugging hardware components together is like cobbling bits and pieces of software together of course.

That said I can program a bit, and like to fool around with Python and Apache/modpython to make interactive web pages. UI is the key here of course. And it's hard to do right, really hard.

And there is a lot gogn on in UI land, outside Apple (lets forget about MS here). Linux distros are doing crazy experimental stuff. I am now using Easy Peasy Linux on my EEE, really cool. It mostly (unfortunately not completely) does away with windows - all is run full screen. No borders, no space lost, no other apps partly visible and taking attention. Seems very suitable as basis for slate type computers as well. And trying out the latest E17 on another box, doesn't make me excited yet but I think there is much more about it than I have seen.

All interesting developments, and while I won't have any influence in it and probably will never make a living with it, it's fun to keep track of.

Re:P4 and MythTV

[bingoUV]

You are right, and in a sense, I cobble together software pieces for efficiency too. Though I seldom write complete programs for my use (except scripts, of course)

But solutions to the I/O problem are not all hardware either. Tinkering with RAID (software), LVM, filesystems, intelligent caching, RAM filesystems etc. make a perceptible difference too.

Anandtech 'Bench' compares ALL recent CPUs...

[distantbody]

And its constantly growing. check it out: http://www.anandtech.com/bench/default.aspx?b=2&c=1

Re:Anandtech 'Bench' compares ALL recent CPUs...

[Anonymous Coward]

It's also interactive.

Re:Anandtech 'Bench' compares ALL recent CPUs...

[nurb432]

You mean all recent ix86 style CPU's.

Re:Anandtech 'Bench' compares ALL recent CPUs...

[jittles]

It doesn't have my Xeon X3220 quad-core on there at all. :( So the only info I have on it is that it scores a 7.2 out of 7.9 possible on the windows experience index. Would be nice to compare it to an i7. Maybe I should overclock...

Re:Anandtech 'Bench' compares ALL recent CPUs...

[Rudeboy777]

The Xeon X3220 *IS* the Core 2 Quad Q6600. Same socket, clockspeed, cache, FSB. The performance results will be identical.

Dual P3 here.

[that this is not und]

I am typing this on an old Dell Precision which sports two P3 processors.

Running a NetBSD smp kernel.

Re:Dual P3 here.

[cbope]

I have an HP Kayak dual-proc workstation at home with two 1GHz PIII's running Ubuntu with 640MB of RDRAM. Great machine. Unfortunately it's not cost-effective to upgrade the memory thanks to Rambus...

Yawn

[Fotograf]

so new CPUs are a bit faster than old tech? Wow...

Re:Yawn

[Anonymous Coward]

Heheh yeah... I started to read the article thinking "Well the new cpu's are probably about 5 or 6 times as fast... happens every 5 years..." and guess what...

news at 11... *big yawn indeed*

I must seem stone age...

[Terminus32]

I'm still using my five year old 2.93GHz & 470MB RAM box, running Xubuntu....can't afford much else at the moment! :-(

"...no single-page view in sight"

[macraig]

There's an easy way to thwart that advertising blackmail for users of Firefox: the AutoPager extension. Antipagination would probably still work for older versions of Firefox.

Re:"...no single-page view in sight"

[Anonymous Coward]

advertising blackmail

"Blackmail"? You do realize that advertising, while perhaps annoying, is the primary driving force behind lots of websites being able to pay their bandwidth and hosting bills. Calling advertising "blackmail" is nonsensical drama.

Oh, I forgot, this is Slashdot, where advertising is treated like leprosy (ignoring the fact that Microsoft advertises on Slashdot).

Re:"...no single-page view in sight"

[Dr.Syshalt]

If you annoy your visitors to the point they start looking for ad-blocking/skipping solutions - there's something wrong with your business model. Either you fix it or you leave the business, it's just that's easy. Whining about your customers not liking some part of your business doesn't make sense. Google was probably the first to realize there is no point in annoying the hell out of surfers - that's why their ads are text-only and quite modest.

Re:"...no single-page view in sight"

[Anonymous Coward]

Which is still a far cry from "blackmail". OP was being overly and unnecessarily dramatic. Annoying ads is nowhere near coercion.

Re:"...no single-page view in sight"

[macraig]

I'm sorry, but breaking that ONE article into SIXTEEN fucking pages, just so they can force people to at least peripherally view sixteen different ads, is pretty damned near coercion. I wouldn't need tools like Autopager if it wasn't for gross excesses like that one. Their extreme behavior begets reciprocally extreme behavior.

Re:"...no single-page view in sight"

[Anonymous Coward]

People getting thrown in modern day political prisoner camps in North Korea is coercion. Some whiny computer geek bitching about being "force" to click thru some web pages because of ZOMG ADS is not coercion. Get some perspective and cut the drama, moron.

Re:"...no single-page view in sight"

[macraig]

Stop trying to shift the specific context out of focus, moron. It's about the article, not its effect on the Universe.

Re:"...no single-page view in sight"

[Joe Snipe]

Chrome + Adblock + FastestChrome = a readable article.

P3 Pride!

[AliasMarlowe]

I still have a P3 working at home - it's a Dell Dimension XPS T450 from about 1998. It came originally with Windows 98, and over the years it has received extra RAM, new graphics, and so forth, so it now boasts 384MB RAM and an ATI Rage Pro, as well as a 20GB disk.

Actually, it's really in semi-retirement, as it's a bit slow for modern applications, but it is still on our LAN and occasionally roused from its grave^Wslumber. At one time, it had Win2000, which it could run OK, but it was a little sluggish running Office2000. Nowadays, it dual boots between Ubuntu/Gnome and PCLinuxOS/KDE, which are about as responsive as Win2000 was. It's fine for most web browsing, IRC, file viewing (graphics, PDF, PS, etc.), text editing, and suchlike. It can handle Gimp and Inkscape once the files being edited aren't too big, and can even run LaTeX well enough, but it sucks rocks trying to run OpenOffice.

Re:P3 Pride!

[Anonymous Coward]

My 2nd PC was a P75 Packard Bell. I continuously upgraded it and it has become a P4 2.8Ghz w/3GB DDR and a 512MB AGP Radeon 3850. It plays Fallout 3 just fine, but my Core 2 laptop kicks its ass with a GeForce 9600.

Re:P3 Pride!

[Corporate Troll]

My parents server is a P-III 800MHz with 768Meg RAM. It used to be my desktop 10 years ago. When I bought it it only had 128Meg RAM. It does run OpenBSD, though.... It's really overkill for what it does: load averages: 0.17, 0.16, 0.18

Re:P3 Pride!

[dcam]

Until very recently I had a number of dell P3 ~900MHz desktops that ran quite happily. I still have one, currently running debian and acting as my fileserver. Been running for ~4 years with no issues. Shortly to be replaced with a VM running on a dedicated VM host.

Re:P3 Pride!

[Anonymous Coward]

Had to be done, P2!: :)

roulette:~# uptime
  15:11:08 up 1219 days, 1:51, 17 users, load average: 0.00, 0.00, 0.00
roulette:~# cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 5
model name : Pentium II (Deschutes)
stepping : 2
cpu MHz : 400.967
cache size : 512 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr
bogomips : 790.52
 
roulette:~# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/hda1 73G 17G 53G 24% /
tmpfs 125M 0 125M 0% /dev/shm
roulette:~#

Re:P3 Pride!

