Athlon Xp 3200+ 400FSB is Coming

SoDaLaS writes "Athlon 3200+ Coming: According to CNET The Athlon 3200+ with a 400MHz FSB is on the way in the next two weeks. It'll be interesting to see how well the processor overclocks at that high of a bus speed...it didn't seem to hamper the new 800MHz FSB Pentium 4, which many people were worried about too."

Comments

400 MHz, 800 MHz

[Gothmolly]

Lets make sure we're comparing apples to apples. The 400 MHz bus on the Athlon is a DDR doublepumped bus, so its really 200 MHz. The 800 MHz FSB on the P4 is a quadpumped bus, so its really 200 MHz.

Re:400 MHz, 800 MHz

[EvilBudMan]

We'll yes and no, It's still going to be pretty fast.

Re:400 MHz, 800 MHz

[VAXman]

Yes, and the peak bandwidth of the double-pumped 200 MHz Athlon bus is 3.2 GB/s, and the peak bandwidth of the quad-pumped 200 MHz P4 bus is 6.4 GB/s.

So factoring the double/pumped into frequency gives the more realistic performance picture.

Re:400 MHz, 800 MHz

[mrtroy]

Not really. Thats like saying a RAID is slow because each individual drive is slow. Thats like saying a car with a turbocharger still has the same horsepower as a car without. Clearly, this is poor reasoning.

Re:400 MHz, 800 MHz

[AccUser]

Why compare apples with apples? Why not apples with pcs? After all, the Athlon is built around the x86 architecture... :-P

Re:400 MHz, 800 MHz

[GigsVT]

Yes, but calling it "Mhz" is misleading. That's like saying the PCI bus runs at 1056Mhz because it is 32 lines running at 33Mhz.

We already have enough misleading and confusing computer terms, we don't need to add another one.

Re:400 MHz, 800 MHz

[arvindn]

In that case, wouldn't the AMD thing to do be to call it the "800+" FSB? :)

Re:400 MHz, 800 MHz

[GigsVT]

Mhz is NOT a measure of throughput or performance. It has a very specific definition, Million cycles per second. Don't let them confuse the term and water it down until it is meaningless.

Re:400 MHz, 800 MHz

[Anonymous Coward]

Does it give the more realistic picture? I'm not a hardware designer or anything, but I've always been under the strong impression that the front-side bus speed mainly matters when you have a memory access. In that case, what really is going to make a difference isn't bandwidth. Instead, it's latency.

I'm assuming that when something isn't in the cache, the processor is going to read a whole line at a time, but will start with the word actually needed and then fill the rest of the cache line after that. In that case, the processor only has to sit and wait while just one word is fetched from RAM. So what really matters is not how fast you can stream in multiple words, but how fast you can read in one word. (Of course, there are times when streaming in lots of data is helpful, such as when the processor isn't mainly hitting its cache but instead is copying a large block of memory...)

As I said, I am not an expert, and there is always the possibility that doubling or quadrupling the data rate will reduce the latency of memory reads as well as increasing the bandwidth. If so, I'd love to be enlightened about this topic...

Re:400 MHz, 800 MHz

[MentlFlos]

Where is

+1, Bad Joke... BOOOOOOO

when you need it?

Re:400 MHz, 800 MHz

[drinkypoo]

While it may be misleading, it is less misleading than, say, PR numbers given for AMD CPU speeds. At least this is actually based on the number of clocks per second.

Re:400 MHz, 800 MHz

[jandrese]

I don't know about you, but I find the PR numbers to be quite handy for comparison purposes. It's certainly a better metric than Mhz.

Re:400 MHz, 800 MHz

[drinkypoo]

I don't use PR numbers or MHz ratings when comparing CPUs. I visit a review site and peer at many images containing various colored bars and lines. I especially check out the multimedia content creation benchmarks, which represent the most CPU-intensive operations I generally apply my CPU to, and the gaming benchmarks, which are the second most CPU-intensive things... Well, except maybe Unreal Tournament, that engine has always been terribly CPU-bound.

Re:400 MHz, 800 MHz

[Anonymous Coward]

What a retarded post. Using your reasoning:

The engine in a 911 Carrera is a normally aspirated H6 with 320HP. The engine in a 911 Turbo is the same H6 turbocharged to 456HP, so its really 320HP.

Re:400 MHz, 800 MHz

[CTho9305]

There are multiple phase-shifted clocks. These clocks run at 200MHz, but features of interest (edges) occur at 400 or 800 million times per second. The end result is functionally indistinguishable from an 800MHz clock.

