Slashdot Mirror
Two Approaches to the Next-Generation Desktop
puppetman writes: "Tom's Hardware has a review up of a pre-production P4/2666 using 533 mhz Rambus memory (and shows it stomping the competition). The Pentium 4 needs memory bandwidth, and DDR doesn't supply it. Or does it? Anandtech, ironically, has a preview of the E7500 chipset from Intel - dual channel DDR with support for up to 16 gig of RAM. With a new bus architecture, this looks perfect for high-load databases that need wide pipes to hard-drives, memory, and ethernet. Both of these technologies look great for mid-range database servers.
Anandtech claims that dual DDR200 will provide 3.2 gig/second bandwidth, where Tom claims that DDR266 (single channel) offers only 2.1 gig/second. Intel is sure hedging their bets. I wonder what AMD has up their sleeves."
I am partly curious what kind of OC'ing results you will be able to get out of the 2666 MHz P4 w/ the 533 Mhz RDRAM, I would like to see it't benchmarks compared to the OC'd 2200 (to 3760 MHz) w/ slower FSB that was posted not so long ago.
Saying Java is nice because it works on all OS's is like saying that anal sex is nice because it works on all genders.
The Pentium 4 needs memory bandwidth, and DDR doesn't supply it.
Do *users* need this memory bandwidth or does the proverbial Quake benchmark need it?
Show me "desktop" (as the headline implies) application that requires this. Even the most cutting edge 3D games don't use current 3D processors to their potential, these days.
Life is the leading cause of death in America.
But you know there is at least one photoshop filter that would run faster on a 1ghz g4 and i'm sure we'll see steve jobs demonstrating it at macworld 2004 proving that macs are still twice as fast. :)
yes i run a goth/punk/emo porn site.
The vast majority of systems that are being sold today are somewhere around the 1Ghz mark. They represent the "sweet spot" on the price/performance curve, and quite frankly, users just don't need anything better. Open source OS users, such as most of us here, don't need to ratchet up the speed to 1.5Ghz unless they're running a bleeding edge release of the bloated KDE 2. Windows XP runs just great (well, as well as Windows XP can run, anyway ;) on my Duron 900.
Desktop users don't need anything faster than 1Ghz. So what's Intel's brilliant strategy? Why, they're going to develop chips that are even faster than the overpriced 2Ghz P4s they're having difficulties unloading right now.
And that, my friends, is why AMD is well on its way to winning the war. Intel is putting a product on the market without bothering to notice that nobody needs anything faster. They will lose a lot of money doing this (a friend at Intel pegged the development costs for this chip at $3.7 billion). AMD is sitting tight and refining their core business: solid, stable, speedy, and inexpensive chips that consumers can afford and that consumers actually want to buy.
If I were a stock broker, I would be telling all of my clients to short Intel and go long on AMD right about now. The revolution is underway and the underdog is winning.
Mr. Uptime
Free Open Source Naked Ladies!
Give me a desktop with no fan, lots of pixels and video RAM, and a reasonable-sized disk and a CD-burner. In a small case. And put the disk in one of those removable-drive drawers so it's easy to replace. If it needs more than 500 MHz, it belongs on the server in the back room. Desktops are for running X (or VNC if you don't have a real OS), and doing light development, and running MP3s. If I need to have a dedicated machine to do development on instead of a shared environment, (which I don't), it almost certainly needs to be a slower machine to emulate a random customer.
Actually, my current desktop is a laptop running Win98. There's never enough RAM, and often not enough disk, but the 450MHz CPU is almost always fast enough.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
To access data in a Rambus module, the request must pass through all modules in sequence up to the module that has the data and then must pass back through those modules to deliver the data to the northbridge. This is, BTW, why continuity RIMMs are required.
As one can derive, this greatly increases latency as the number of modules increases. Servers, being systems that generally have lots of RAM, often have at least 8 modules available.
Due to this increased latency as a function of the number of modules (and other factors), Rambus is therefore poor memory for servers.
Note that this is per channel, meaning a dual channel Rambus system with eight modules has the memory latency of a four module system because the modules are split between the Rambus channels.
Computer Science is no more about computers than astronomy is about telescopes. --E. W. Dijkstra
>What exactly am I supposed to do with a machine like that? I develop Java software. My IDE, app server and build scripts each open their own JVM instance. I really haven't seen any performance problems with a 450mhz with 512MB ram.
:), just ask any hobbyist 3d animator for example, and no, buying a lot of cheap machines to do a renderfarm doesn't always cut it, at least not when you want to preview some effects like volumetrics before sending them to a final render.
---
I'm doing 3D animation, the fastest, the less hours I spend waiting for my renders to come out. That's ONE application... it's not because you're still playing tradewars in ascii on an XT that some other people won't benefit from advances in technologies.