[bloodninja]

...but it sucks rocks trying to run OpenOffice.

Try the Go-OO 3.2 builds. Go-oo is an improved OpenOffice.org, and 3.2 is about twice as fast as OOo 3.0 (5.2 seconds to startup, vs 11.3 for v3.0).

Re:P3 Pride!

[lowrydr310]

My desktop machine is a P3 800 with 256MB RAM, running Win2K. It has a sleek 20" 1600x900 display too!

It's used primarily for light web browsing (chokes on today's flash), Office XP, downloading, and media storage. I actually still use it quite a bit to play Half Life or Medal of Honor online; the TNT2 video card still works great for those applications.

I've been tempted to upgrade so many times, but I have three decent modern laptops in my house, and don't really use the desktop enough to justify it.

Re:P3 Pride!

[nauseum_dot]

I have to add to the P3 pride list: At home, my file server is a Dell Poweredge 1300 with dual P3 500 Mhz CPUs and 768 Mb of ram. It has a software raid 1, from two 80 Gig HDs running Ubuntu 6.06. In my parents basement, I left a Pentium II 300 with 384 Mb and an 8 gig HD running Centos 4 for remote access, ssh tunnelling, etc.

Re:P3 Pride!

[Anonymous Coward]

When my last laptop (now replaced) was stolen at work (less than a year ago), I started carrying my old laptop again. Most of what I did with it was old abandonware games anyway, so my 486sx with 40MB RAM running win98 was just fine.

Drew a lot of odd stares, though :)

Re:P3 Pride!

[Hymer]

It can't be a Pentium III since it arrived early in 1999. You may have a Pentium II system upgraded to Pentium III (we replaced several P II 350-450 with P III 850).

Re:P3 Pride!

[Hatta]

I'm suprised you have such good results. I also have a P-III running at home. It's got dual 450 mhz processors, 512mb ram, and a Geforce 4MX 400 running XFCE. It's servicable for playing xvids and mp3s, but I don't like it for web browsing one bit. Not even Opera seems to render fast enough to be pleasant. I've really been thinking about replacing it with an Atom box.

Re:P3 Pride!

[Anonymous Coward]

A pentium 3 you say, your penis must be very small.

I have a Dell Dimension XPS 266 - Pentium 2 with 512 MB RAM, A Voodoo III 16 MB, an SB Gold 12 MB and 3x 40 GB hard drives running MS-DOS 6.22, Ubuntu Linux 9.10 and Windows XP.

It can be used for web browsing and some office work, but no modern stuff like flash and video etc. I use it mainly for playing old games and determining the minimum system requirements for software I create.

Re:P3 Pride!

[hesaigo999ca]

I remember buying my P3 laptop with 512 mb ram and 68 gb hdd (dual bay) and cd burner with
15@ screen and using it for school to learn programming (at a whopping 5k$ back then...ouch, saw one today on ebay for 100$)

It still competes with all the others when you know how to set up winxp to run smoothly.
I had sqlserver dbs on it, had asp.net websites running, used vs2005, and even have multi boot, with
6 diff OS (those modular bays are great)

I have one instance of win2003 and vs2008 running, to see if it can handle it, and yes there is a small lag, but man, I can still program with the best of them on this thing.

M$ has always forced to go and upgrade hardware and software in a vicious circle.
Problem is, we do not need to, even a P4 which I have 3 of personally, run winxp and are great, some faster then others, and 1 dual core...just to play WoW.

Unless the world ends or someone has broken or stolen my computers, I do not see myself replacing those machines for another 20 years. Why run a multi core processor just to read your emails, or surf the web, it is ridiculous....but get the latest game, and you are screwed into getting a 1gb video card and multi core 2gb ram machine just to see "the cool" graphics.

After WoW, I am not getting any other game, I prefer just sticking with WoW or any of my older games.
As for my machines, I use them as I need them, I even have a p2 laptop running xp just to use for playing music, but it has sqlserver dbs and vb6.0 on it as well.

Re:P3 Pride!

[Larryish]

We have a P3 here that works as a headless file server, it is 900 Mhz with 256MB RAM and contains the "big" hard drive. It also serves as a testing bed for website scripts.

The thing was built out of parts from the "leaning tower of upgrades". We control it via Virtualmin/Webmin.

It ran fine with Debian Etch, went to Lenny last weekend and it still runs great.

My personal machines are P4, a Gateway laptop and a Dell Dimension 2400 desktop, and Ubuntu 8.04 likes them a lot.

Old machines work, very well in fact. They do everything now that they did then, and then they were just fine.

Re:P3 Pride!

[BrokenHalo]

If we're going to get into a pissing match, I've still got a "turbo" XT (8088 processor, with an 8087 math coprocessor) with a Hercules monochrome screen. It even still works; on its HDD there's a copy of WordPerfect 5.1, the Microsoft Fortran and C development suites and an unlicenced copy of AutoCAD. That machine used to really have to work for its living. :-)

Re:P3 Pride!

[AchilleTalon]

Okay, I must admit, I am a sinner. I still own a 486 bought in 1993, running at 25 MHz and having 16MB RAM. And if not enough for you to hang me I did never ever upgraded it, except a huge extra 350MB disk drive. Within this long period of time, the box received a brand new PS fan, since the former exhausted. This box received Linux 0.92 SoftLanding on these old days and many others since then. It is still running Linux 24/7 and is used as a firewall/router without any problems.

Now, I must also admit it is completely unrelated with the original discussion as are all those about the P3 and the likes.

Re:P3 Pride!

[Anonymous Coward]

I still have a P3 working at home - it's a Dell Dimension XPS T450 from about 1998. It came originally with Windows 98, and over the years it has received extra RAM, new graphics, and so forth, so it now boasts 384MB RAM and an ATI Rage Pro, as well as a 20GB disk.

Actually, it's really in semi-retirement, as it's a bit slow for modern applications, but it is still on our LAN and occasionally roused from its grave^Wslumber. At one time, it had Win2000, which it could run OK, but it was a little sluggish running Office2000. Nowadays, it dual boots between Ubuntu/Gnome and PCLinuxOS/KDE, which are about as responsive as Win2000 was. It's fine for most web browsing, IRC, file viewing (graphics, PDF, PS, etc.), text editing, and suchlike. It can handle Gimp and Inkscape once the files being edited aren't too big, and can even run LaTeX well enough, but it sucks rocks trying to run OpenOffice.

Hey I have exactly the same machine as yours. My memory config is also 128MB+256MB. Do you need any parts from it :=)? I recently took out the HD and crashed it just to get the little magnet.

Re:P3 Pride!

[oatworm]

Oh yeah? Well, I have an Apple //e with a Microsoft CP/M Softcard in it. It gives me a grand total of 128k of RAM and two processors (the Apple's 6502 and the Softcard's Zilog Z-80), which is perfect for when I want to run WordStar or work in TurboPascal.

I hate to admit it, but it doesn't get used much these days. :-)

Re:P3 Pride!

[theheadlessrabbit]

...but it sucks rocks trying to run OpenOffice.