Re:400 MHz, 800 MHz

[Ark42]

That peak bandwidth assumes mostly sequential reads. In reality the latency figure is based around the 200MHz clock and probably more important towards overall performance.

Re:400 MHz, 800 MHz

[Anonymous Coward]

Its not misleading. The data transfer rate of the FSB is 800MHz. Hz = cycles/second, this can apply to many things besides clocks. Saying that the bus is 200MHz is more misleading. Consumers should see a number that represents real performance.

Re:400 MHz, 800 MHz

[Anonymous Coward]

That must be why the 800MHz FSB performance comparisons made such a poor showing. /sarcasm

Oh, you said 'probably more important'. I guess you were probably wrong.

The numbers show the 800MHz FSB as the hands down performance winner. No questions, it wasn't even close. Look up the reviews on tomshardware and sharkeyextreme.

Re:400 MHz, 800 MHz

[Bedouin X]

And interestingly enough, UT is the benchmark that is owned by the slower (compared to P4) Athlons.

Re:400 MHz, 800 MHz

[BlowChunx]

Sure, memory latency has an effect. That's overhead to get to the first word that's needed by your cache.

Then bandwidth is needed because on your processor, we have to fill the rest of the cache line. It would be painful to have low latency (OOhhh! we get the first word quickly), and low bandwidth (dang, now we have to wait to fill the rest of the cache line...).

Re:400 MHz, 800 MHz

[polarkittycat]

Your right, MHz. is mostly meaningless. But as far as the bus goes look at it this way.

AMD bus = 200Mhz. x 2 bits per cycle = 3.2GB/s bandwidth
Intel bus = 200Mhz. x 4 bits per cycle = 6.4GB/s bandwidth

You cannot argue that Intel doesn't have an advantage here, having twice the memory bandwidth of the AMD chip.

Re:400 MHz, 800 MHz

[alienmole]

I would say it's potentially confusing, but not necessarily misleading, if you put it in context. I think it's reasonable to talk about "memory access frequency" in terms of MHz, i.e. memory access cycles per second.

The "cycles" are logical, in that electrically they're actually half-cycles, but if you want to do calculations involving memory bandwidth, MHz is actually the right unit - since the multiplier (2) has no natural unit.

Re:400 MHz, 800 MHz

[GigsVT]

Well, as I understand it, the "quad pumped" intel is done by actually doubling the number of lines, and both sets run with data on the rise and fall of the clock. This is the part I think is most misleading, and the biggest abuse of the word.

Re:400 MHz, 800 MHz

[alienmole]

You're right, they've effectively increased the bus width, but they're folding this into the MHz number. I think there's some logic to that, though. It makes some sense to stick to talking about the memory bus width as 64 bits, because that's true even with QDR, it's just that there are two 64-bit ports (read and write). You could describe the situation as:

200 MHz clock x 2 accesses per port per clock cycle x 2 ports

...but if you multiply that out, taking the above units into account, you get 800M accesses/sec, i.e. a "memory access frequency" of 800MHz. The point being that once you've made the choice to stick to talking about a 64-bit bus, and to quote a single figure for the effective memory bus frequency, MHz is a valid unit in that case.

I'm not saying it's ideal, but I think there are both technical and marketing reasons for the choice - it's not totally off-base.

Re:400 MHz, 800 MHz

[let_freedom_ring]

"So what really matters is not how fast you can stream in multiple words, but how fast you can read in one word"

No, it matters how fast you can stream. Most data i/o from the memory to the processor cache are instructions rather than user data. When your processor jumps to a new routine (or an ISR) and the cache is cold there will be a high level of latency for the first instruction but the rest of the routine will be streamed into cache fairly quickly. This is why high bandwidth is very important and why latency can be tolerated.

Finally....

[DarkBlackFox]

AMD can take advantage of DDR 400 for synchronous system performance. Expanded front side bus + more work per clock cycle= damn good performance. Great stuff.

Re:Finally....

[sigep_ohio]

I would think that it isn't going to do much more than extend the Athlon's lead in some areas(business content) and close an already large gap in others(multimedia) between the P4.

But it is nice to see that AMD moved the Athlon to the 400MHz bus. Now hopefully they will give the Athlon64 the same support, along with a dual-channel memory controller like in the Opteron.

Re:Finally....

[Omega's Wildfire]

I am happy to see the FSB go up as well. If the chip will only over clock by a small margin, then they could have simply just increased the quality control of the chips to ensure function at the higher bus speed. Feel free to correct me if I am wrong, but isn't this just a minor improvement which the chip houses have been doing for years. I want to see the 800 FSB on the AMD side. :)

Re:Finally....

