Most likely, it is the BLX Godson-2, a MIPS-like 64-bit processor developed entirely in China. This would mesh well with the Chinese government's intent to encourage the Chinese people to buy Chinese electronics.
Unfortunately, we have no idea what the architecture is like. I know a lot of you think MIPS is inherently "more efficient" than x86 per clock, but that's just foolish fanboyism. MIPS-compatible processors come in a variety of speeds with different Instruction Per Clock capabilities. Some are VERY simple, single pipelined designs intended for embedded markets, while others are aggressively clocked, with superscalar designs like those from SGI.
Just making it MIPS doesn't guarantee better performance. BLX has never publically stated exact performance comparisons, and they could be comparing it to a "200 MHz P3 (yes, I know this doesn't exist)" for all we know. Think about how difficult it would be for an updtart chip maker to compete with a Pentium 3 clock-for-clock: the processor would require supercalar design with out-of-order execution, including dual integer pipelines and a pipelined FPU. This is very unlikely from a project that has only been designing processors for 5-6 years.
More likely, the processor is more like the older Winchip, with a single integer pipeline and a low-latency non-pipelined FPU. This would give the 600 MHz Godson-2 performance on the order of a P3 at 200-300 MHz.
But if DSPs and some generations of software allowed to squeeze >25MBit through a pair of copper lines which where said to top at 56kbit, they may do something similar to EEGs.
Just a note. Older modems used the audible frequencies of the phone line, int he range of around 300 Hz to 3500 Hz. Later modem standards could take advantage of "all digital" voice lines with 15-bit samples 4000 times per second (56k). But all of these were limited to the VOICE band, because that's all the equipment at the interchange propogated.
ISDN takes advantage of up to TWO of these digital voice lines.
BROADBAND modems take advantage of the fact that copper can carry a LOT more bandwidth than just voice, but equipment must be added at the interchange to connect to a real network.
I only have 1GB and have never had an issue. Before swinging your e-penis around on forums, you should probably take some time to investigate your PC. Noob.
Noob? Investigate my PC? You think I DIDN'T investigate the problem?
Before patch 1.12, OTHER players complained of memory leaks that I NEVER experienced. I could play for hours on-end (and did) without a single memory issue or crash.
Patch 1.12 fixed memory leaks for a LOT of people, but it actually INTRODUCED a memory leak for me.
Specifically, the game used up memory that could not be reclaimed by the OS. At a point, the game was using so much memory that it begins swapping to disk. It was at this point that the game either turned into a slideshow, crashed to desktop or hard locked. If the game crashed to desktop, a couple hundred megs of memory could not be reclaimed by the OS (obvious memory leak). Mind you, this hardware had ZERO issues on any other game I ever purchased, and had ZERO issues in all the patches leading up to version 1.12.
The game was working fine for most people, but I did find other folks who had the same problem. The only things we could find in common were 1GB ram and 6600 GT series video cards...but that didn't quite pan out, because other folks with similar hardware had no issues. In the end, we just figured the game had issues with some 128MB cards. The folks with 2GB ram wouldn't notice it, because it would take HOURS of continuous play for that meory leak to raise its head, so that made it even harder to track down the issue.
Still, we tried to fix it. The community tried out different beta and WHQL certified driversets to try and find a fix. They tried different sound configurations and different acceleration settings to try to corner the issue. In the end, I said fuck it, and went on to other things.
Funny thing, I recenty upgraded my system, and for the hell of it I installed Battlefield 2. Just as I expected, no memory leak - there is no unreclaimed memory upon exit.
Also, how exactly are you teaching EA a lesson by giving them $60 and not playing the game?
Let's just say I got my money's worth before patch 1.12 was released. I played the game for almost 6 months. I wasn't teaching EA a lesson - no, this was an exercise in self-control. I don't upgrade just because a game suddenly stops working, not when my new PC is less than a year old.
I figured I would let it rest until my next upgrade, because BF2 is fairly popular, and the player base isn't going anywhere. I figured the game wouldn't have the same memory leak on different hardware - and what do you know? It works fine on the new machine, and feels fresh now that I put it on the shelf for a year.
THAT is the point I want to get across to these WoW players. You don't have to quit PERMANANTLY. But if you're sitting around on their servers just waiting for a miracle to happen, it's not gonna. MMORPG players are like breadcrumbs: once a few key players start leaving one game for another, others follow in droves, but until that happens nobody moves a muscle. The key is, someone has to have the willpower to take that first step.
I'm sick and tired of people whining about poor playing conditions on WoW, and having to wait hours or sometimes days to get in some play time. This ISN'T the first time I've seen WoW whining stories on the front page of Slashdot, and it only seems to be getting WORSE. These are issues that would KILL any other MMORPG, it is simply the massive inertia that keeps people from leaving. I love it that I was modded troll just for telling some whiner to either suck it up and take the beating from Blizzard (stop whining), or forge their own road and leave WoW.
Boo-hoo. You're spending $15 per month to wait an hour every night, and not play on Tuesdays at all. Woe is you! Save it for someone who cares. You've had a year and a half to realize things just aren't getting any better, and meanwhile you could have been pumping your money i
Exactly, here is a real-world example of a mid high-end X2 3800+ system built to be quiet. I recently built the following:
Athlon 64 X2 3800+ w/retail cooler (3600 RPM full speed, not exactly quiet). GeForce 7900 GT (with a LOUD stock cooler) Asus A8N5X (with an annoying 6000 RPM fan on the chipset).
To make this system quiet, I used the proper components, and voided a couple warranties:
* Antec Sonata II case w/450w powersupply. VERY quiet, stock ($100). Is priced competitively with other quality cases.
* Zalman vf900-Cu video card cooler, to replace the "miniature vacuum" stock cooler on the 7900 GT ($50). Alternatively, I could have used the vf700-Cu, or an Artic Cooling fan ($30).
* Zalman passive northbridge heatsink, to replace the 6000 RPM screamer ($5).
* Scythe 120MM 800 RPM liquid bearing fan, to be mounted on the front fan mount to cool the hard drives and the new passive heatsink ($15).
* Used the Q-fan feature of the ASUS motherboard to throttle the retail CPU fan. Even at max load it doesn't top 2400 RPM, and at idle it is a whisper at 1100 RPM ($0).
Total cost to make my system quiet (assuming you could have bought a crappy, loud case + powersupply instead for $50):
It AMAZES me how much shit people will put up with to keep playing MMORPGs. If the system is broke that badly, why don't you leave? Why the hell do you think the problems keep getting WORSE?
I mean, I certainly wouldn't put up with that shit, especially if I was paying a monthly fee.
Hell, I stopped playing Battlefield 2 after they released the version 1.12 patch because it suddenly caused a memory leak on my computer. With only 1GB of RAM, it would crash to desktop after about 30-45 minutes. I like to play Battlefield for much longer stretches, so this was unacceptable.
I LOVE BF2. I *CRAVE* the experience...it's even more addicting than BF1942 was. But I quit playing, because I have standards. I am NOT going to upgrade to 2GB RAM just to "fix" THEIR problem.
QUIT the game. Let Blizzard have a chance to catch their breath, then maybe they'll resolve the issues.
Exactly. These days, there is no such thing as future-proofing.
Take the Athlon / Athlon 64 line, for instance. In the past, we've seen nice upgrades in bus speed (100, 133, 166, 200), and if you invested in a motherboard with faster bus speed support, and good enough ram, you could upgrade for cheap.
Socket 939, with the introduction of the X2 processor, also offered an opportunity for a (relatively) cheap upgrade from single core.
But where does that leave us now?