A computer can play Crysis with the highest possible settings flawlessly, and still suck rocks trying to run OpenOffice.

Re:P3 Pride!

[The Mighty Buzzard]

Everything sucks rocks (and quite a few other things) trying to run OpenOffice. It makes MSOffice look blazingly fast and I absolutely hate saying anything less than derogatory about MS.

Re:P3 Pride!

[MrCrassic]

I use the same PC as my file server, and it works excellently with Windows Server 2003. I found a newer computer recently, though, so I may decommission it. (Unfortunately, the new computer uses a Celeron processor...kind of sucks, but it's probably much faster than a Pentium 3 running at 450 MHz.)

Re:P3 Pride!

[1729]

I remember buying my P3 laptop with 512 mb ram and 68 gb hdd (dual bay) and cd burner with
15@ screen and using it for school to learn programming (at a whopping 5k$ back then...ouch, saw one today on ebay for 100$)

My first laptop was a 25MHz 486 with a 250MB HD, 4MB of RAM (upgraded to 20MB at some point) and a monochrome VGA screen. It came with Windows 3.1, and I later I installed Slackware on it from a stack of floppy disks.

Re:P3 Pride!

[commodore64_love]

>>>over the years it has received extra RAM, new graphics, and so forth, so it now boasts 384MB RAM and an ATI Rage Pro, as well as a 20GB disk.
>>>

That sounds like my laptop. I replaced the Win98 with an Ubuntu Linux install (actually kubuntu), and it runs very nice now. I can run all the latest software.

Re:P3 Pride!

[iphinome]

Okay I'll play, I have a sparcstation 2 with 32 megs of ram, a 512meg scsi hdd mono framebuffer and a 19" mono monitor. Back when I was younger and more carefree I used to run netbsd on it with qvwm and then run solarir x86 on my PC just to confuse people.

Re:P3 Pride!

[blackpig]

[yorkshire accent]Luxury![/yorkshire accent]

In my shed now...

IBM PC (5150)
64kB RAM + 256kB memory card
Cassette port
2 x 5.25inch double sided(!) floppy drives
CGA card + green screen CRT

Bought second hand in 1988 for AUD$500
including Wordstar and dBaseII

Able to play LeisureSuit Larry and Space Quest

Still boots but the DOS 3.1 floppy is just about worn out.

Re:P3 Pride!

[CAIMLAS]

I have a 700MHz Celeron laptop I still use; it has 386Mb of RAM. It runs OO.org just fine (Windows XP and Linux) - though load time is a bit long (not over 20s).

over the years it has received extra RAM, new graphics, and so forth, so it now boasts...

... a massive percentage of your power bill.

Really, though. The P3 was a good chip. The P4 was smoking silicon slag. I'd rather have a P3 today than a P4 - unless the P4 came with a bunch of RAMBUS memory. That, I'd sell.

There's not much memorable or even appreciable from that era of x86 other than being able to say "I was there", and maybe being able to actually gain a leg up against competitors by using AMD.

Re:P3 Pride!

[Anonymous Coward]

i have a similar box as a home server- with more ram (640 mb, in 3 sticks) and a bigger hard drive (well eventually i'll probably mount up all my obsolete ide drives on it... just for the hell of it). at one point it used to perform admirably well with kde 3.5 (actually about as well as some newer systems). I've been fighting the itch to throw a nvidia fx card and try a modern distro on it... maybe ubuntu lucid ;p

Where's the P4 vs. Modern CPUs conclusion ?

[MasJ]

Isn't this what the article summary gets at ? I couldn't find anywhere in the conclusion how the P4 actually compares to present day processors.
I'm not about to read through 17 pages of all of that just to open my eyes.

Oh, and for CPU comparisons, I usually use:
http://www.cpubenchmark.net/cpu_list.php

It's quite reliable for my choices. I just need everything to boil down to a number these days. Too much choice out there. Was simpler when you could just look at Ghz and know which is better. Now a P7700 and T8600 (examples I just made up..) could be at the same clock speed, be called Core 2 Duo and have totally different performance numbers. Confusing!

Re:Where's the P4 vs. Modern CPUs conclusion ?

[Quirkz]

The P4 appears to be included in every one of the performance benchmarks (or at least on the one performance page I bothered to check on). The headline here is badly skewed. It's a new chip comparison that includes benchmarks for a lot of older chips, including the P4. Not a "how far have we come" article remembering the bygone days of P4 yore. Bad /. headline.

Re:Where's the P4 vs. Modern CPUs conclusion ?

[ColaMan]

I can't remember when I changed my .sig - it was quite a few years ago now. Certainly before computers were measured in GHz.

ZX Spectrum

[Ralz]

My ZX Spectrum doesn't handle Crysis very well, its probably because of the 16K memory. Maybe I should get something newer with more memory, 640K ought to be about enough for anybody.

P4 is quite enough...?

[indre1]

P4 3,2Ghz Northwood @3,6Ghz and a decent graphics card can easilly run Modern Warfare 2 @ 1280x1024 - what else do you need from a processor on a desktop computer?
All these multicores barely give any real advantage to a regular gamer/desktop user at the moment.

Re:P4 is quite enough...?

[Pojut]

All these multicores barely give any real advantage to a regular gamer/desktop user at the moment.

One very significant advantage they provide to people like me is driving a game on my main display while playing back a video on my secondary monitor.

Dragon Age + Aqua Teen Hunger Force = Made of win

Re:P4 is quite enough...?

[thejynxed]

You do realize that overclocking Northwood core CPUs is a bad idea, right?

They have been known to suffer from random heat death, even with water cooling. They also tend to have computational errors and actually suffer worse performance when overclocked. This last bit is very batch dependent though - it really depends on where the chip was manufactured. The heat issue is still valid for every Northwood. There's a good reason most OEMs blocked overclocking in BIOS for their Northwood equipped systems.

Re:P4 is quite enough...?

[FreonTrip]

Yes, this was a widely discussed issue back when the Northwoods were in vogue. "Sudden Northwood Death Syndrome" was caused by electromigration within the CPU, and voltages beyond 1.7V always eventually killed the proc. I won't stand here and say that you shouldn't overclock, but I'd suggest picking up a spare on eBay just in case it fries out...

Re:P4 is quite enough...?

[indre1]

It runs at default 1.55v, so I doubt it will suffer this syndrome. And it has been running like this for years.

Re:P4 is quite enough...?

[Archangel Michael]

Core 1 SVHOST.EXE
Core 2 Antivirus
Core 3 Virus/Malware/Crapware
Core 4 IE8

That is why you need quad cores on XP/VISTA/7

A P4 doesn't cut it any longer, as you'd never be able to run IE8 because the other three processes taking over your life.

Re:P4 is quite enough...?

[Anonymous Coward]

Except for all the new games that require multiple cores.

Re:P4 is quite enough...?

[Anonymous Coward]

Why did this get marked interesting. If you actually look at the benchmarks, you'll see nearly 100fps difference! While this is overkill for most games, I, and most PC gamers I know, can't *stand* to play games under 50fps or so. To get this, everyone ends up turning the settings wayyy down. So sure, you can play something like Modern Warfare 2, but, it'll more look like Quake 2. I guess if you're all about the game input and dynamics, and don't care at all about graphics or immersion, then yes, you're correct. Pffft.