[shamilton]

Answer me this, where did this whole notion that P4s are somehow faster than AMD chips come from?

Canadian dollars:

1800, AMD: $95, Intel: $215
2400, AMD: $160, Intel: $240
2800, AMD: $349, Intel: $499

That is, comparing the 1800+ to a 1.8 P4, etc. And yes, those AMD readings are usually pretty conservative. If you compare performance per tic, AMD continues to beat the living crap out of Intel, has since the K7, and likely will continue to do so for generations yet.

Point 1: If you want a cheap CPU, an 1800+ for a hundred bucks (60 USD) is a damn fine deal. If you want to be loyal to Intel, that will buy you a 1.7 Celeron, which is comparable to a T-Bird 1333 of two years back.

Point 2: If you want to shell out, then you are getting more bang for your buck by buying a high end AMD, although this point is a bit weaker as they tend to get closer in price.

Point 3: If you REALLY want to shell out some coin, you could buy one of the really high end Intels, which pull ahead of AMD chips due to lack of an AMD offering in the same range. But then, if you are going to shell out, why not purchase a Xeon, or dual AMDs, or Sun hardware, or a data processing centre to run Quake? The scale is only relevant where you can actually compare the two, so in my opinion, this point is moot.

This can all be explained if you consider Intel isn't so interested in making a great processor as it is making great fabrication processes, and patenting them. The processor is more of a testbed. Much like how id is mostly a technology company, but Carmack has said if they didn't make a game, they would end up missing things in the engine.

Re:Finally....

[sigep_ohio]

Ok, chill. Breathe.

I was really only refering to the Performance differences between the AMD 3000+ vs Intel P4 3.06GHz. These processors are about the same price, and like I said perform similarly. P4 wins some, Athlon wins some. I wasn't bashing anyone, let alone a lowly processor.

Your right lower priced Athlon's cost less and perform better than comparable priced intel alternatives. Unfortunately at the top of the line, there is no real price difference in processors. And in reality there isn't much of a performance difference, overall. But in terms of individual tests, P4's excell in some areas while Athlons excell in others. Why do you need to jump down my throat over it. Blame Intel and AMD for building their processors so differently.

Re:Finally....

[shamilton]

My reply was not so much directed at you as anyone who assumes there is an "already large gap" between AMD and Intel chips. There isn't.

Re:Finally....

[yomegaman]

You are totally missing the point with this "performance per tic" business. The P3 beats the P4 in efficiency also, but so what? The P4 trades off efficiency for the ability to run at higher clock rates, and it was a good choice since the fastest P4 beats anything out there in integer performance (including the expensive 64-bit RISC chips, and the Itanium2) and has perfectly respectable floating-point performance.

Really, the AMD fanboys are starting to remind me of Apple zealots, always blathering about how a 3HGz G4 would crush a 3GHz P4. If we restrict ourselves to discussing parts that actually exist then there's no question that the P4 is fastest consumer-level CPU you can get.

Re:Finally....

[shamilton]

I think it's more likely the less-per-tic thing was marketing. At the very least it has forced AMD to use their "comparable-to" reading, which just makes them look bad to the uninformed, which is most consumers.

As for the G4, the point I am stressing is not that an AMD is faster per tic, but that is faster per dollar -- certainly not the case with the G4. (A coworker of mine recently paid $300 to upgrade his G3 300 to a G4 500... and noticed no difference.)

Re:Finally....

[ProtonMotiveForce]

1. The performance readings are _not_ conservative. In real world performance (and modern benchmarks) the 3.06GHz P4 will trounce the 3200+ AMD. The 3.0 certainly beat the 3000+ in the majority of the modern benchmarks I've seen.

2. At the high end (i.e. you spend a little dough) an Intel based system will always outperform the AMD.

All in all, you're spewing typical contrarian garbage. It used to be that Intel sucked because they had this CISC architecture that didn't scale, and had a high IPC but alow clock - RISC ruled! Now that the shoe's on the other foot, and Intel has a high clock and AMD now has a better IPC AMD rules!

Fact is, Intel chips perform better - end of story. IPC is meaningless.

People just love to hate the one at the top, and they just plain love an underdog.

Hence the fanaticism you see towards AMD and Linux (though Linux does rule, I admit).

Re:Finally....

[workindev]

Answer me this, where did this whole notion that P4s are somehow faster than AMD chips come from?