Over the past two years, AMD has gained only 800 MHz at the top-end, and this is all thanks to the tweaked 130nm, and later, the tweaked 90nm process. It sounds like a lot, but percentage-wise, it has not been impressive. Only the introduction of dual-cores has kept performance improvements in line with market demand.
Slow performance gains means one of two things: expand to quad-core processors, or redesign for better Instruction Per Clock (IPC) rates.
So what happens when we go quad-core? Not so much. Most multithreaded applications are designed to take best advantage of dual-core processors. Beyond that, there are only so many people who do constant hardcore multi-tasking, and even then you run into I/O limitations hindering multi-tasking gains. Also, unless AMD finds a way to magically pump up the HT bus and improve the performance of their dual-channel DDR2 controllers, the 4 cores will be data-starved. Quad-core is only going to find a foothold in servers, and ONLY if AMD releases better performing memory and HT busses.
So what does that leave us? Like Intel's Conroe, AMD will have to redesign the Athlon 64 for greater IPC, and like Intel, AMD will have to prioritize process improvements. Neither of these seem to be happening, so the future of "upgrades" on the AM2 platform is pretty bleak.
Me, I just bought a 3800+ X2 plus 2GB ram. The prices I've seen for AM2 X2s are exactly the same as those for Socket 939, and DDR 2GB is not any more expensive than performance DDR2 (ie, DDR2 800). I'll get more out of overclocking the 3800+ than I ever would purchasing an upgrade in the future.
Even if I didn't overclock, the performance gains of upgrading to a 2.6, 2.8 or maybe 3.0 GHz X2 a year or two down the road aren't exactly impressive.
To say nothing of the nightmare of getting printers to work properly
I will say I had zero trouble setting up printing under OS X, and this was on an Intel Mac. Like Windows, I simply selected the printer from a list (or provide drivers), and enabled sharing.
I can say without a doubt in my mind that it was ten times easier than setting up the same printer in Linux via CUPS. After hours of crawling support forums, and my local Linux guru having a look, we discovered we had to replace lpr, because the default lpr included in Debian was some older BSD version that didn't support USB. At least, that's how he explained it to me.
It's funny, really...Apple uses CUPS, but that's because it ACTUALLY WORKS once you get rid of all the backend *NIX crap. It worked well under Linux too, once we wasted hours tracking down the lpr issue.
I also couldn't get my Matrox G400's OpenGL acceleration working in X under Debian (even with said local Linux guru), despite the fact that Mandrake configured it correctly out of the box. Tell me, if I'm following the FAQs to the letter, how the hell can one distro get it right the first time, and another distro with similar components leave me completely puzzled? Just more inconsistency to piss me off.
You can understand why I left. I had this kind of patience when I was a kid, sure. MSDOS was almost as insane a place as Linux in terms of tracking down settings, and inconsistent behavor, and I imagine if I had gotten started on Linux instead of DOS, I'd know it back-to-front. But now I'm getting older, have less time on my hands, and I don't have time to learn EVERYTHING...so I tried something simpler....instead of having ghastly shit like Spotlight foisted on me
What's wrong with Spotlight? Honestly, I always though that locate and updatedb were decent, but clumsy, because I don't like opening a terminal window just to search for some files. Spotlight is right there on the bar, and even categorizes the data, something locate didn't do.
It's as fast as locate, has more meta-data, and you don't have to open a console. What's not to love?
or frigging menu bars that are ten miles across the screen from where I'm working,
Yeah, I'm not so sure I like this "one menu for all" thing either, but then I also must admit that, for those applications where I really make use of the menus, I usually keep those applications maximized, and that means no more movement to reach the menu than Windows or Linux.
Mind you, it does annoy me that some 3rd-party widgets actually have options hidden in the menu, which is utterly stupid. For instance, the option to cancel or run print jobs on my printer is embedded in a menu, not a checkbox or button on my printer widget. Is it Brother's fault for making such a boneheaded design decicion, or is it Apple's fault for allowing them to do it? Good question.
I will say that Apple sets a good example for third parties: I have yet to find a configuration widget that is part of the OS that has functions hidden in the menus.
Then I get to re-experience the same reason why I just recently left Linux for OS X.
Stuck inside their bubble, open source zealots think it's ok to have thousands of branches of THE SAME OS, COMPETING, INCOMPATIBLE WINDOWING LIBRARIES that aren't even standardized as part of the OS, and ten thousand window managers to make absolutely certain that your OS has no recognizable "look" or "personality" whatsoever. When they add new features, they add them in the "cleanest possible" manner (ie, make everyone patch and recompile EVERYTHING), rather than the "most usable" manner (add compatibility layers). When they change features, they don't do so gracefully, they break old code and expect everyone to recompile.
Here's just one example of how open source just gets it wrong: a few years ago, I was looking to play some emulators on my Linux box. I figured it would be as easy as emulation on Windows, but boy was I wrong.
See, I wanted to use the same USB gamepad I'd been using for the last few years on Windows. Only problem was, when Linux added support for *USB* gamepads, they used a different interface. Thus, emulators designed for *ANALOG* gamepads could not use my USB gamepad. Unfortunately, most of these emulators had been abandoned, and nobody had bothered to add USB gamepad support, so I was up shit creek unless I wanted to hack it in myself (sound familiar?). This is an example of adding a new feature CLEANLY, but in a manner that is completely UNUSABLE without extensive reworking.
I'm sick of it. It's little things like this that made using my Linux box for anything besides web browsing and basic office tasks a pain.
WINDOWS, by contrast, has supported USB gamepads since Windows 98, and has taken all the guesswork out of the issue. Regardless of whether you use an analog or USB gamepad, an application can use the same hooks to communicate with the pad regardless, and the user can use the same setup widget to configure ALL pads. Now, that's not going to be very CLEAN code, but it is a damn sight more USABLE. I can't speak for OS X on this issue, simply because I've only been using it a few weeks.
This is just one example of why I don't need yet another open-source operating system. Open source applications are just fine - the ones that are mismanaged or get caught up in their own self-image eventually get replaced by better development projects. But operating systems have momentum, and don't just disappear. Open source zealots could do a lot of damage to the USABLILITY of OS X, and it would still take a long tome for it to fade away.
I used to be a Matrox junkie. I loved the 2D quality. I loved the high performance of the G400 series. But Matrox took too long to relase a followup. The G450 was late and unimpressive due to the low clock speed and 64-bit unoptimized memory bus.
The G550 was late and unimpressive due once again to the low clock speed and 64-bit unoptimized memory bus.
This whole time, Matrox was sitting on their ass, not improving on the DualBus originally released with the G400. I knew what was going on thanks to MURC (Matrox User's Resource Center, no longer exists), which was frequented by disgruntled current / former Matrox employees. They said that after the G400, Matrox management did not have a vision, and as a result their best digital and analog folks left. The recovery time is why the G450/550 were so late, and not at all innovative Matrox never regained their G400-era fire.
Meanwhile, ATI and Nvidia were learning that memory bandwidth was becoming a severe performance limiter, and out of necessity they learned tricks like texture and Z compression, and other memory bandwidth savers. ATI and Nvidia also switched from wide 128-bit unified memory busses to crossbar memory architectures to improve overall bandwidth utilization.
I didn't wait around. I bought a Radeon 8500 to replace my G400 MAX about 6 months before the Parhelia was released.
When I saw the Parhelia, I knew I had made the right decision. Matrox used the same old bus technology they had been relying on for years, and upgraded it to 256-bit wide to try to make up for their technology deficit. It performed a little better than cards with 128-bit busses, but wasn't all that impressive.
When the 9700 Pro was released with a 256-bit bus, and ran circles around the Parhelia, it dawned on people just how bad Matrox's memory tech was. Fragment AA was a nice carrot to attract people, but it wasn't enough to make up for the car's incredibly poor performance.