In your everyday life, other technologies benefit from it, CAD benefit from it, movies studio benefit from more power, Science, etc. I can't beleive some people are SO much self-centered that they pull out comments like this (neither moderators modding the parent up), I mean, if you have the IQ to come here and read the articles, how can you think like that?
Granted, these changes are kinda pointless for most people, after 1GHZ cpu and a geforce2, you don't need much more to enjoy what most end user technologies have to offer, but there are still DESKTOP users out there that enjoys powerfull machines for other things than showing off
$0.02
--- Metamoderating abusive downgraders since my 300th post.
The next-generation desktop which I'm thinking of doesn't need a single linuxkernel-in-less-than-one-minute-building numbercruncher. I would like to have a seamless multi-host cross-platform desktop, shared among e. g. a Sun running Solaris, a GNU/Linux workstation, a PDA, some recycled underpowered P100-class machines, an Apple Macintosh, maybe even a (ugh) w1nd0ze box. All of them would run different operating systems on many kinds of hardware.
;-)
A modern desktop environment is built on many layers, lots of processes and daemons, many interfaces and abstractions, most of which could be delegated to and shared among other hosts. Poor performance? No need to throw away the old box, just add a new one. With open and interopable interfaces like X11, CORBA, XML, HTTP or whatever, a next-generation desktop of this kind should be possible, especially with Free software.
In my view the most promising solution towards this concept is the GNU Network Object Model Environment (GNOME), largely based on CORBA, using only a few remaining locks which are likely to disappear within the next few years. If finally a common object model between GNOME, KDE, GNUstep and other backends can be established, the seamless multi-host cross-platform desktop could become reality.
The 2.6 GHz machine could then be used to build SETI packages and Linux kernels to heat up the office
With all of the posts saying that our 1GHz's are fast enough, I say until Quake n looks like Final Fantasy (the movie!) we don't have fast enough CPU,RAM,Video,[Insert Bottleneck Here].
There is no longer anything that can be done with computers that is nontrivial and clearly legal. -- Paul Phillips
Of course a not-yet-released equivalent of an overclocked P4 is going to beat the competition vs. AMD's AthlonXP which is out and available NOW.
I would like to note that while the P4 did pounce the AthlonXP, take a look at the numbers (and i'm not talking about price, as I don't even want to know how much that P4 will cost!)
AthlonXP 2000+ runs at 1,666MHz at a bus which is the equivalent of 266MHz.
The P4 is running at 2666MHz (a full Gigahertz higher frequency) with a bus at the equivalent of 533MHz.
The (essentiually overclocked) Pentium 4 has a full SIXTY PERCENT CPU clockspeed advantage and a ONE HUNDRED PERCENT front side bus (FSB) advantage, yet look at its real-world performance:
MP3 encoding: 6.2% faster than the Athlon. (woop)
DivX encoding: 30% (note that the program is highly optimized, by Intel themselves, for the P4. How many programmers have an Intel engineer handy?)
Xinema 4D: 12.8%
3DMark 2001: 4.9%
Note that that Lightwave was not included--the only common test that runs faster on the P4 is the raytracing test. Guess which one Tom's Hardware used?
I just thought I'd point out that the only conclusion that you can really draw from these tests is that, as many in the hardware community know, the P4's architecture is designed for high clockspeed, with zero regard to actual real-world performance. Which matters more to you?
Computer Science is no more about computers than astronomy is about telescopes. --E. W. Dijkstra
Don't be too impressed with the numbers of the Lightwave rendering benchmark, the scene used is heavily Radiosity-based, which Newtek (makers of lightwave) publicly said that was SSE2-optimized, if they'd run the same application benchmark but with any other math-intensive scenes like raytrace, etc.. the gap wouldn't be that impressive. I use Dual Xenon and Dual MPs at work, I've noticed the difference, and Tom being tom, he still goes on doing flawed benchmarks (flawed because he doesn't mention that little fact even if a lot of people told him).
At least he does other benchmarks to round-up the possibilities of errors.
--- Metamoderating abusive downgraders since my 300th post.
Depending on luminance levels, contrast, etc. the eye can perceive at much higher rates than 40Hz. Film appears "okay" at 24fps because the film itself has motion blur, and because most people are used to it (and filmakers work within it's limitations). But I work in a vision research lab where we have a 240Hz monitor (120 Hz for each eye with high-speed shutter stereo). For some things, it does matter.
"It's overkill, of course. But you can never have too much overkill." - Anonymous Slashdot Coward
AthlonXP 2000+ runs at 1,666MHz at a bus which is the equivalent of 266MHz.
The P4 is running at 2666MHz (a full Gigahertz higher frequency) with a bus at the equivalent of 533MHz.