You might want to check this out. Find me a single benchmark that AMD took the crown in. You will soon discover that the closest AMD got to winning a benchmark was 2nd place on the Sysmark Office 2000 test. The rest of the tests placed AMD in 4th, 5th, and sometimes even 6th place (behind the slowest P4 in the test, the 2.53GHz).

This can all be explained if you consider Intel isn't so interested in making a great processor as it is making great fabrication processes

So basically, even though Intel isn't interested in making a great processor, they still make one that is superior to what you are promoting?

The only fair comparison would be a dollar amount comparison (a $200 AMD processor vs a $200 Intel processor), and Intel still takes the crown with that. An XP 2800+ is about the same price as a Pentium 4 2.6Ghz, and the benchmarks still show Intel ahead on majority of the tests. The only thing we have to thank AMD for is the fact that we don't pay $5,000 for a superior Pentium 4 now.

Re:Finally....

[PMuse]

To present this argument convincingly, you might try a graph of benchmark score (y) v. price (x). Ignore the clock speed. If what you're saying is correct, AMD should show a higher performance rating at most or all price points.

The question to answer for people is this: if they have $175 to spend on a processor, what can they get for it?

Re:Finally....

[Uberbah]

I haven't looked at the entire review, but I smell the wiff of bullshit comming from Tom. First I checked out the game benchmarks...where he only ran Quake 3!!! If you didn't know, Quake 3 has always run faster on P4's than on Athalons. To be fair, he should have also benched something like Serious Sam 2 which has a similar advantage on AMD chips. I would be similarily suspicious of his other benchmarks, like mp3 and mpeg encoding. If he knowingly uses a game that's better optimized for one processor in a benchmark, whats to stop him from doing the same in others?

Remember that Tom plays favorites with Intel and AMD, depending on who's giving him more free stuff these days. A year or two ago he was loudly claiming there was no way Intel would be able to compete with AMD and that they'd end up exiting the consumer processor business.

So basically, to paraphrase: "there are lies, there are damn lies, and then there's benchmarks"

Re:Finally....

[DinZy]

This isn't even an upgrade. Most AMD chips can do 200 FSB now, or 400 if you count like they do. They are just going to officially support the speeds that people have been getting for some time now. Its actually the chipset manufactureres that made it possible. Some people have been able to get T Birds to hit 200+ FSB on nForce 2 motherboards.

Re:Finally....

[workindev]

Fair enough -- benchmarks can certainly be tweeked. But I still have yet to see any comprehensive review (Toms, Anandtech, Hardware Central, anywhere) that has declared AMD the speed king. Intel just seems to be flat out faster on an overwhelmingly majority of the benchmarks.

Re:Finally....

[Anonymous Coward]

----
1800, AMD: $95, Intel: $215
2400, AMD: $160, Intel: $240
2800, AMD: $349, Intel: $499

That is, comparing the 1800+ to a 1.8 P4, etc.
----

Wow are you this stupid? First of all, an 1800+ Athalon is NOT 1.8 GHZ. You're wonderful AMD tricks morons like you with this marketing technique. So comparing the 1800+ to a 1.8ghz is not really accurate, but evenso you are still wrong.

Secondly, your prices are flawed, and no it's not just because you are canadian. Let's look at the Athlon XP 2800+

AMD 2800 (2250mhz 333FSB) - $215 INTEL 2.26ghz 533FSB - $166

Hmm, which one is cheaper now? Gasp! Intel!

For the same price as your AMD 2800 which is only 2.25ghz at 333FSB you could get a 2.6ghz 800FSB with HyperThreading for $203, $12 cheaper than the AMD one and much faster.

Athalon XP 3000+ cock in around 2.4ghz and cost $317. Again, a 2.6 and 2.8ghz Intel processors with 800mhz FSB and HyperThreading are --cheaper-- and outperform the XP 3000.

So which is the better value again?

Next time try and get your facts straight, and atleast keep them within the threshold of this reality.

Re:Finally....

[Captain Pedantic]

Let me guess: you've recently bought an Intel chip (a 2 ghz Celeron maybe?) and you want to convince yourself that Intel are still the best?

Re:Finally....

[juhaz]

Dunno where Tom takes his numbers, or whether or not there is something fishy about his methods, I prefer Anandtech for reviews.

XP 3000+ review would indicate that in most of the things it indeed does lose to P4 3.0 (though the differences are rather small), but XP2800+ beats P4 2.66 in 15 tests out of 22. And price/power becomes even more clear on slower speeds, 2100+ outperforms 2.2GHz P4 and is about $60 cheaper. "Bang for the buck" argument stands.