The thing is, Apple supplements their boring, typical hardware by releasing unique devices that are ahead of their competition.
Just look at the Mac Mini. When it was released, NO vendor had a desktop anything like it, and even today the Intel Mac Mini is competitively priced with Intel Core-based knockoffs. The difference is, the Mac Mini is ESTABLISHED thanks to Apple taking a risk, and the rest are scrabbling for marketshare.
The original Ipod was smaller than anything else, but offered the capacity of a hard drive player. It was ungodly expensive, but it became rooted because the industry took too damn long to respond to the threat.
The Nano is another recent product that still has no competition. I don't know how, but Apple managed to get a deal for flash devices twice the capacity of their competitors (32Gb, while the rest of the industry was on 8Gb). The result was the 4GB Nano, with room for Apple to upgrade the line to 8GB in the future (once the 32Gb chip becomes cheap).
The fact that the Nano 2GB was priced at only $10-30 more than most other 1GB players shows just how lazy and wussified the whole industry is outside of Apple. Apple set the standard for pricing of 1GB players with the release of the Shuffle, but then instead of the industry pushing prices down and capacity up, they coasted while charging $10-30 less than an equivilant Shuffle. Then, when the 2GB Nano was released for $10-30 more than a 1GB player (despite the fact that the Nano used TWO 8Gb chips), was still competitively small AND had a smashing screen, the rest of the industry looked painfully stupid.
THIS is why Apple owns the industry...because they're the first manufacturer since DIAMOND MULTIMEDIA's RIO PMP to really PUSH the industry.
I have purchased Creative Labs Muvo players in the past because I WANT drag & drop without having to use a media library, but with Rockbox now I can turn any Ipod into exactly the interface I want. Suddenly, I don't have to compromise on larger size and less capacity for about the same price just to get some key features I need...Rockbox makes that possible.
Full Disclosure: I bought my first apple product EVER last week, a Mac Mini. The OS is decent, but I could care less about the apps. I don't like Safari, so I have Firefox. I don't want to be forced to use a media library with Itunes, so I went through the trouble of installing XMMS. I bought the Mini because it was tiny, VERY quiet at nominal room temperature, competitively priced with other small Intel Core solutions. It also offered me the chance to try OS X, but fall back to Windows or Linux if I really hated the experience...that's something no other small Core box could offer.
If the industry continues to lag Apple concerning small flash players, I may purchase a Nano or equivilant sometime in the future.
Don't get me wrong, Rosetta is decent (for an emulator), but Firefox has just been ungodly slow on my Intel Mac Mini. I've been waiting for an official Intel build.
It's been almost two full years since the consoles were in their swing, it has even been over a year and a half since big PC games like WoW, HL2, Doom 3, etc were released.
The industry has nothing BIG. Oblivion is decent, but it doesn't have the same momentum as a wider-audience game series like Quake / Doom / UT / (War)(Star)craft, and the cost of entry is VERY high (either an Xbox 360 or a relatively well-specced PC). With the unimpresive, high-priced start of the Xbox 360, and with a similarly high-priced start expected for the PS3, what does your average gamer (who can't afford or justify a 360) have to look forward to? Interesting new releases on existing consoles are rare compared to a couple years ago.
To put this in perspective: this is the second generation in a row where mainstream gaming system prices have gone UP significantly. Before the PS2 / Xbox, most gaming systems debuted at around $200-250. When you took inflation into account, that meant the system was always cheaper than the previous generation.
But the PS2 / Xbox pushed that envelope to $300, and the Xbox 360 and PS3 are pushing that number to $400 or more (yes, you can buy a $300 Xbox 360, but it will really cost you $340 with the official memory card, and well over the $400 mark if you add a hard drive in the future). Gaming is getting significantly more expensive, and it's no wonder people are playing less games. People have done the same thing in response to high prices, say, in theaters.
That would be due to several "lessons learned" as Intel developed Itanium.
1. The instruction overhead due to extra hint bits, etc, means Itanium instructions are much larger than x86 32/64 instructions. With the addition of poor branch performance (read: more wasted instruction bandwidth), the need for large, high-bandwidth caches makes Itanium expensive.
2. The compilers have not caught up. EPIC lacks OOOE, and has poor dynamic branch prediction hardware, so it is at the mercy of the compiler.
Core retains Intel's original insights made with the P6:
1. x86 is hard to decode (takes more silicon), but it takes less bandwidth than other instruction formats. Bandwidth is even more expensive than the cost of more complex decoders, just look how expensive it was for Intel to add full-speed cache to the original Pentium Pro, and how pricey the Itanium is with huge, fast on-chip cache.
2. OOOE + Branch Prediction + internal RISC is king. One reason the original Pentium never performed well is because it could RARELY execute more than one instruction per cycle. Thus, it performed like a fast 486 unless the code was recompiled as Pentium optimzed. The P6 was designed to avoid the reliance on compilers to improve performance, as it could optimize code in any condition. Funny, we didn't start seeing Pentium-optimized code on the market until the P6 started taking over.
Core is just a logical extension of this concept. The predictor is more accurate, there are more instruction decoders, more ALUs and SSE units, and more retirement units. The only reason Core seems to groundbreaking is because we didn't see it in small, evolutionary steps.
"Just one addition, branch instrucctions don't hold up processors, memory load and floating point instructions do. The problem with the branch instructions is that the branch decision may come too late, and the penalty depends on the depth of the pipeline."
No. For someone who claims to be educated in computer architecture, you are sadly lacking in understanding. Unfortunately, all you're thinking about is the mispredict penalty.
Floating-point multiplies are FAST these days. Memory load issues can hold up a pipeline, especially if the cache heiarchy has not been designed for high hit rates (too low capacity, bad cache block design, etc), but that's not typically the case these days.
Branches are one of the biggest issues for the x86 32/64 platform to handle because of the non-uniform instruction size, and because ACCURATE branch predictors are SLOW (multi-cycle). What this means is instruction decode + prediction ends up taking multiple cycles, and during those cycles the processor keeps reading instructions as if a branch wasn't taken. By the time you have a branch decoded, fed into the accurate branch predictor, and predict taken, what do you do with all the instructions in the decode pipeline?
You flush them. That's wasted fetch bandwidth, plus you have bubbles in the execution pipelines while the fetch pipes refill. THAT IS COSTLY.
The Pentium 4 uses a trace cache to get around this issue. It keeps a record of all instructions executed in actual branch order, unfragmented, based on their previous branches. Unfortunately, there is a big hit if the trace cache mispredicts or doesn't contain your instructions, because then you have to go to L2 cache. It's not optimal, but it does work well.
The optimal solution? You have FAST but inaccurate branch predictors that can work in a single cycle, and you make fast decicions about branches. You also have to predict the branch target address using a branch target buffer, because calculating it would take far too long. Thus, you can keep fetching immediately, even if the branch is "taken".
The final two pieces to this system are a pipelined accurate branch predictor that can override the fast predictor, and of course, the real branch verification hardware (which causes the pipeline flush you commented on if the direction predictor is wrong). This system guarantees you much less instruction pipe stalls while you wait on an accurate branch prediction. Unfortunately, it is expensive to implement, so most processors use some sort of compromise system (like the Pentium 4).
Right, there are two camps for the "high-end" branch prediction concept:
Camp 1: devise adaptive, multi-component prediction systems that offer both fast and accurate branch prediction. Waste hardware purely for branch prediction.
Camp 2: Use the compiler hint if available, otherwise execute both paths, and throw away the incorrect processing path. It seems cheaper on the surface, but you have to realize: all that extra fetching to process both paths in reasonable time mean more fetch bandwidth and more execution units required just to keep up.