How come so many people rant and rant about how clockspeed isn't everything, then they go and use the same argument in a different way to establish the "clear superiority" of the Athlon? Who cares how many Hz one is than the other? (Don't argue about consumers here, that's for another discussion...).
Sorry, but if you're going to paint it as an achievement that the Athlon performs so well 1000MHz slower than the 2.6GHz P4, then why can't the Intel fanboys paint the fact that the P4 runs at 2.6GHz as an achievement?
The (essentiually overclocked) Pentium 4 has a full SIXTY PERCENT CPU clockspeed advantage and a ONE HUNDRED PERCENT front side bus (FSB) advantage, yet look at its real-world performance:
"Essentially overclocked" Pentium 4? It's not a new Pentium 4 chip, it's a new motherboard. Of course it's an "essentially overclocked" Pentium 4. Why add in the negative connotations?
I just thought I'd point out that the only conclusion that you can really draw from these tests is that, as many in the hardware community know, the P4's architecture is designed for high clockspeed, with zero regard to actual real-world performance. Which matters more to you?
I dunno, looking at these benchmarks I'd say the Pentium 4's architecture is damn fast. It's scaling up incredibly fast. Remember when it was first released and everybody called it a disaster?
Intel could easily release those 2.6GHz chips today, but they aren't doing it for marketing reasons. The architecture of the Pentium 4 is incredibly fast, but the management of the company is spreading out the releases over time. You can get a 2GHz today and overclock it to 2.6GHz. People are doing that all over.
The Athlon is a different design: It's very fast. The Pentium 4 is another design: It's very fast. The Athlon is cheaper, by a fair margin, especially at the highest end chips. But painting the picture that the Pentium 4 is so very much slower than the Athlon, especially with benchmarks like this, are just plain stupid.
His entire conclusion is absurd. Piece by piece:
:)
"Our detailed tests show that forthcoming P4 CPUs with 133 MHz FSB clock used in conjunction with the 845E chipset (DDR SDRAM support) will effectively be castrated."
Intel castrated it their selves. Compare its performance to VIA's P4X266 Chipset's performance and you will see that Intel crippled it to prevent it from competing with Intel's Rambus chipset. Notice that Intel is suing VIA for that chipset because it ruins the facade that RDRAM is better than DDR. Also note that Intel has refused nVidia's request for an Intel license for a DDR chipset. Intel knows that a dual channel DDR chipset would show RDRAM for what it is: A fraudulent attempt to maintain a high performance monopoly. Whatever company "causes to be sold" the most RDRAM gets to own a controlling interest in Rambus Inc. At this point, Intel is the clear winner even though Sony made a race out of it by packaging Rambus with the Playstation 2. Intel suppresses their own DDR performance to make people believe that RDRAM is the fastest stuff out there. AMD would be committing suicide by using RDRAM to capitalize on Intel's marketing hype because that would place them directly under Intel's thumb.
"This is because the Pentium 4 has a problem: the increase in clock speed (e.g. P4/2533 or P4/2666) will be rendered useless by the slow DDR SDRAM memory bus of the 845 platform".
Again, this is Intel's doing for product placement purposes as was done with the Celeron when it competed with the Pentium III and was done by Apple on the new iMac's 100fsb 800mhz G4. A 133fsb does not cost any money, it is just an easily achievable clock frequency with available current chipsets.
"And one shouldn't forget that even a dual DDR platform for P4 should be priced at a level that is similar to a Rambus system, considering that it's from Intel."
Rephrased: "And one shouldn't forget that even a dual DDR platform for P4 will be priced as high as an RDRAM system because Intel will not license the platform to nVidia and Intel KNOWS it will outperform a Rambus system, ruining 2 years of carefully crafted marketing and gamesmanship" The fact is, a dual channel DDR chipset from Intel may be available for the Pentuim 4, but only for the Xeon, a processor not available except from Intel's favored OEM Parteners, such as Dell.
Before you defend Intel remember that Craig Barrett, after AMD went from 10% market share to 40% in a year, said "the market is dropping" to justify Intel's reduced profits. Well, Intel is a bellwether stock and the market believed everything Craig said. The market did drop. We all lost our jobs. We can now say in hindsight that at least a part of the market was due to drop. But because of Craig's statement, it was the tech sector that was hit first, and hardest. Instead of simply saying, "Intel has reduced profits because of competitive pressure", he brought the entire tech sector down with him. The recession that was due could have been placed entirely on Enron's shoulders. The energy sector was in fact dropping. Enron's insiders were cashing out at the same time Craig made his statement. People got scared and pulled their money out of the market. There was less money in the market than there had been and it came out of the tech sector when it should have come out of energy.
Go ahead and defend Intel. They have made poor greedy choices, sold inferior products at exorbitant prices and done it at the expense of all our livelihoods. Shame on them.