Re:Finally....

[Slack3r78]

You won't have seen them because they're not. I'd agree that AMD's prorating can't be described as conservative, but it's generally in the right ballpark. I've never understood why people act shocked when a P4 does manage to beat a prorated AMD chip on a given benchmark -- the AMD chip is running considerably slower, it should be expected.

I've been a bit leery of the prorating system since AMD put it in place, and there have been times that they've had to tweak it because it's been obviously off. But I would certainly say that "3000+" certainly gives you a better idea of how the chip is going to perform in comparison to a P4 than it's true 2.17GHz speed is going to. (that number's offhand, don't shoot me if I'm a little off :)) I think people forget that people forget the whole prorating system is meant to be a rough guide, and this is probably because PR decided to spin it in their own special way.

And as I side note, I realize THG is a popular portal for hardware news/reviews, but personally, I just can't bring myself to trust any of the info that comes from it. There've been too many times I've seen his testing methods and conclusions vary quite badly from other reviewers' for me to take him seriously any more. Honestly, I didn't check the link, but skewing the test to favor a given side would be par for the course for THG.

Re:Finally....

[Slack3r78]

I'm not an expert on the subject, but I believe that you're more or less right. What's interesting to me though is the fact that AMD has only been on a 333MHz FSB for about 6 months, and they're already bumping to 400Mhz. Remember that they spent, what, almost 2 years at 266MHz before that? Logically, my guess would be that the new Barton core is able to handle the higher bus speed easier than the older cores could. Again, I don't have any hard data, just trying to reason this out.

To me, this isn't really exciting so much because it's on the new 3200+ so much as it means that the Athlon 64 could quite likely ship with a memory controller set up to run at 400MHz instead of the 333MHz we've been expecting up to this point. Remember, the memory controller is built-in to the CPU on the Hammer chips, so the concept of FSB doesn't really apply apples-to-apples, but the idea is still to have the CPU and memory running at the same clock. This means faster memory, which means better performance still. x86-64 just keeps getting sexier all the time. =)

Re:Finally....

[Duds]

Well, last time I did a quick price calc here in the UK when I was considering an upgrade.

MSI board + P4 2.4 + 512MB of appropriate RAM
MSI board + Athlon 2400+ and 512MB of appropriate RAM

Same price.

The Athlon JUST matches it doing most things. You try clever video encoding etc and that Athlon performs at its clock speed, not its PR.

For someone that's bought AMD and only AMD since my K6-2-300, that was a shock.

Overclocking

[sigep_ohio]

Unless they shrunk the Athlon core, I don't see a lot fo room for overclocking. The 3000+ isn't an overclocking dream, so simply moving to a faster bus ain't gonna make the 3200+ any better.

Re:Overclocking

[keesh]

Only 'performance' freaks will buy the high-end processors now anyway. Anyone with any sense will save themselves a hundred quid by losing a few MHz...

Re:Overclocking

[Musashi Miyamoto]

Not only that, but the typical increase of speed when overclocking a motherboard is around 20%. The speed difference is rarely noticable!

If you had just waited two more weeks, the cost of a faster chip and motherboard would have dropped to the same price that you paid for yours, and you wouldn't need to overclock, add any cooling units (which cost extra money anyway), and worry about the stability of your system.

Re:Overclocking

[Anonymous Coward]

I think the point was that with the high FSB speed there would be little room to increase it. With current intel processors the only way to overclock is via the FSB, but this is not true for current AMD chips that come multiplier unlocked...

Of course the highest end chips are never great overclockers, they are already quite close to the "edge".

Current AthlonXP 1700+ on the other hand... cost me $70 and runs at 2500MHz.

Re:Overclocking

[Maverick2219]

The thing with overclocking usually isn't to get the newest, and fastest CPU and then overclock the hell of it. It's to get something in the midrange and then overclock. For instance, when the Barton core was first released you didn't see curious overclockers buying 3000+ parts to overclock, no they (myself included) purchased the 2500+ part at a savings of roughly $400. I can overclock my 2500+ up to what SiSoft Sandra says is 3200+ performance specs with very little noise a-la my watercooling system. I needed a watercooler anyway due to the high temperatures in the summer in my room, and guess what? It's completely stable. People won't buy the 3200+ CPU's en mass to overclock them, but they will buy the boards updated to run the 400Mhz bus so that they can overclock their old 266 and 333Mhz bus CPU's in a more stable environment.