Obviously, if your code contains lots of branches that cannot be predicted by the compiler hints, the Camp 2 solution is going to perform worse. The advantage of active branch prediction is that you never have to recompile the code to keep the branch hints "optimized" if your datasets change.
It doesn't really matter which camp you choose, because both camps waste space on a Branch Target Buffer (predicts the TARGET of the branch) anyway, and that's often more costly than the branch direction predictor. Even the Itanium has a BTB, that's how it can instantly start executing the "branch taken" case.
The Itanium is just taking advantage of a serious architectural flaw to perform branch prediction. Even modern compilers are inserting 20% or more "noops" into the instruction stream, why not take advantage of that underutilization. On any other platform, it would be a very stupid approach to branch prediction.
"1T" SRAM is DRAM with an SRAM front-end cache (thus, the "1T" name...DRAM requires one transistor per cell). Normal SRAM is "6T," which indicates it requires 6 transistors per cell.
So long as the front-end cache is large enough, the latency is close to that of a real SRAM cache. The benefit is "1T" SRAM uses a lot less die real-estate, and a lot less power.
Sure, it is an x86 processor, based on the Winchip.
Let me just put this in perspective for you:
For anything other than MPEG2 or MPEG4 viewing, or encryption, this processor will perform like a PII 350 or slower. Imagine running Gnome or KDE on a PII 350 with a TINY (64k) L2 cache, and you'll come to understand just how much this chip sucks.
Via claims that their philosophy is low-power computing, but the real deal is Via does not want to pay their design team to completely redesign the Winchip. The Winchip is not even a real superscalar processor: it only has one integer unit, and one floating-point unit. Think of it as a highly-optimized 486, with performance much closer to 1 Instruction Per Clock.
It wasn't a bad design way back when the Pentium ruled the roost, because the Pentium typically executed one instruction per cycle. It was pretty rare to keep both instruction pipes on the Pentium filled, even with a decent optimized compiler. But then the P6 came along with support for out-of-order execution, making the Winchip obsolete overnight.
The sad thing is, modern processors like the Athlon 64 and Pentium M can compete with the Nehemiah on power usage AND performance. You can take any Athlon 64, clock it to 1.0 GHz and 0.8-1.0v, and it will have twice of the performance of any Nehemiah processor and use less power. You can do even better with the Pentium M, which supports voltages as low as 0.7v.
Your noisy Opterons, for instance, can probably be underclocked and undervolted, and you can potentially cool them passively (if they are of the 90nm variety).
They figure you'll look up that information, their biggest concern was spacing out a three page review over ten pages.
That said, Via normally names their boards with a number that is 10x the CPU speed. The board is named the N8000, so that would indicate 800 MHz, like all the 130nm passive MiniITX boards. The benchmarks agree.
It's really quite sad that NanoITX only uses the old crappy CN400 chipset with the older 130nm Nehemiah core. I mean, the 90nm rev still sucks, but at least they can clock it higher, and at least the bus isn't holding it back. You would think after all this delay they'd be able to put their new C7 platform in NanoITX form-factor.
800 MHz on a CN400 chipset is about the performance of a PII 350.
Not THAT much. You can eye the power consumtion results and know that Hot Hardware did not enable Cool 'n Quiet. Otherwise, there should have been a 50-60w difference between idle and load.
The Intel chip, on the other hand, has Enhanced Speedstep enabled. You can tell this by the fact that they mention the prcessor's voltage is from 1.2v to 1.35v. With Speedstep, the processor at idle goes to 2.8 GHz at 1.2v. AMD's Cool 'n Quiet is even more drastic, dropping the processor speed to 1 GHz at 1.1v.
The fact that AMD can ALMOST match the power consumption of Intel using Enhanced Speedstep, while not using Cool 'n Quiet, shows you just how much better the Athlon 64 is.
Hey, I agree wholehartedly, and I've been an AMD fanboy since the release of the P4.
Anyone who can look at this breakdown of the new Core design, understand it, and STILL proclaim AMD the performance leader is retarded. The extra simple decoder means potentially 33% more thoroughput out the gate, and the fused micro-ops can add another 5-10% performance improvement (assuming you have enough execution units to use all this). The 128-bit SSE unit, plus the ability for simple decoders to handle packed SSE instructions, also means double the speed at vector operations.
That said, at least I had my just desserts. I always said superpipelined Netburst was a retarded design, and the fact that Intel went and developed Conroe only validates my claim.
I am still curious to see the power usage of Core. It should be less than the P4, but whether it is competitive with AMD may be another story. Hopefully AMD will finally get off their ass and improve their own design, which hasn't changed much since the K7 (onboard memory controller aside). Who knows, I may end up buying Conroe, and becomean Intel fanboy again.
Because the Pentium 4 was never made for powerful floating-point thoroughput using plain old x87 FP instructions. To give it better floating-point performance, a 128-bit vector unit (SIMD) was strapped on. With it, the Pentium 4 can perform 4 32-bit floating point operations every 2 cycles.
This is the one part of the Pentium 4 chip that is very much like the parallel execution units of a video card. If you want to maximize performance of the SIMD unit, you want a simple program with packed SIMD instructions. The key point is, the video card can do that better.
I just finished talking to a "normal" gamer buddy of mine today. You know, he likes to game, and his kids like to game. He was telling me about how much he and his kids love Call of Duty 2, and how smooth and amazing it looks on a Radeon 9200. As I sat there fighting to keep my mouth closed, he went on to mention he was buying a gaming PC "just for him," with a fancy new GeForce 6600 (no, not the 6600 GT).
I almost laughed at him, but then I realized: he doesn't care about fancy TRMSAA or Angle-dependent Anistropy, he doesn't really notice fancy real-time lights or shadow-maps. The only thing that matters to him and his kids is the game looks decent (ie, at least 640x480 resolution, reasonable resolution textures, and billinear filtering).
And why not? A whole half of the gaming population bought into the PlayStation all because it could do fancy new 3D rendering, even though the unfiltered textures and 100 polygons per scene looked like crap. Most of them traded up to the Playstation 2 because it could do much better polygonal detail and allowed higher video and texture resolutions, plus DVD.
So, here's the problem: most people think that the PS2 looks great, and it is in fact almost as good as you can get on a standard TV. So, why should people upgrade when the current console does everything they think they need?
If you don't have an HDTV, the PS3 doesn't really offer anything over the PS2 except Bluray, and again, the DEMAND for Bluray depends on you having an HDTV. All the selling points for the PS3 come back to this HDTV thing.
And normal people don't care about HD. Take my sister and her husband: they have Sony everything, a nice Sony DLP widescreen, plus a matching Sony 5.1 receiver and speaker set. It's great, except most of the things they watch are in SD, and the speakers are set up in two clumps on either side of the room. The closest they get to watching HD is through their DVD player, but I know better than to argue with them.
Look at the XBox 360. IT LOOKS MUCH BETTER THAN PREVIOUS GENERATION CONSOLES, there is no doubt of that. The problem is, it doesn't look GOOD ENOUGH to warrant the upgrade, so people stick with their current-generation consoles and games. You can be sure that the PS3 will encounter the same problem.
I mean, people convince themselves all the time that Sony equipment is somehow "better" than other brands, but what happens when it's Sony versus Sony? I expect all those eagar PS2 buyers won't be so eagar to go PS3. The only console to offer anything really DIFFERENT is the Revolution, but that gives it as much risk of sinking as swimming. One thing I am sure of: console sales this generation, overall, will be lower than last generation.
Slashdot Mirror
Most likely, it is the BLX Godson-2, a MIPS-like 64-bit processor developed entirely in China. This would mesh well with the Chinese government's intent to encourage the Chinese people to buy Chinese electronics.