Re:P4 is quite enough...?

[FreonTrip]

In that case, you should be fine. Just make sure that airflow remains good, and that all relevant fans are in proper working order, and keep rocking the old hardware. :)

something missing....

[pjrc]

Did anyone else notice how the Q9550 and Q9650 are absent from this article?

Probably the last thing Intel wants is these previous generation (and attractively priced) chips appearing in the "overall performance per dollar" chart on "Page 17 - The value proposition". Instead, we get a graph where only the i5 and i7 chips appear to perform well beyond any of the older options, but it's a carefully crafted illusion because the faster (and attractively priced) versions of those older chips weren't tested.

Re:something missing....

[Asian Freud]

It includes the older Q9400, and i5-750 beats it quite handily.

An older article comparing Q9650 and i5/i7:

http://techreport.com/articles.x/17545/14

Probably better

[Snaller]

Haven't read TFA but probably better, this dual core crap is slow as hell.

Other factors

[HalfFlat]

The article makes a strong case for the i3-530 and the i5-750, but unlike the comparable AMD processors, they have no support for ECC.

If you're using a computer just for game playing and email, that's fine. On the other hand, if you are doing anything which requires reliability — both in terms of machine stability and the consistency of results and data — ECC is a must. The premium that Intel charge for what should be a standard feature prices them out of the value computing market.

Re:Other factors

[moeinvt]

"AMD processors, they have no support for ECC."

Interesting. I'm an EE, but not a microprocessor architecture guy. Is this ECC for on-board cache memory where the processor is implementing an encoding mechanism between the processor and onboard cache read/write? Does it do something similar between onboard cache and DRAM or external cache, or would that be something implemented at the OS level? I'm guessing that the SER for built in cache has to be ridiculously low (a few per year?). If you can say, what type of applications are you doing that requires this level of robustness?

Re:Other factors

[Timothy Brownawell]

AMD processors and the newer (i3, i5, i7) Intel processors have the memory (DRAM) controller built in. The ECC here is for the DRAM, I have no idea about internal cache. Google released a study a few months ago with various information about actual observed memory error rates... after a bit of crunching on their numbers, I came up with an expected 1/5 chance of a single random bit-flip over a 6-year lifespan, and a 1/3 chance of part of your memory going bad (and causing crashes, corruption, etc, if not caught with ECC) after a couple years.

Re:Other factors

[Anonymous Coward]

This isn't about the machine that runs the LHC, it's about machines that 99% of people use. Thus, no ECC.

Re:Other factors

[SIGBUS]

The ECC support involves the motherboard RAM itself - each DIMM has extra chips to carry the error-correcting information. It's mainly used in servers that run 24x7. Single-bit errors are automatically corrected, and, if they occur, multiple-bit errors are at least detected. The point is of course to keep the server from crashing, or worse, silently corrupting data.

Up until about the mid-1990s, most PCs had parity memory, which provides error detection but not correction. But, in the rush to make things cheaper, computer makers realized that they could shave a few pennies off the cost of a machine by eliminating the parity chips. By doing so, they made it very easy for bad RAM to cause all sorts of hard-to-diagnose problems. Unfortunately, non-parity, non-ECC RAM became the standard, and there are very few places you can buy ECC DIMMs off-the-shelf.

Socket AM2+ and AM3 boards can easily support ECC RAM, but it's up to the motherboard maker to enable it. My recent Asus board for an OpenSolaris box has 4G of ECC memory installed.

Re:Other factors

[Entrope]

To be precise, multi-bit errors are *usually* detected. Any ECC scheme will accept faults that happen to convert stored data from one valid pattern (called a codeword in the literature) and another. They just trade off the likelihood of correctable, detectable-but-not-correctable and undetected faults (according to some model of what causes faults) versus the space and time overhead. The fault origin models are pretty good at matching what most servers see, and the standard ECC schemes are enormously valuable for long-running servers.

Re:Other factors

[petermgreen]

From what I remember a characteristic of DRAM is that the actual raw memory arrays are generally one bit wide. So different bits in the same word will usually end up in very different places on the chip.

This makes the chance of a bitflip or fault hitting multiple bits of the same word very low.

More interesting question: Pentium M vs Atom etc?

[RichiH]

I have a X31 (see http://www.thinkwiki.org/wiki/Category:X31 ) and I am thinking about upgrading to a X100e, X200, X201/X210 -- but I am not sure how my trusty X31 compares to current low-end hardware.

Hard requirements:

* At _least_ 3-4 hours of run time with normal workload (KDE4, konsole, half a dozen ssh sessions, no flash)
* TrackPoint - I hate touchpads
* sturdy - those things are there to be used, not pampered. I don't abuse them needlessly, but I will not go out of my way to make sure the purty purty thing does not get a scratch, either.

My old work computer

[British]

I'm still using a HP zd7000, a P4 laptop from several years ago as my main PC. The battery has long since died, but it's still perfect for general use with the docking station.

I've considered plunking down $300 for a modern laptop, but it never seemed to be an issue. This laptop is still "good enough".

Re:My old work computer

[Anonymous Coward]

Thats nothing, i'm still using a 486 with an awesome turbo button on the front.

Duke Nukem has never run better.

Re:My old work computer

[BitwizeGHC]

P4 laptops tend to suck for portability. I had one. Bitch weighed eight pounds, had two huge fans, and much of the case interior was heatsink.

I gave it to my parents, who use it around the house. Good job they don't have to take it anywhere.

Re:My old work computer

[Anonymous Coward]

A $300 budget laptop isn't what I would consider "modern". Sure, it fits one definition of modern (in that it was built very recently and is for sale now) but a budget laptop is not "characterized by or using the most up-to-date techniques, ideas, or equipment".

Re:My old work computer

[Anonymous Coward]

A $300 laptop is disposable rubbish. A sub $1000 laptop is rubbish. Do yourself a favour and buy something you'll like.

Force Factor

[Barrington21]

I read this post. I totally agree with you......... http://ezinearticles.com/?Force-Factor-Reviews---Dont-Buy-Until-You-Read-This-Review&id=3016296

Love to see a true comparison

[filesiteguy]

I remember all the PC World/PC Magazine/Computer Shopper articles on the Pentium, P-II, P-III and the numbers they threw out. The numbers made sense, given a baseline of a 100 MHz Pentium or even a 66 MHz DX/2.

I would like to see the exact same tests run with these chips. The software may be old - Word 2.0/Photoshop 4.0 - but it should still work.

Mirror

[plaincorgi]

Site appears to be down;Coral mirror: http://techreport.com.nyud.net/articles.x/18448

TLDR

[petes_PoV]

Too Long Didn't Read

Thank you for that link. All I wanted was "the answer", 17 pages of verbiage just to get a 1 paragraph conclusion was just too much

Mod parent up

[turing_m]

If you're using a computer just for game playing and email, that's fine. On the other hand, if you are doing anything which requires reliability -- both in terms of machine stability and the consistency of results and data -- ECC is a must. The premium that Intel charge for what should be a standard feature prices them out of the value computing market.