Re:Overclocking

[DarkBlackFox]

If I'm not mistake, they affixed a heat spreader to the Opteron. What would stop them from/why wouldn't they slap one on the higher power Athlons? Seems to me that's what keeps the P4's "cooler," along with heatsinks of twice the mass... simple thermodynamic physics.

Re:Overclocking

[mahdi13]

I don't understand why anyone would even want to overclock a CPU that is already running over 2Ghz...it made since back when you could overclock a 400Mhz to a 600Mhz, but with the new CPUs these days there is no need to overclock!

Re:Overclocking

[Anonymous Coward]

Your argument is flawed.

The reason many CPU's run at stock speeds far below their maximum is supply and demand. Intel and AMD can't sell all their chips as the fastest speed grade even if they all pass the tests! Not everyone is prepared to pay $800 for a CPU you know!

Also, buying a low-end CPU and running it at the same speed and voltage as a high-end model does not generate any more heat than the high-end model normally does and so does not require any more cooling, unless you are pushing the limits of the chip and need to keep it extra cool.

Re:Overclocking

[humming]

Agreed.

And most overclocking gains you a couple of percent faster computer while you /will/ get gcc going segmentation fault all over your ass while compiling glibc...

Overclocking was a good idea on my 486sx/25 (going to 33), on the Celeron300A and some of the earlier Durons but nowadays when common processors are breaking 2GHz? It's not the processor that's the bottleneck in todays systems, not the memory, not the bus, not the graphics card and not the harddrive, but a combination of all these parts. Sure, you can overclock all of these things, but then the failure percentage will start to add up REAL quickly and you'll end up with a system more instable than the leaning tower of Pisa.

YMMV. ;)

Re:Overclocking

[Vicegrip]

Overclocking is a decent measure of hardware stability. I know I regularly look at that data when comparing hardware in reviews.

It does say something about the potential reliability of a system if you can increase its clock by a good factor and not start to have system failures.

Re:Overclocking

[fobbman]

I'm going to assume that you are trolling, but as I'm an angler myself I can appreciate a good set of inline blades.

What I don't think you understand about the CPU business is that when Intel, AMD, whomever makes CPUs, they make them without knowing which one will end up going what speed. It's not until they test them that they find out, and then they put them into a bin based on the max speed they run.

Well, let's say that they have a good run and they get 60% of them to go at 3000+ speeds, with the rest waterfalling down from there. That's great, but the market isn't demanding a bunch of 3000+ chips. Turns out the big push is for, say, 2400+ chips. So, to fill those orders they set many of those faster chips to run at the 2400+ speeds via the cutting of bridges.

Why not just release all those 3000+ chips at 3000+ speeds? Profit, dear troll, profit. If they flooded the market with those higher-priced chips, then the price would go down. Better to make a large profit on those fewer faster chips.

At least, that's how I understand it.

Re:Overclocking

[hendridm]

Perhaps the overclocking hicks are the loud minority? I never understood it either. Reliability is very important to me, and quietness is to a lesser extent - neigher of which mix well with OC.

I'm waiting for the day I see these burned out systems on blocks in people's front yards...

Re:Overclocking

[Maverick2219]

Purchasing the right part WILL get you a massive overclock though. The prime example is the Athlon XP 2100+ Thoroughbred B part. This CPU is actually running at 1730Mhz, however many overclockers are getting these CPU's up to 2500Mhz on stock air-cooling and 100% stable. It's unknown why these particular parts are doing such a high overclock, but anyway you slice it a 44% overclock is pretty significant.

Also, my own CPU which runs at 1830Mhz and overclocks to 2200Mhz is still getting a 20% overclock.

Re:Overclocking

[svirre]

"Overclocking is a decent measure of hardware stability. I know I regularly look at that data when comparing hardware in reviews."

No overclocking is an indication of the margins in the particular device in the particular environment you are overclocking on.

This can be an indicator of either large process spread (That is the manufacturer has a number of devices performing better than specified but not good enough/many enough to mark them up)

It can also be an indication of the manufacturers politics w.r.t. introducing faster devices. This was particularly noticeable back before the athlon when intel had a lot less contested position. They then chose to introduce faster devices a lot slower then process technology allowed them to, resulting in larger design margins.

In the last 5-7 years the process spread has been reduced considerably, thus reducing the first effect (of cource CPUs has allways been binsorted which reduces spread in a single grade, but this also gives you another aspect where politics affect overclockability as CPUs from better bins are marked down to meet demand)

Overall I'd say the reduced yield in overclocking today is due to competition: Current CPU's are much closer to process limits than they were back in the 300A days.