Unfortunately, we have no idea what the architecture is like. I know a lot of you think MIPS is inherently "more efficient" than x86 per clock, but that's just foolish fanboyism. MIPS-compatible processors come in a variety of speeds with different Instruction Per Clock capabilities. Some are VERY simple, single pipelined designs intended for embedded markets, while others are aggressively clocked, with superscalar designs like those from SGI.
Just making it MIPS doesn't guarantee better performance. BLX has never publically stated exact performance comparisons, and they could be comparing it to a "200 MHz P3 (yes, I know this doesn't exist)" for all we know. Think about how difficult it would be for an updtart chip maker to compete with a Pentium 3 clock-for-clock: the processor would require supercalar design with out-of-order execution, including dual integer pipelines and a pipelined FPU. This is very unlikely from a project that has only been designing processors for 5-6 years.
More likely, the processor is more like the older Winchip, with a single integer pipeline and a low-latency non-pipelined FPU. This would give the 600 MHz Godson-2 performance on the order of a P3 at 200-300 MHz.
But if DSPs and some generations of software allowed to squeeze >25MBit through a pair of copper lines which where said to top at 56kbit, they may do something similar to EEGs.
Just a note. Older modems used the audible frequencies of the phone line, int he range of around 300 Hz to 3500 Hz. Later modem standards could take advantage of "all digital" voice lines with 15-bit samples 4000 times per second (56k). But all of these were limited to the VOICE band, because that's all the equipment at the interchange propogated.
ISDN takes advantage of up to TWO of these digital voice lines.
BROADBAND modems take advantage of the fact that copper can carry a LOT more bandwidth than just voice, but equipment must be added at the interchange to connect to a real network.
I only have 1GB and have never had an issue. Before swinging your e-penis around on forums, you should probably take some time to investigate your PC. Noob.
Noob? Investigate my PC? You think I DIDN'T investigate the problem?
Before patch 1.12, OTHER players complained of memory leaks that I NEVER experienced. I could play for hours on-end (and did) without a single memory issue or crash.
Patch 1.12 fixed memory leaks for a LOT of people, but it actually INTRODUCED a memory leak for me.
Specifically, the game used up memory that could not be reclaimed by the OS. At a point, the game was using so much memory that it begins swapping to disk. It was at this point that the game either turned into a slideshow, crashed to desktop or hard locked. If the game crashed to desktop, a couple hundred megs of memory could not be reclaimed by the OS (obvious memory leak). Mind you, this hardware had ZERO issues on any other game I ever purchased, and had ZERO issues in all the patches leading up to version 1.12.
The game was working fine for most people, but I did find other folks who had the same problem. The only things we could find in common were 1GB ram and 6600 GT series video cards...but that didn't quite pan out, because other folks with similar hardware had no issues. In the end, we just figured the game had issues with some 128MB cards. The folks with 2GB ram wouldn't notice it, because it would take HOURS of continuous play for that meory leak to raise its head, so that made it even harder to track down the issue.
Still, we tried to fix it. The community tried out different beta and WHQL certified driversets to try and find a fix. They tried different sound configurations and different acceleration settings to try to corner the issue. In the end, I said fuck it, and went on to other things.
Funny thing, I recenty upgraded my system, and for the hell of it I installed Battlefield 2. Just as I expected, no memory leak - there is no unreclaimed memory upon exit.
Also, how exactly are you teaching EA a lesson by giving them $60 and not playing the game?
Let's just say I got my money's worth before patch 1.12 was released. I played the game for almost 6 months. I wasn't teaching EA a lesson - no, this was an exercise in self-control. I don't upgrade just because a game suddenly stops working, not when my new PC is less than a year old.
I figured I would let it rest until my next upgrade, because BF2 is fairly popular, and the player base isn't going anywhere. I figured the game wouldn't have the same memory leak on different hardware - and what do you know? It works fine on the new machine, and feels fresh now that I put it on the shelf for a year.
THAT is the point I want to get across to these WoW players. You don't have to quit PERMANANTLY. But if you're sitting around on their servers just waiting for a miracle to happen, it's not gonna. MMORPG players are like breadcrumbs: once a few key players start leaving one game for another, others follow in droves, but until that happens nobody moves a muscle. The key is, someone has to have the willpower to take that first step.
I'm sick and tired of people whining about poor playing conditions on WoW, and having to wait hours or sometimes days to get in some play time. This ISN'T the first time I've seen WoW whining stories on the front page of Slashdot, and it only seems to be getting WORSE. These are issues that would KILL any other MMORPG, it is simply the massive inertia that keeps people from leaving. I love it that I was modded troll just for telling some whiner to either suck it up and take the beating from Blizzard (stop whining), or forge their own road and leave WoW.
Boo-hoo. You're spending $15 per month to wait an hour every night, and not play on Tuesdays at all. Woe is you! Save it for someone who cares. You've had a year and a half to realize things just aren't getting any better, and meanwhile you could have been pumping your money i
Exactly, here is a real-world example of a mid high-end X2 3800+ system built to be quiet. I recently built the following:
Athlon 64 X2 3800+ w/retail cooler (3600 RPM full speed, not exactly quiet).
GeForce 7900 GT (with a LOUD stock cooler)
Asus A8N5X (with an annoying 6000 RPM fan on the chipset).
To make this system quiet, I used the proper components, and voided a couple warranties:
* Antec Sonata II case w/450w powersupply. VERY quiet, stock ($100). Is priced competitively with other quality cases.
* Zalman vf900-Cu video card cooler, to replace the "miniature vacuum" stock cooler on the 7900 GT ($50). Alternatively, I could have used the vf700-Cu, or an Artic Cooling fan ($30).
* Zalman passive northbridge heatsink, to replace the 6000 RPM screamer ($5).
* Scythe 120MM 800 RPM liquid bearing fan, to be mounted on the front fan mount to cool the hard drives and the new passive heatsink ($15).
* Used the Q-fan feature of the ASUS motherboard to throttle the retail CPU fan. Even at max load it doesn't top 2400 RPM, and at idle it is a whisper at 1100 RPM ($0).
Total cost to make my system quiet (assuming you could have bought a crappy, loud case + powersupply instead for $50):
$120
Was it worth it? HELL YES.
My god.
It AMAZES me how much shit people will put up with to keep playing MMORPGs. If the system is broke that badly, why don't you leave? Why the hell do you think the problems keep getting WORSE?
I mean, I certainly wouldn't put up with that shit, especially if I was paying a monthly fee.
Hell, I stopped playing Battlefield 2 after they released the version 1.12 patch because it suddenly caused a memory leak on my computer. With only 1GB of RAM, it would crash to desktop after about 30-45 minutes. I like to play Battlefield for much longer stretches, so this was unacceptable.
I LOVE BF2. I *CRAVE* the experience...it's even more addicting than BF1942 was. But I quit playing, because I have standards. I am NOT going to upgrade to 2GB RAM just to "fix" THEIR problem.
QUIT the game. Let Blizzard have a chance to catch their breath, then maybe they'll resolve the issues.
Exactly. These days, there is no such thing as future-proofing.
Take the Athlon / Athlon 64 line, for instance. In the past, we've seen nice upgrades in bus speed (100, 133, 166, 200), and if you invested in a motherboard with faster bus speed support, and good enough ram, you could upgrade for cheap.
Socket 939, with the introduction of the X2 processor, also offered an opportunity for a (relatively) cheap upgrade from single core.
But where does that leave us now?
Over the past two years, AMD has gained only 800 MHz at the top-end, and this is all thanks to the tweaked 130nm, and later, the tweaked 90nm process. It sounds like a lot, but percentage-wise, it has not been impressive. Only the introduction of dual-cores has kept performance improvements in line with market demand.