Yep. Basically if you want to do anything reliable and minimize cost, you need AMD. You also have to take care to get the right motherboard - I've only discovered the utility of ECC now and my Gigabyte motherboard doesn't support ECC. I think Asus generally support ECC. You wouldn't happen to know the sorts or reductions in errors running registered memory brings (compared to just ECC)? If you must run registered as well, it's a comparison between Opterons and Xeons.

If you are concerned about data integrity you might also want to look at an operating system that has ZFS - which means OpenSolaris or FreeBSD, and running mirrored or RAIDZ.

And now for reality.

[Anonymous Coward]

As somebody working in the gaming industry, let me correct you on each of your points.

1) A great many game-related problems can be parallelized quite well. It differs by genre, but most games today could easily split graphics, audio, input processing, game logic and AI into separate threads. Some gaming engines have started to do this. AI is one area that really benefits from multiple threads of execution, so that we can simulate several different outcomes at a time.

2) This was true in the 1970s. We've come a long way since then. From compiler-assisted technology like OpenMP to a variety of higher-level approaches and techniques, multithreaded programming doesn't have to be difficult. Even just making your data immutable, like functional programmers have been trying to teach us for decades, removes many of the IPC woes you mention.

3) This isn't a problem at all. Aside from netbooks, most consumer laptops and virtually all consumer desktops sold since 2006 have had at least two cores. Intel's Core i7 has been out for over a year now, and has seen very good adoption rates. The average number of virtual CPUs (ie. physical, cores or threads) on the average gaming PC today is roughly 2.7. Besides, games shouldn't care how many CPUs are present. They adapt to the available resources. If you have one CPU, we do everything on it. If you have 8, we'll distribute the load appropriately.

4) Where did you hear this from? Again, this was true in 2003, but things have changed a lot since then. Virtually every engine written since then, by a half-decent team, has included mulitprocessor support.

Re:And now for reality.

[cgenman]

Virtually every engine written since then, by a half-decent team, has included mulitprocessor support.

How many PC engines have been written from scratch or properly updated, compared to the overall number of games released in a year? Say that the benchmarks are coming from games that are a year old (2008 holiday season), and those were major titles (3 year dev cycles)... That puts us back to how we were developing games in 2005. I remember a lot of platform panicking around how to get the 360 and PS3 working, how to make DS ports of all code, and laughing at the useless Wii that was never going to go anywhere. I also remember that PC gaming was dying except for the casuals. If you count that the majority of those were licensed engines, and you may very well be looking at code that started in 2003.

I don't know. In the 90's, the PC was the lead platform for a lot of projects. Now, at least in the developer's houses that I've seen, the development effort for the past few years has focused on the more sexy, confusing, and profitable console side of things. I have no doubt that games written and released now are properly multithreaded. But considering the amount of engine licensing going on from old code (not everyone is AAA), and hesitations around investing additional effort in PC versions of projects, I'm not surprised that it has been trickling in more slowly than it should.

Re:And now for reality.

[alan_dershowitz]

If you're involved in the game industry then you should know that just because the technology is out there doesn't mean it's cheap to license or even available for a particular platform. 1970's is pushing it way back. If that was true then it should have been no problem for the Sega Saturn or any other platform in the 90's that could use multiple CPU's, but it was an issue because there was no developer support or people didn't want to program games in FORTRAN to take advantage of multiprogramming. OpenMP goes way back to the 1990's but by 1997 we had exactly one fully compliant C++ compiler, cross-platform C++ was a joke, and half the libraries didn't compile here or there or were not lean enough for or supported features that were not practical on embedded or game systems. Library support in general on game platforms, even the devkits, sucked ass all the way through the playstation 2, in particular at launch.

I remember using a set of macros that let you define blocks of C code that could run in separate threads, but it wasn't cross platform at all. There were other products that were too bloaty or we could not incorporate into a game without violating licensing with either the library or with any of the platforms. Nobody I knew was even using C++.

Basically starting in around 2006 as you said, because the PC platform as you said was starting to become multi-core, was when it started that you had all the libraries you might need, C++ was commonly used on every platform, code was pretty portable. Even now it is totally realistic to license a game engine that is still single-core for your game that predated 2003 in origin, like Unreal Engine. What I am getting at is that now is the multiprogramming time, and it didn't really start becoming practical until we got to the convergence C++ ubiquity, C++/library portability ubiquity, C++ feature ubiquity, multiprocessing library ubuiqity, multiprocessing platform ubiquity, and multiprocessing engine ubiquity (still not there yet but close.)

I am not considering client-server and offloading stuff like network/sound to other threads/processes because if you were on a platofrm that supported that that already had been capable for a long time and that is not the stuff that it killing your performance. On older single CPU systems using a multithreaded engine was just stupid because even if you shunt off something you need to another thread you are just gonna wait around for the result or drop frames if it took too long. So why bother changing the model to something confusing and unnecessary and just put everything in a single-threaded loop. That's the way everybody had been doing it since forever, and there was no reason to give a shit until recently so why spend all the money and time. Well as you said, the time is now, and things are changing. But I think saying things like it's been possible FOR THE GAMING INDUSTRY since the 1970's is just ridiculous. I can't see any reason why anybody should have cared until the last couple of years.

Re:And now for reality.

[Anonymous Coward]

As somebody working in the gaming industry, let me correct you on each of your points.

So you do what exactly? Play-testing?

1) A great many game-related problems can be parallelized quite well. It differs by genre, but most games today could easily split graphics, audio, input processing, game logic and AI into separate threads. Some gaming engines have started to do this. AI is one area that really benefits from multiple threads of execution, so that we can simulate several different outcomes at a time.

Yes, but the UI threads still need to be kept reasonably in sync. Besides, much of that is handled by other on-board hardware. What gaming rig today doesn't have dedicated chips for graphics, audio, networking and input? AI though, is certainly an area that can benefit from multiple cpus.

2) This was true in the 1970s. We've come a long way since then. From compiler-assisted technology like OpenMP to a variety of higher-level approaches and techniques, multithreaded programming doesn't have to be difficult. Even just making your data immutable, like functional programmers have been trying to teach us for decades, removes many of the IPC woes you mention.

See the part above where you mentioned input processing? By definition input is mutable. The input stream is not even known. You can't make all the processed data immutable.

3) This isn't a problem at all. Aside from netbooks, most consumer laptops and virtually all consumer desktops sold since 2006 have had at least two cores. Intel's Core i7 has been out for over a year now, and has seen very good adoption rates. The average number of virtual CPUs (ie. physical, cores or threads) on the average gaming PC today is roughly 2.7. Besides, games shouldn't care how many CPUs are present. They adapt to the available resources. If you have one CPU, we do everything on it. If you have 8, we'll distribute the load appropriately.

3.1: average 2.7 cores/gaming PC. [citation needed].
3.2 How are games going to adapt to the available resources if they don't care how many CPU's are present? If a game is designed to work flawlessly on a single core CPU, what is the point in making it run on multiple cores? It already works. OTOH if a game relies on multiple cores to handle it's processing, what is the developer going to do to make the game fail gracefully when less power is available? Does the game run slower? Is the AI markedly dumber? There comes a point when the game developers are simply going to have to abandon single-core. But while single-core is still a large demographic, games will still have to run on them.

4: Skipped entirely since I'm not a game engine developer.

P2 rulez.