Overclockers should also be aware that the 'stability' tests they commonly perform is woefully iadequate. There may be rarely used paths that fail to function in your overclocked devices. You do not have any overview of how good testcoverage your stabilitytests have but I'd be suprised if they are much better than 90%

Re:Overclocking

[cehf2]

Overclocking does not mean having lots of noisy fans or shortening the life of a chip. I have a Celeron 300A, which although over 3 years old has performed perfectly at 450Mhz for all of its life.

As for fans, I started out adding them, but recently I have resorted to removing them and adding some ducting to keep everything cool. The only nosiy thing in my PC is the hard disks which are 5 years old :)

The nice thing about the Celeron for o'clocking is that as you only increase the FSB from 66 to 100, you only o'clock the chip, not the m'board or RAM.

Re:Overclocking

[JuddN]

The reason many CPU's run at stock speeds far below their maximum is supply and demand. Intel and AMD can't sell all their chips as the fastest speed grade even if they all pass the tests!

I believe that your cynicism is correct in this case. I have a Pentium 4 1.8 chip and a Pentium 2.4B chip. Both of these CPUs have an 18x multiplier.

I have run both of them at a 140 MHZ FSB speed, on the same motherboard (GA-8IRXP), and they have identical performance and operate at the same average temperature. The only difference is that the 1.8 is _supposed_ to run at only 100 MHZ FSB (instead of 133 or 140), but in reality it can do everything the 2.4 can do. Oh, and it has been crunching Seti@home constantly now for almost 11 months at 40% more than its rated clock speed, with no signs of degradation.

The P4 pricing is designed so that all the stoopid rich people with more money than brains can go out and get the fastest processor in the range, not realizing that they can still get the same performance from a CPU 2 or 3 notches lower down.

Re:Overclocking

[Surak]

Damnit why, everytime a new board comes out, overclocking is brought up.

It's a good measure of performance and stability. The clock speed is a function of the crystal driving the CPU. Any CPU can (in theory) run at any clock speed. The label on the chip is a function of what the manufacturer feels is the ideal speed based on their testing.

First, overclocking works decent for a few people, but is not available to the masses for several reasons including technical difficulty and noise issues

Overclocking is fairly easy nowadays. Gigabyte, for instance, sells motherboards (the 7VAXP is one of them) that allow you to set the board to overclock right in the BIOS setup. No soldering or hacking is necessary.

Third, maybe everyone doesnt want their computer to sound like a jet is going off from the cooling needed to overclock

Can you say 'liquid cooled'? :) Quieter than 'standard' cooling even. :)

Of course the biggest reason NOT to overclock is simple: it voids your manufacturer's warranty on the chip and makes your system less stable, even if only a little bit.

Re:Overclocking

[ImaLamer]

Personally I think if you want to add more power just go with more machines.

Really, if one task can't get done, then upgrade! But if you need to - spread out, do so and donate extra clock cycles/routing time (favorite p2p?).

More computers all just combined and connected is fun.

Re:Overclocking

[jandrese]

Overclocking is another metric though. How far you can overclock a processer gives you an idea of how the processor line is going to perform in the future. Processors that are hard to overclock are tuned up closer to the limits of the design, because of that, it will probably take longer to come down in price (because the yields won't be quite as high).

Re:Overclocking

[Anonymous Coward]

I don't think a heat spreader does much for the cooling actually, not if you mount the coolers properly. The heat spreader adds an extra thermal junction...

It does keep clumsy people from crushing the processor.

Re:Overclocking

[muckdog]

The topic of overclocking came up in the gentoo forums. While an overclocked system may run perfectly stable running for playing Quake, compiling code can be a more complex task. A lot of people that claimed to have "stable" overclocked system where having a lot of trouble when they needed to compile.

Re:Overclocking

[Anonymous Coward]

You have to be kidding...

If you play games, edit video, do 3d rendering or CAD or anything more demanding than web surfing your computer is *never* fast enough.

Overclocking is a great way for people with more technical skills than money (like me) to have top-end++ performance for middle/low-end price.

Re:Overclocking

[svirre]

Wrongo.

Heat spreaders hinder thermal dissappation, they do ot help.

What they di is to privide a measure of physical protection for the fragile core.