Slow performance gains means one of two things: expand to quad-core processors, or redesign for better Instruction Per Clock (IPC) rates.
So what happens when we go quad-core? Not so much. Most multithreaded applications are designed to take best advantage of dual-core processors. Beyond that, there are only so many people who do constant hardcore multi-tasking, and even then you run into I/O limitations hindering multi-tasking gains. Also, unless AMD finds a way to magically pump up the HT bus and improve the performance of their dual-channel DDR2 controllers, the 4 cores will be data-starved. Quad-core is only going to find a foothold in servers, and ONLY if AMD releases better performing memory and HT busses.
So what does that leave us? Like Intel's Conroe, AMD will have to redesign the Athlon 64 for greater IPC, and like Intel, AMD will have to prioritize process improvements. Neither of these seem to be happening, so the future of "upgrades" on the AM2 platform is pretty bleak.
Me, I just bought a 3800+ X2 plus 2GB ram. The prices I've seen for AM2 X2s are exactly the same as those for Socket 939, and DDR 2GB is not any more expensive than performance DDR2 (ie, DDR2 800). I'll get more out of overclocking the 3800+ than I ever would purchasing an upgrade in the future.
Even if I didn't overclock, the performance gains of upgrading to a 2.6, 2.8 or maybe 3.0 GHz X2 a year or two down the road aren't exactly impressive.
Just a counter-point.
...instead of having ghastly shit like Spotlight foisted on me
Certainly, I have my opinion, and you have yours.
To say nothing of the nightmare of getting printers to work properly
I will say I had zero trouble setting up printing under OS X, and this was on an Intel Mac. Like Windows, I simply selected the printer from a list (or provide drivers), and enabled sharing.
I can say without a doubt in my mind that it was ten times easier than setting up the same printer in Linux via CUPS. After hours of crawling support forums, and my local Linux guru having a look, we discovered we had to replace lpr, because the default lpr included in Debian was some older BSD version that didn't support USB. At least, that's how he explained it to me.
It's funny, really...Apple uses CUPS, but that's because it ACTUALLY WORKS once you get rid of all the backend *NIX crap. It worked well under Linux too, once we wasted hours tracking down the lpr issue.
I also couldn't get my Matrox G400's OpenGL acceleration working in X under Debian (even with said local Linux guru), despite the fact that Mandrake configured it correctly out of the box. Tell me, if I'm following the FAQs to the letter, how the hell can one distro get it right the first time, and another distro with similar components leave me completely puzzled? Just more inconsistency to piss me off.
You can understand why I left. I had this kind of patience when I was a kid, sure. MSDOS was almost as insane a place as Linux in terms of tracking down settings, and inconsistent behavor, and I imagine if I had gotten started on Linux instead of DOS, I'd know it back-to-front. But now I'm getting older, have less time on my hands, and I don't have time to learn EVERYTHING...so I tried something simpler.
What's wrong with Spotlight? Honestly, I always though that locate and updatedb were decent, but clumsy, because I don't like opening a terminal window just to search for some files. Spotlight is right there on the bar, and even categorizes the data, something locate didn't do.
It's as fast as locate, has more meta-data, and you don't have to open a console. What's not to love?
or frigging menu bars that are ten miles across the screen from where I'm working,
Yeah, I'm not so sure I like this "one menu for all" thing either, but then I also must admit that, for those applications where I really make use of the menus, I usually keep those applications maximized, and that means no more movement to reach the menu than Windows or Linux.
Mind you, it does annoy me that some 3rd-party widgets actually have options hidden in the menu, which is utterly stupid. For instance, the option to cancel or run print jobs on my printer is embedded in a menu, not a checkbox or button on my printer widget. Is it Brother's fault for making such a boneheaded design decicion, or is it Apple's fault for allowing them to do it? Good question.
I will say that Apple sets a good example for third parties: I have yet to find a configuration widget that is part of the OS that has functions hidden in the menus.
Then I get to re-experience the same reason why I just recently left Linux for OS X.
Stuck inside their bubble, open source zealots think it's ok to have thousands of branches of THE SAME OS, COMPETING, INCOMPATIBLE WINDOWING LIBRARIES that aren't even standardized as part of the OS, and ten thousand window managers to make absolutely certain that your OS has no recognizable "look" or "personality" whatsoever. When they add new features, they add them in the "cleanest possible" manner (ie, make everyone patch and recompile EVERYTHING), rather than the "most usable" manner (add compatibility layers). When they change features, they don't do so gracefully, they break old code and expect everyone to recompile.
Here's just one example of how open source just gets it wrong: a few years ago, I was looking to play some emulators on my Linux box. I figured it would be as easy as emulation on Windows, but boy was I wrong.
See, I wanted to use the same USB gamepad I'd been using for the last few years on Windows. Only problem was, when Linux added support for *USB* gamepads, they used a different interface. Thus, emulators designed for *ANALOG* gamepads could not use my USB gamepad. Unfortunately, most of these emulators had been abandoned, and nobody had bothered to add USB gamepad support, so I was up shit creek unless I wanted to hack it in myself (sound familiar?). This is an example of adding a new feature CLEANLY, but in a manner that is completely UNUSABLE without extensive reworking.
I'm sick of it. It's little things like this that made using my Linux box for anything besides web browsing and basic office tasks a pain.
WINDOWS, by contrast, has supported USB gamepads since Windows 98, and has taken all the guesswork out of the issue. Regardless of whether you use an analog or USB gamepad, an application can use the same hooks to communicate with the pad regardless, and the user can use the same setup widget to configure ALL pads. Now, that's not going to be very CLEAN code, but it is a damn sight more USABLE. I can't speak for OS X on this issue, simply because I've only been using it a few weeks.
This is just one example of why I don't need yet another open-source operating system. Open source applications are just fine - the ones that are mismanaged or get caught up in their own self-image eventually get replaced by better development projects. But operating systems have momentum, and don't just disappear. Open source zealots could do a lot of damage to the USABLILITY of OS X, and it would still take a long tome for it to fade away.
I used to be a Matrox junkie. I loved the 2D quality. I loved the high performance of the G400 series. But Matrox took too long to relase a followup.
The G450 was late and unimpressive due to the low clock speed and 64-bit unoptimized memory bus.
The G550 was late and unimpressive due once again to the low clock speed and 64-bit unoptimized memory bus.
This whole time, Matrox was sitting on their ass, not improving on the DualBus originally released with the G400. I knew what was going on thanks to MURC (Matrox User's Resource Center, no longer exists), which was frequented by disgruntled current / former Matrox employees. They said that after the G400, Matrox management did not have a vision, and as a result their best digital and analog folks left. The recovery time is why the G450/550 were so late, and not at all innovative Matrox never regained their G400-era fire.
Meanwhile, ATI and Nvidia were learning that memory bandwidth was becoming a severe performance limiter, and out of necessity they learned tricks like texture and Z compression, and other memory bandwidth savers. ATI and Nvidia also switched from wide 128-bit unified memory busses to crossbar memory architectures to improve overall bandwidth utilization.
I didn't wait around. I bought a Radeon 8500 to replace my G400 MAX about 6 months before the Parhelia was released.
When I saw the Parhelia, I knew I had made the right decision. Matrox used the same old bus technology they had been relying on for years, and upgraded it to 256-bit wide to try to make up for their technology deficit. It performed a little better than cards with 128-bit busses, but wasn't all that impressive.
When the 9700 Pro was released with a 256-bit bus, and ran circles around the Parhelia, it dawned on people just how bad Matrox's memory tech was. Fragment AA was a nice carrot to attract people, but it wasn't enough to make up for the car's incredibly poor performance.
The thing is, Apple supplements their boring, typical hardware by releasing unique devices that are ahead of their competition.