[leuk_he]

I have a working and active PII 350 (beat that 400 Mhz of you) that runs as a games computer for my 7 year old. Games from that era still work well. It is not cot connected to the internet because I bet something will fail when it get beaten by the upgrade crazyness.....

Re:Mod parent up

[Timothy Brownawell]

You wouldn't happen to know the sorts or reductions in errors running registered memory brings (compared to just ECC)? If you must run registered as well, it's a comparison between Opterons and Xeons.

My understanding is that registered memory is less about error correction and more about being able to plug in way too many DIMMs per memory channel, so you don't want it unless you need ridiculous amount of memory.

If you are concerned about data integrity you might also want to look at an operating system that has ZFS - which means OpenSolaris or FreeBSD, and running mirrored or RAIDZ.

Or use Btrfs; ZFS isn't the only option with integrity checks.

Re:Pentium Pride!

[Gr8Apes]

I've got a Pentium Pro 180, retired, sitting next to my desk.

Next to it is a Pentium 166. Upstairs is a Sun Ultra 2.

They're all still functional, but no real reason to keep them going. They will probably head out the door shortly for the recycle bin or whatever I can do to make them not have a huge impact on landfills.

I tossed the 286 a while ago.

Good form sir!

[Anonymous Coward]

Good, intelligent response, that respectfully corrects and didn't flame down the parent; wish we could get more of that on slashdot.

Everything's an Xbox port...

[rsilvergun]

it's 5 years old, so it's no surprise to me that a high end P4 is all you need. Heck, I just bought a $80 GT 240 to go with my 2 year old Athlon and I can play anything on the shelf on high.

Re:Everything's an Xbox port...

[Fallingcow]

Everything's a bad x-box port, though.

Jade Empire, for instance, is unplayably slow on my (high end, 1.5 y.o.) laptop, despite the fact that its graphics look about as good as Quake 2 and I can play plenty of much better looking games (e.g. L4D) at reasonably high detail.

You've gotta have a machine 3x as powerful as it ought to have to be just to make sure you can play all these half-assed ports. It is a dark time for the PC gamer.

Pentium2 333Mhz

[gatkinso]

Runs PuTTy and Xming like a charm, best yet has a real serial port!

I guess this proves I am old, and should probably open up a tackle shop.

Re:More interesting question: Pentium M vs Atom et

[Dystopian Rebel]

I have an X32 and an Atom-based Netbook. Is there something in particular you want to compare?

Re:Mod parent up

[swordgeek]

"Or use Btrfs; ZFS isn't the only option with integrity checks."

Oh yeah, because nothing screams "reliable" like filesystem that is still in beta.

Re:More interesting question: Pentium M vs Atom et

[Anonymous Coward]

According to notebookcheck.net (http://www.notebookcheck.net/Mobile-Processors-Benchmarklist.2436.0.html), pretty much any P-M CPU is going to be faster than almost every Atom CPU. They'll trade blows in benchmarks, but if you've got one of the faster P-Ms (1.6 GHz and up), you'll probably want to jump to a dual-core non-Atom CPU to see a notable performance improvement. If you can find something similarly clocked to your P-M (e.g. actual GHz wise), it'll be an even larger improvement.

An SSD will also give a very noticeable boost to interactive performance. Your X31 is SATA, so there's not as many SSD options as there are SATA, but they're out there. There's even companies using good controllers like the Indilinx Barefoot on PATA SSDs. I've also heard that it should be possible to use an Intel X18-E (which was available in PATA form) in a 1.8" to 2.5" adapter, but I've never heard of anyone actually doing it.

Re:Mod parent up

[Anonymous Coward]

If you are concerned about data integrity you might also want to look at an operating system that has ZFS - which means OpenSolaris or FreeBSD, and running mirrored or RAIDZ.

Or use Btrfs; ZFS isn't the only option with integrity checks.

OpenSolaris makes using ZFS brain-dead easy, especially since it's the default root file system. Further, for those which consider ECC a requirement, both OpenSolaris and Solaris 10 include many other "self-healing" technologies not present in other operating systems.

Emotiveness

[Anonymous Coward]

...non-emotive...

You mean like the: Linux vs. Windows vs. OS X, which is the best desktop OS debate? We all know that debates between computer geeks about the merits of different software or hardware systems are always the very embodiment of non emotiveness.

Mod this down, wrong!

[multimediavt]

If all your processes must communicate with each other constantly, you lose the benefits of having each process processed by a different core.

This statement is just flat wrong, and hardly insightful. The only time this condition is true is if you are dealing with processors *completely isolated* from each other's memory resources. To my knowledge, there is no such beast (cluster or multi-core system) and hasn't been since the days before MPI and OpenMP (or their predecessors) existed. The only bottlenecks in the above quoted situation are latency and bandwidth so that each process CAN communicate simultaneously with any other process, running on any core, tied to the same high-speed bus/network. There are actually other ways to create data parallelism within a system so that even discrete processors can still contribute work toward a larger problem.

I build supercomputers, so I should know!

P4 has power!

[Anonymous Coward]

I would say the P4 and Pentium D series of processors have kept up quite well. I primarily work with CAD (Solidworks and AutoCAD depending on location) and have noticed that the old processors chew through the programs quite well, I have found RAM (or lack there of) and GPU to be much more critical when working with high demand applications.

Work PC:
Pentium 4
1 GB RAM
Crappy GPU
OS: WinXP x32
Solidworks Startup time: Long enough to cross the office and get coffee.

Home PC:
Pentium D Dual Core 64-bit
4 GB RAM
GPU: Nvidia 8600 GT 256mb DDR3 VRAM
OS: Win7 x64
AutoCAD Startup: Near Instant
AutoCAD Inventor Startup: Long enough to get preheated coffee
Solidworks Startup: Near Instant
***Plays EVE Online and COD:WaW Fine

Looking at these systems I think it should be clear that processor makes little difference to CAD users which are considered to be one of the heavier uses of PC's.

Re:P4 has power!

[CompMD]

I've taught CATIA classes for a major aerospace company and worked closely with the company formerly known as UGS. I've been impressed what will run complex CAD programs. Here's a good example:

My cheap CAD box:
CPU: P3-550
RAM: 640MB
OS: WinXP
GPU: nVidia QuadroFX 1000

Runs CATIA V5 effortlessly (but takes about 45 seconds to start), and as long as you aren't working with huge assemblies it works fine. There are no problems making parts, and rendering is fluid. Runs NX4 just fine as well.

Here's a machine that can be a big laggy sometimes:

Dell Precision 530
CPU: 2x Xeon 2.8GHz
RAM: 2GB RAMBUS
GPU: nVidia Quadro4 700XGL

This machine runs NX5 *ok* but nothing spectacular. For simple parts there isn't much difference between it and the P3. And yes, I did use the SMP extensions for the Parasolids engine.

The performance comes mainly from the graphics card.

Atomic replacement of entire game state

[tepples]

You can't safely update that information without making sure that its not being read.

Think of game physics as a function from the current state of the world and the inputs (from the player and the AI) to the next state. Every frame, the physics would not modify the state in place but instead build a new object containing the current state of all objects under the control of physics. Under such a system, which would look familiar to anyone comfortable with Haskell-style functional programming or SQL's COMMIT statement, updates happen by replacing the entire set of positions, orientations, etc. of all objects between the end of one frame and the start of the next using one atomic operation. So you'd need to lock exactly one thing: a reference to this one game state object.