Intel's 1.7 trillion dollar market cap has been cut by Tom Pabst on more than one occasion. A series of articles he has had deriding Rambus, causing the 1.13 Ghz recall, and showing the Pentium 4 for the paper tiger it is has seriously hurt Intel. But Tom, like all hardware websites is cash poor. Tom's hardware has resorted to doing marketing research among their readership for Socratic Technologies. Sometimes they have been overt, sometimes they have sent readers to secure servers just for simple popularity polls. Tom's latest revenue generation technique is the introduction of "Editorial Content Sponsorships" which I'm going to guess prompted the recent editorial change of heart toward Rambus. Please notice that in the most recent article no AMD processors were over clocked according to their projected roadmaps and the test is presented as if it was fiction. Unfortunately, it seems we have lost another fair and unbiased journalist. Another because Sharky's Extreme was the first to go into Intel's pocket, prompting Sharky himself to leave the website. Sharky's is owned by INT Media Group. Noteable investors in INT media include Dell Computer Corporation, International Business Machines Corporation, Lucent Technologies Inc., Macromedia Inc., Microsoft Corporation, Nortel Networks Corporation and Oracle Corporation.
Expect wonderful reviews of Intel hardware on Sharky's and unfortunately now, Tom's. Look to [H]ard OCP, The Inquirer, The register, Anandtech and Ars Technica for relatively unbiased hardware news.
Post Intelligently, Thanks
If voting were effective, it would be illegal by now.
It is important for a variety of reasons not to let up upon the current technological pace occurring today. There are so many factors to consider economically, scientifically, and sociologically. If we allowed a slow down of the current pace of technological advancement it could have a devastating impact on our society at large.
First off, it is naive to think that current users wouldn't use or enjoy more powerful computers. It is the software industries fault that end users are unable to fully utilize the more powerful machines being built. Already plenty of comments have suggested a variety of applications from facial recognition to video editing that all would benefit from faster more powerful computers.
It is actually important to me that regardless of the 'need' the average user has for more powerful computers, that the software industry does its job to drive the users to want more power.
Only by nurturing and then feeding the publics appetite for technology does the industry continue to push us forward technologically. If millions of people and companies didn't demand the upgrades and new features that are available with more powerful systems we risk losing all the potential gains for the future that these desires produce.
Screw that, I _NEED_ the extra performance.
/PAINFUL/. Running any sort of complex filters meant going out to a friggin lunch break, bleh).
:) ) we should have MP3 encoding speeds of a few MINUTES per second passed.
;) ) that cost me $40 per CPU (ok so at that price I bought two, wish I'd bougthen four or five. :) ) will have to suffice.
Hell, I was DISAPOINTED with the ABYSLEMAL results that came in.
Huh?? You ask?
Well yah.
You see how that MPEG4 video took TIME to encode? Time that could be measured in MINUTES per video?
Tell me when I can do MPEG4 encoding at over 1000x real time speed with shitloads of Virtual Dub filters running and without out my CPU even going up to 10% utilization, and THEN I will say that we have (maybe) gone fast enough.
As it is I still have to hit the render button, wait... wait... wait... wait... wait... wait... wait... wait... wait... wait... Run it through my post production filters and repeat the waiting (seen above) if not for an even longer time, and THEN I get to compress it down to some sort of video stream (choose a codec folks).
Ooooh great...
Royal pain in the arse when rendering takes longer then creation.
Oh yah, and did I mention that I am not even using over $1k of software here? I am not even running some sort of fancy high end effects house, I am just doing regular quicky animations. But rendering those Terrains sure is a pain in the arse, and then those realistic clouds, ooh ouchies MAJOR performance hit there folks.
Heck even photoshop still takes times to run filters. Not even complex filters either, just single ones. (It has gotten A LOT better since the 'old days' of running Photoshop on those "brand new Pentium 166mhzs!!!" Oh man, that was
What about even transfering images from a digital camera? You know how bleeping long it takes to load previews of all images in a folder? You know, all 100-200 images? Or more? Most likely of varying resolutions to boot. How lovely.
That _IS_ enough to annoy a Grandama and encourage her to upgrade to a new machine.
You think people want to WAIT to encode their MP3 streams? Why? By the time we hit 10ghz or so (and if HD speeds hopefuly start scaling up a bit faster.
Or at least we sure as friggin better, heh.
Until then my 700mhz Duron OC'd to 950mhz/1ghz (depending on time of year
(well that and my 80gb + 20gb Hds which are quickly filling up. Screw CPU time, I can always play Gameboy, but I _NEED_ more HD space damnit! I filed up 40GB in two weeks, and I wasn't even trying!! )
Need help treating your acne? Come here!