Just look at the Mac Mini. When it was released, NO vendor had a desktop anything like it, and even today the Intel Mac Mini is competitively priced with Intel Core-based knockoffs. The difference is, the Mac Mini is ESTABLISHED thanks to Apple taking a risk, and the rest are scrabbling for marketshare.
The original Ipod was smaller than anything else, but offered the capacity of a hard drive player. It was ungodly expensive, but it became rooted because the industry took too damn long to respond to the threat.
The Nano is another recent product that still has no competition. I don't know how, but Apple managed to get a deal for flash devices twice the capacity of their competitors (32Gb, while the rest of the industry was on 8Gb). The result was the 4GB Nano, with room for Apple to upgrade the line to 8GB in the future (once the 32Gb chip becomes cheap).
The fact that the Nano 2GB was priced at only $10-30 more than most other 1GB players shows just how lazy and wussified the whole industry is outside of Apple. Apple set the standard for pricing of 1GB players with the release of the Shuffle, but then instead of the industry pushing prices down and capacity up, they coasted while charging $10-30 less than an equivilant Shuffle. Then, when the 2GB Nano was released for $10-30 more than a 1GB player (despite the fact that the Nano used TWO 8Gb chips), was still competitively small AND had a smashing screen, the rest of the industry looked painfully stupid.
THIS is why Apple owns the industry...because they're the first manufacturer since DIAMOND MULTIMEDIA's RIO PMP to really PUSH the industry.
I have purchased Creative Labs Muvo players in the past because I WANT drag & drop without having to use a media library, but with Rockbox now I can turn any Ipod into exactly the interface I want. Suddenly, I don't have to compromise on larger size and less capacity for about the same price just to get some key features I need...Rockbox makes that possible.
Full Disclosure: I bought my first apple product EVER last week, a Mac Mini. The OS is decent, but I could care less about the apps. I don't like Safari, so I have Firefox. I don't want to be forced to use a media library with Itunes, so I went through the trouble of installing XMMS. I bought the Mini because it was tiny, VERY quiet at nominal room temperature, competitively priced with other small Intel Core solutions. It also offered me the chance to try OS X, but fall back to Windows or Linux if I really hated the experience...that's something no other small Core box could offer.
If the industry continues to lag Apple concerning small flash players, I may purchase a Nano or equivilant sometime in the future.
And a good thing, too.
Don't get me wrong, Rosetta is decent (for an emulator), but Firefox has just been ungodly slow on my Intel Mac Mini. I've been waiting for an official Intel build.
Hmm. Your ideas are intriguing to me and I wish to subscribe to your newsletter.
It's been almost two full years since the consoles were in their swing, it has even been over a year and a half since big PC games like WoW, HL2, Doom 3, etc were released.
The industry has nothing BIG. Oblivion is decent, but it doesn't have the same momentum as a wider-audience game series like Quake / Doom / UT / (War)(Star)craft, and the cost of entry is VERY high (either an Xbox 360 or a relatively well-specced PC). With the unimpresive, high-priced start of the Xbox 360, and with a similarly high-priced start expected for the PS3, what does your average gamer (who can't afford or justify a 360) have to look forward to? Interesting new releases on existing consoles are rare compared to a couple years ago.
To put this in perspective: this is the second generation in a row where mainstream gaming system prices have gone UP significantly. Before the PS2 / Xbox, most gaming systems debuted at around $200-250. When you took inflation into account, that meant the system was always cheaper than the previous generation.
But the PS2 / Xbox pushed that envelope to $300, and the Xbox 360 and PS3 are pushing that number to $400 or more (yes, you can buy a $300 Xbox 360, but it will really cost you $340 with the official memory card, and well over the $400 mark if you add a hard drive in the future). Gaming is getting significantly more expensive, and it's no wonder people are playing less games. People have done the same thing in response to high prices, say, in theaters.
That's a very good question. I'm getting a Mac Mini delivered on Saturday, and I may be able to try this out during the week.
My guess is it only works with XP specifically.
That would be due to several "lessons learned" as Intel developed Itanium.
1. The instruction overhead due to extra hint bits, etc, means Itanium instructions are much larger than x86 32/64 instructions. With the addition of poor branch performance (read: more wasted instruction bandwidth), the need for large, high-bandwidth caches makes Itanium expensive.
2. The compilers have not caught up. EPIC lacks OOOE, and has poor dynamic branch prediction hardware, so it is at the mercy of the compiler.
Core retains Intel's original insights made with the P6:
1. x86 is hard to decode (takes more silicon), but it takes less bandwidth than other instruction formats. Bandwidth is even more expensive than the cost of more complex decoders, just look how expensive it was for Intel to add full-speed cache to the original Pentium Pro, and how pricey the Itanium is with huge, fast on-chip cache.
2. OOOE + Branch Prediction + internal RISC is king. One reason the original Pentium never performed well is because it could RARELY execute more than one instruction per cycle. Thus, it performed like a fast 486 unless the code was recompiled as Pentium optimzed. The P6 was designed to avoid the reliance on compilers to improve performance, as it could optimize code in any condition. Funny, we didn't start seeing Pentium-optimized code on the market until the P6 started taking over.
Core is just a logical extension of this concept. The predictor is more accurate, there are more instruction decoders, more ALUs and SSE units, and more retirement units. The only reason Core seems to groundbreaking is because we didn't see it in small, evolutionary steps.
"Just one addition, branch instrucctions don't hold up processors, memory load and floating point instructions do. The problem with the branch instructions is that the branch decision may come too late, and the penalty depends on the depth of the pipeline."
No. For someone who claims to be educated in computer architecture, you are sadly lacking in understanding. Unfortunately, all you're thinking about is the mispredict penalty.
Floating-point multiplies are FAST these days. Memory load issues can hold up a pipeline, especially if the cache heiarchy has not been designed for high hit rates (too low capacity, bad cache block design, etc), but that's not typically the case these days.
Branches are one of the biggest issues for the x86 32/64 platform to handle because of the non-uniform instruction size, and because ACCURATE branch predictors are SLOW (multi-cycle). What this means is instruction decode + prediction ends up taking multiple cycles, and during those cycles the processor keeps reading instructions as if a branch wasn't taken. By the time you have a branch decoded, fed into the accurate branch predictor, and predict taken, what do you do with all the instructions in the decode pipeline?
You flush them. That's wasted fetch bandwidth, plus you have bubbles in the execution pipelines while the fetch pipes refill. THAT IS COSTLY.
The Pentium 4 uses a trace cache to get around this issue. It keeps a record of all instructions executed in actual branch order, unfragmented, based on their previous branches. Unfortunately, there is a big hit if the trace cache mispredicts or doesn't contain your instructions, because then you have to go to L2 cache. It's not optimal, but it does work well.
The optimal solution? You have FAST but inaccurate branch predictors that can work in a single cycle, and you make fast decicions about branches. You also have to predict the branch target address using a branch target buffer, because calculating it would take far too long. Thus, you can keep fetching immediately, even if the branch is "taken".
The final two pieces to this system are a pipelined accurate branch predictor that can override the fast predictor, and of course, the real branch verification hardware (which causes the pipeline flush you commented on if the direction predictor is wrong). This system guarantees you much less instruction pipe stalls while you wait on an accurate branch prediction. Unfortunately, it is expensive to implement, so most processors use some sort of compromise system (like the Pentium 4).
Right, there are two camps for the "high-end" branch prediction concept:
Camp 1: devise adaptive, multi-component prediction systems that offer both fast and accurate branch prediction. Waste hardware purely for branch prediction.