Re:Pentium Pride!

[makapuf]

I have a P4 / 256MB fully functional under my TV as a HTPC.

I tossed (put to recycle) my Amstrad PC2086, 8086 + 20 Gb HDrive (with at least 5% bad sectors ..) a month ago. (SVGA + 14' CRT, no less)

(that and an overclocked abacus)

Re:More interesting question: Pentium M vs Atom et

[RichiH]

Definitely!

What OS do you run?
If Linux, what WM/DM do you use?
If KDE 4, which is faster?

When compiling stuff, which of them is faster?

What is your overall feeling about their relative responsiveness?

Anything else I missed and you deem important :)

Re:P4 and MythTV Details

[mpapet]

Watch those details before you dust off the P4. Parent is using a firewire tuner box.

That means the firewire tuner box is doing all of the rendering then sending the video straight to the P4 hard disk storage. Putting most PCI tuner cards in a P4 and attempting HD rendering generally speaking will not work.

Having used a pre-HD external Hauppauge device, I can attest that it is a very nice device and well supported in Linux/Mythtv.

If one wanted to push things, I'd be interested to hear if a P3 can write an HD stream to disk fast enough to make one usable.

You only have a Pentium 4?

[rrhal]

Ha! I have a Pentium 90! - 2 of them infact. Old DEC servers.

performance per dollar can be significant

[Chirs]

If you've got XXXXXX dollars to build a large-scale number-crunching cluster, performance per dollar is VERY important.

Re:Pentium Pride!

[mj01nir]

I'm still running a Pentium Pro 200 as my firewall. It just keeps chugging along.

It's sitting next to an IBM 5150 that I just can't bear to throw out. It still runs. I have Minix on it.

P1 Pride!

[jmanforever]

I still have an original Pentium 100 MHz box, with 96 MB of RAM, and a 4GB IDE hard drive, running Windows 98SE.

Yes, it IS still in service, although I only turn it on when needed.

I call it my "Antique Media" server. Drives "A", "B", and "E" are all shared on the network, and all I use it for is when I need to get something off from a 1.44M or 720K 3.5" floppy, a 1.2M or 360K 5.25" floppy, or a 100MB Iomega ZIP disk.

Re:P4 and MythTV Details

[colinnwn]

Any digital (ATSC or ClearQAM) tuner on the PCI slot should not take any more horsepower than writing the firewire tuner stream to disk. This is because the digital TV signal is already an encoded MPEG-2 stream that can also be sent pretty much straight to disk also.

A Hauppauge HD-PVR or PVR-x50 should also cause no problems, because they hardware encode a MPEG-2 or 4 stream and present it to your computer. Now if you have an analog tuner with software frame grabber, it will take a lot of horsepower. I am surprised the parent said he could do 1080i without VDPAU.

With VDPAU it would be cake walk for a P4 to write 4 hd streams to disk while watching one. The problem I would worry about is saturating the hard drive controller. You might need to be sure you have 2 disks on different controllers.

Re:Mod parent up

[Anonymous Coward]

Same thing. If you have a lot of memory, like 32 gigs for a database cache on some server, you're going to need a lot of memory. That's not wayy to much memory...that's an amount of memory that's required for that specific application with hardware that was made to hold that much memory the same as your desktop was made to hold 4Gb or whatever. The chances of getting the error for a stick of memory are the same no matter how much memory you put in, it's just, now you have a lot more memory, so the chances of seeing an actual error are multiplied by the number of sticks.

Re:More interesting question: Pentium M vs Atom et

[Anonymous Coward]

I gave my X31 to my better half in favour of an X200s. She loves it as much as I did :-)

The 200s is v.nice, but I can only get 2 hours out of if with the 4-cell battery. Get the 9-cell as an extra and you've got the best of both worlds.

Re:Pentium Pride!

[fifedrum]

I have a sun enterprise 4500 in the garage holding up a stack of paint cans. works great, computes OK too with loads of CPU and several 36GB SCSI drives.

Re:P4 and MythTV Details

[Yeechang Lee]

I am surprised the parent said he could do 1080i without VDPAU.

Playing MPEG-2 high-definition streams (whether from over-the-air or FireWire) is easy. To oversimplify, video playback involves 1) decoding the compressed video signal and 2) rendering, or displaying, it. As mentioned, my Pentium 4 was fast enough to decode MPEG-2 streams in 2005, and the Xv hardware-assisted renderer (usable from Linux via any Nvidia or Intel[1] video card/chipset made in the past many, many years) quite nicely displayed the video with the more-than-decent Bob 2X deinterlacer. The resulting 50-70% CPU usage I saw is perfectly adequate for a box that doesn't do anything else, and of course the usage would be less with a faster CPU. Before VDPAU, software decoding and Xv render is what the vast majority (I'd guess 95%) of MythTV users used for high-definition playback.

Decoding high-definition h.264 video (such as produced by the Hauppauge HD-PVR, which shipped in May 2008) is much more difficult. My Pentium 4 was able to just barely play 720p 6Mbps h.264 recordings, but no more; people on mythtv-users were reporting in mid-2008 that a the fastest Core 2 Duo boxes were just barely adequate to play 13Mbps (the best quality, more or less indistinguishable from the original) HD-PVR recordings, and sometimes were overstretched even then. In other words, MythTV users were beginning to create recordings they could not play back!

VDPAU has the video card handle everything. The card itself, not the CPU, decodes both MPEG-2 and h.264 streams and renders the resulting video using excellent deinterlacers. Given the dilemma that the HD-PVR created, VDPAU could not have arrived later (late 2008/early 2009) than it did.

[1] There's still no adequate Xv support using ATi, from what I understand. I don't know whether current ATi Linux drivers have finally solved this; most sensible people on mythtv-users just throw up their hands and buy a $30 Nvidia card.

12 pages to reach punhline: 5x faster

[peter303]

The gaming benchmarks ran five times faster on the latest multicores.

Re:P4 and MythTV Details

[colinnwn]

My Myth frontend/backend with a Celeron 430 was unable to smoothly play HD MPEG2 recordings from live TV without VDPAU. It would sometimes do ok if it was a previously recorded, but I'd still get the occasional studder. I've also never been able to get any 3 of the HD digital capture cards I've tried set up and recording programs on the correct channel. So my experience with that is limited. I fell back to analog again.

Of course I should have realized VDPAU was deinterlacing. But my PVR-150 recordings frequently have jagged edges in the on screen playback. I hoped playing around with the deinterlacer would fix this, but I guess not. I wonder what could be the cause?

Party like it's 1999!

[wayward_son]

I have a P3 laptop that sees regular use around the house as a portable DVD player/video player.

IBM Thinkpad 600x It's a PIII-500, upgraded to 384MB RAM and a 40GB disk. I found the optional DVD drive online and bought one.

It is currently running Mythbuntu 8.04. I found an X.org configuration file on teh interwebz to get the accelerated driver for the Neomagic 256zx to work properly instead of having it using the crappy default SVGA driver. It decodes DVDs without any problems. I can also play most video files, although not the latest H.264 encoded ones.