Camp 2: Use the compiler hint if available, otherwise execute both paths, and throw away the incorrect processing path. It seems cheaper on the surface, but you have to realize: all that extra fetching to process both paths in reasonable time mean more fetch bandwidth and more execution units required just to keep up.
Obviously, if your code contains lots of branches that cannot be predicted by the compiler hints, the Camp 2 solution is going to perform worse. The advantage of active branch prediction is that you never have to recompile the code to keep the branch hints "optimized" if your datasets change.
It doesn't really matter which camp you choose, because both camps waste space on a Branch Target Buffer (predicts the TARGET of the branch) anyway, and that's often more costly than the branch direction predictor. Even the Itanium has a BTB, that's how it can instantly start executing the "branch taken" case.
The Itanium is just taking advantage of a serious architectural flaw to perform branch prediction. Even modern compilers are inserting 20% or more "noops" into the instruction stream, why not take advantage of that underutilization. On any other platform, it would be a very stupid approach to branch prediction.
"1T" SRAM is DRAM with an SRAM front-end cache (thus, the "1T" name...DRAM requires one transistor per cell). Normal SRAM is "6T," which indicates it requires 6 transistors per cell.
So long as the front-end cache is large enough, the latency is close to that of a real SRAM cache. The benefit is "1T" SRAM uses a lot less die real-estate, and a lot less power.
Sure, it is an x86 processor, based on the Winchip.
Let me just put this in perspective for you:
For anything other than MPEG2 or MPEG4 viewing, or encryption, this processor will perform like a PII 350 or slower. Imagine running Gnome or KDE on a PII 350 with a TINY (64k) L2 cache, and you'll come to understand just how much this chip sucks.
Via claims that their philosophy is low-power computing, but the real deal is Via does not want to pay their design team to completely redesign the Winchip. The Winchip is not even a real superscalar processor: it only has one integer unit, and one floating-point unit. Think of it as a highly-optimized 486, with performance much closer to 1 Instruction Per Clock.
It wasn't a bad design way back when the Pentium ruled the roost, because the Pentium typically executed one instruction per cycle. It was pretty rare to keep both instruction pipes on the Pentium filled, even with a decent optimized compiler. But then the P6 came along with support for out-of-order execution, making the Winchip obsolete overnight.
The sad thing is, modern processors like the Athlon 64 and Pentium M can compete with the Nehemiah on power usage AND performance. You can take any Athlon 64, clock it to 1.0 GHz and 0.8-1.0v, and it will have twice of the performance of any Nehemiah processor and use less power. You can do even better with the Pentium M, which supports voltages as low as 0.7v.
Your noisy Opterons, for instance, can probably be underclocked and undervolted, and you can potentially cool them passively (if they are of the 90nm variety).
They figure you'll look up that information, their biggest concern was spacing out a three page review over ten pages.
That said, Via normally names their boards with a number that is 10x the CPU speed. The board is named the N8000, so that would indicate 800 MHz, like all the 130nm passive MiniITX boards. The benchmarks agree.
It's really quite sad that NanoITX only uses the old crappy CN400 chipset with the older 130nm Nehemiah core. I mean, the 90nm rev still sucks, but at least they can clock it higher, and at least the bus isn't holding it back. You would think after all this delay they'd be able to put their new C7 platform in NanoITX form-factor.
800 MHz on a CN400 chipset is about the performance of a PII 350.
Not THAT much. You can eye the power consumtion results and know that Hot Hardware did not enable Cool 'n Quiet. Otherwise, there should have been a 50-60w difference between idle and load.
The Intel chip, on the other hand, has Enhanced Speedstep enabled. You can tell this by the fact that they mention the prcessor's voltage is from 1.2v to 1.35v. With Speedstep, the processor at idle goes to 2.8 GHz at 1.2v. AMD's Cool 'n Quiet is even more drastic, dropping the processor speed to 1 GHz at 1.1v.
The fact that AMD can ALMOST match the power consumption of Intel using Enhanced Speedstep, while not using Cool 'n Quiet, shows you just how much better the Athlon 64 is.
Hey, I agree wholehartedly, and I've been an AMD fanboy since the release of the P4.
Anyone who can look at this breakdown of the new Core design, understand it, and STILL proclaim AMD the performance leader is retarded. The extra simple decoder means potentially 33% more thoroughput out the gate, and the fused micro-ops can add another 5-10% performance improvement (assuming you have enough execution units to use all this). The 128-bit SSE unit, plus the ability for simple decoders to handle packed SSE instructions, also means double the speed at vector operations.
That said, at least I had my just desserts. I always said superpipelined Netburst was a retarded design, and the fact that Intel went and developed Conroe only validates my claim.
I am still curious to see the power usage of Core. It should be less than the P4, but whether it is competitive with AMD may be another story. Hopefully AMD will finally get off their ass and improve their own design, which hasn't changed much since the K7 (onboard memory controller aside). Who knows, I may end up buying Conroe, and becomean Intel fanboy again.
Yeah. It's not just the chip, it has a PCI bus. But you knew that.
Because the Pentium 4 was never made for powerful floating-point thoroughput using plain old x87 FP instructions. To give it better floating-point performance, a 128-bit vector unit (SIMD) was strapped on. With it, the Pentium 4 can perform 4 32-bit floating point operations every 2 cycles.
This is the one part of the Pentium 4 chip that is very much like the parallel execution units of a video card. If you want to maximize performance of the SIMD unit, you want a simple program with packed SIMD instructions. The key point is, the video card can do that better.
I just finished talking to a "normal" gamer buddy of mine today. You know, he likes to game, and his kids like to game. He was telling me about how much he and his kids love Call of Duty 2, and how smooth and amazing it looks on a Radeon 9200. As I sat there fighting to keep my mouth closed, he went on to mention he was buying a gaming PC "just for him," with a fancy new GeForce 6600 (no, not the 6600 GT).
I almost laughed at him, but then I realized: he doesn't care about fancy TRMSAA or Angle-dependent Anistropy, he doesn't really notice fancy real-time lights or shadow-maps. The only thing that matters to him and his kids is the game looks decent (ie, at least 640x480 resolution, reasonable resolution textures, and billinear filtering).
And why not? A whole half of the gaming population bought into the PlayStation all because it could do fancy new 3D rendering, even though the unfiltered textures and 100 polygons per scene looked like crap. Most of them traded up to the Playstation 2 because it could do much better polygonal detail and allowed higher video and texture resolutions, plus DVD.
So, here's the problem: most people think that the PS2 looks great, and it is in fact almost as good as you can get on a standard TV. So, why should people upgrade when the current console does everything they think they need?
If you don't have an HDTV, the PS3 doesn't really offer anything over the PS2 except Bluray, and again, the DEMAND for Bluray depends on you having an HDTV. All the selling points for the PS3 come back to this HDTV thing.
And normal people don't care about HD. Take my sister and her husband: they have Sony everything, a nice Sony DLP widescreen, plus a matching Sony 5.1 receiver and speaker set. It's great, except most of the things they watch are in SD, and the speakers are set up in two clumps on either side of the room. The closest they get to watching HD is through their DVD player, but I know better than to argue with them.
Look at the XBox 360. IT LOOKS MUCH BETTER THAN PREVIOUS GENERATION CONSOLES, there is no doubt of that. The problem is, it doesn't look GOOD ENOUGH to warrant the upgrade, so people stick with their current-generation consoles and games. You can be sure that the PS3 will encounter the same problem.
I mean, people convince themselves all the time that Sony equipment is somehow "better" than other brands, but what happens when it's Sony versus Sony? I expect all those eagar PS2 buyers won't be so eagar to go PS3. The only console to offer anything really DIFFERENT is the Revolution, but that gives it as much risk of sinking as swimming. One thing I am sure of: console sales this generation, overall, will be lower than last generation.