I also have an old 802.11b wifi card in it so it can get to the wireless router for filesharing and updates.

My next project is to get NES, SNES, and Genesis emulators up and running.

No, it can't compete with any computer made this century, but it destroys any portable DVD player out there for a fraction of the price.

Here's to efficient computing.

[Anonymous Coward]

If silicon can only discharge so many times, may the more efficient computer patiently reach it's destination in all wisdom of patience.

DEC ~400MHZ 21164 on a 164LX, with ~512MB RAM... fo'shizzle muh 4chan nizzle.

Beautiful rackmount of SGI Oxygen "MIPS" and Sun Sparc nearby...

The most interesting part of the review:

[yourlord]

Is this little jewel on page 14:

Still, although PC hardware gets faster over time, software often gets slower. If you go look at our review from back in the day, the Pentium 4 670 rendered this same scene in 309 seconds using a single thread. Now it's taken over 600 seconds to do it with POV-Ray 3.7. Just to make sure we didn't have a configuration problem, I installed an old version of POV-Ray 3.6.1 64-bit from March, 2005 on our LGA775 test system. Lo and behold, the P4 670 completed the render in about the same time we'd measured way back when.

This to me is the most telling thing in the review. The bloat that has crept into the software made the same cpu take twice as long to render the same scene. This is why we have machines now that make the machines we used 10 years ago look stupid by the numbers, while they don't really offer that much of an improvement in experience due to the incredible amounts of software bloat eating all the extra resources available. This one little paragraph should make the people involved with POVray bow their heads in shame.

Re:The most interesting part of the review:

[Anonymous Coward]

Did you compare the quality of rendering? Also maybe they made a rewrite which is no more optimized for P4 specifically (because P4 is much different from AMD or new intel CPU's).

Re:The most interesting part of the review:

[thickdiick]

What if the software used to test modern CPUs is better-configured to give meaningful differences between very fast modern CPUs. Perhaps the old software would have all modern processors off the charts and wouldn't give meaningful numbers for comparison.

Re:The most interesting part of the review:

[yourlord]

That's not the point. He's not using a program designed to highlight the differences between CPU's. He's using a program designed to render 3D scenes. And that software, doing what it was designed to do, proved to be half as efficient at the same task as the older version. That translates into executing almost double the amount of code, to accomplish the exact same result. That's shameful.

Re:P4 and MythTV Details

[whimmel]

If one wanted to push things, I'd be interested to hear if a P3 can write an HD stream to disk fast enough to make one usable.

I currently have a couple dual-tuner HDHomeRuns and (they say) it streams from each tuner at 15Mbit/sec max via ethernet. I think a P3 backend would have little trouble writing that to disk.

Re:Mod parent up

[Timothy Brownawell]

"Or use Btrfs; ZFS isn't the only option with integrity checks."

Oh yeah, because nothing screams "reliable" like filesystem that is still in beta.

Hey, silent corruption and catastrophic failure are very different risk scenarios!

Re:More interesting question: Pentium M vs Atom et

[icegreentea]

I while ago I was thinking about picking up a netbook versus a used X61. What I found while digging around is that, as a rough guide, any Pentium M of the same clockspeed of a a given Atom chip will kick that Atom's ass in terms of performance. As for your battery life expencancy, my experience with Lenovo Thinkpads (I'm using a T400 right now), is whatever they advertise, lob 10-20% off it and you'll get a fairly representative number. A lot of it has to do with how low you're willing to turn down the backlight. In those terms, the newer thinkpads are really better, cause they're the only ones that have LED backlight option.

Basically, if your current X31 has maxed out RAM and anything but the slowest CPU option, it's at least at parity, if not better than pretty much any Atom powered netbook out there. It also has a much nicer keyboard.

Nothing wrong w/ that (bit late here though)

[Anonymous Coward]

Most of you guys are concentrating on the "gaming aspect" of things, which is "fine & dandy" & all that!

However, I can tell you 1 thing, point-blank here (yes, from "personal/anecdotal experiences" of my own etc.), about doing "actual work related tasks" & where Core I7 CPU's absolutely ROCK vs. the Pentium 4 & my other previous system, a Dual Core AMD X2 6400 4800+:

E.G. -> When I run a program I wrote that is multithreaded (3 threads - coarse designed threading), for removing HOSTS files duplicate entries & more (when I merge in NEW hosts files, those of others that have been updated, or other blocking lists (like Dancho Danchev's of ZDNet, FireEye, WOT (web-of-trust), & many others as I find them (for "the best possible protection" vs. KNOWN bad sites &/or nameservers etc. et al))?

WELL - When I first wrote this program, it was on a Pentium 4 (fastest 3.2ghz unit, H/T enabled), through an AMD X2 6400 4800+ Dual Core, to currently on an Intel Core I7 920 Quad Core... results??

----

P4 = Took it 10 hours to finish!

AMD X2 6400 4800+ = Took it 4 hours to finish

Intel Core I7 920 = Takes it ONLY 1 HOUR TO FINISH...

----

(That tell anyone here anything? It does me...)

I.E.-> My app's ALL about "String Processing" (e.g.-> List sorts, entries comparisons, deletes/adds of new or old entries, duplicate entries removals, format of each line record changes etc. et al)

That type of processing FLOORS this sucker @ times (especially IF I have other apps running, & that 0's my RAM quite often (working set is usually 90mb++ with the size of HOSTS file I use @ 18mb thusfar, & 14% of CPU (&, that's WITH Sleep API calls (2ms each setting, so the system gets back SOME cpu time too, in relation to apps outside of itself) & Delphi Application.ProcessMessages in its loops (a VB DoEvents analog really, for internal message processing)))

NOW - w/ out "time ceding" API calls &/or Pascal methods/functions like those those in its loops???

Heck - I'd wager it'd "rip up" 50% of my CPU easily (because when I set it initially THAT way????

Well - I could not even move its screen, & the CPU 'sailed' up to that on a DualCore (the AMD I noted above))...

IN THE END:

Intel's "done very well" on their I7's, I'll give them that - they are EVERY BIT AS GOOD AS THEY SAID THEY WOULD BE, @ what they said they would be GOOD at: I.E.-> Actual "work related" tasks...

APK

P.S.=> On gaming? Well, I could "keep up with" Doom III &/or Quake 4 "uncapped" & on "nightmare level" (IDSoftware allows this in their games, to break the 60fps framerate cap) on my previous AMD noted above...

However - By way of comparison, on this Core I7 920 though?

Man - NO way: The enemies are just too, Too, TOO F A S T to keep up with on 'mightmare level'...

(So, that's been my experience in gaming with it (which may or may not "hold true" for various games, especially those where the vidcard matters more than the CPU does (from this test, it appears that Call of Duty Modern Warfare appears to be such a game - where the vidcards really make MORE of a diff. than CPU's do, although going from a P4 to a "state-of-the-art/latest-greatest" Intel I7 Core 9x5 does make a pretty big diff. too))...

Ah, nuff said! apk

Re: Today's Best CPUs Compared... To a Pentium 4

[farmpuma]

Another in depth and highly informative review article in Scott 'Damage' Wasson's always enjoyable style.