Re:Overclocking

[DarkBlackFox]

Within the extra thermal junction though, more suface area is presented for the heat to spread out evenly to be transfered into the heatsink. Rather than all heat being pumped into the sink directly from the die, a spreader would evenly distribute the heat throughout the contact area. This is one of the bigger reasons people argue "Intel runs cooler."

Intel's die temperature is not too far off from AMD's in a chip to chip comparison, but because the die is enclosed in a speader, and a beefier heat sink and bigger fan are used, resulting in the appearance of a "cooler running system."

Re:Overclocking

[10Ghz]

First, overclocking works decent for a few people, but is not available to the masses for several reasons including technical difficulty and noise issues

Ummmm.... How is overclocking going to make the system more loud? I mean, I have 700Mhz Duron with standard fairly standard heatsink/fan. I can push the CPU to about 900Mhz without changing anything, and the system does not make one bit more noise when compared to standard 700Mhz.

Or do you think that those clock-ticks in the CPU's make a noise, and therefore more clock-ticks = more noise?

Re:Overclocking

[10Ghz]

Yes, overclocking propably shortens the life-span of the CPU's. But then again, by the time the CPU's burn out due to overclocking, they would be obsolete regardless.

My Duron has been overclocked right from the start and it runs just fine. And I bet that I will be replacing it because it's just too slow (840Mhz Duron is pretty slow these days) instead of it burning out due to overclocking.

Re:Overclocking

[svirre]

" Personally I think if you want to add more power just go with more machines."

That works until you realize your software licenses cost $20000 pr. CPU. That makes a couple of $100 extra to get the top of the line CPU seem kind of moot.

Re:Overclocking

[quakeroatz]

You obviously have little experience with overclocking. Read this while you try to pry your foot out of your mouth.

Listen and learn:
SAFETY - 90% of the new mobos have thermistors which allow you to monitor the heat of your CPU. If the processor runs reasonably cool (32c-55c) your OC system will last as long as any "retail clocked" processor, ie: indefinitely.

POTENTIAL - Processor manufacturers already know many of the chips the make can be clocked (with stability) at a higher speed than they're sold. They downclock the processors to meet the various price points which are always needed in the market. Once Intel/AMD refines a particular die, they'd be hard pressed to make a chip that can only run at their lowest speed offering. In a sense, Intel/AMD has created a product far superior than the advertised offering. This is a dated fact.

NOISE - Most overclockers love big, loud, fan/heatsink combos. It does help reach that "high end" clock, but most overclocks don't need anything other than your standard, retail AMD/Intel boxed heatsinks. Yes it will run a bit hotter than retail, no it won't be loud or fry your board. I'm talking about fairly large overclocks here, say 1.83ghz Athlon 2500 -> 2.1ghz!

If you actually had *any* experience overclocking modern processors, most of these points would be painfully apparent.

Re:Overclocking

[CTho9305]

See, what you're missing is that when Intel says a chip is only stable at 2.4GHz, they mean that in an improperly cooled case in the Sahara in summer, with a weak power supply, it is stable at 2.4GHz (and should live as long as the warranty). With a proper cooling setup and good power, it may in fact be stable at 3GHz. Overclockers take advantage of these margins to get more for their money.

Re:Overclocking

[LamerX]

Or you could run FreeBSD or Linux. Or is this not Slashdot?

Re:Overclocking

[dreamchaser]

That's why I always do a kernel compile on overclocked machines before I deem them to be stable. All aspects of the system need to be tested, and as you point out compilation is a very CPU intensive operation.

Re:Overclocking

[Rolo Tomasi]

Yeah, but what if one of the SETI packets actually contained the signal of an intelligent alien race, trying to contact us to give us the solutions to all the problems we have on earth, and your CPU glitched out and misinterpreted the data as noise. Failing to solve its problems, humanity eradicates itself in WWIII only a short time from now.

You DESTROYED the HUMAN RACE only because of OVERCLOCKING ! I HOPE it was WORTH IT !

Re:Overclocking

[yokem_55]

And this is precisely why I run a looping chrooted bootstrap with prime95 in the background as a stability test for all of my overclocked systems. If they can consistently produce fully functional code with a really, really intense integer application hammering the system in the background for over 24 hours without croaking, I consider it stable. The sad thing is is that it isn't uncommon for me to come across boards from ECS that can't even hack this test at default clock speeds.....

Re:Overclocking

[Darren Winsper]

I got an 1800+ that does 2600+. It cost me less than 80 quid. This was a couple of weeks ago. Where's my 80 quid 2600+?

Re:Overclocking

[Anonymous Coward]

Sorry dude... but YOU ARE IN WRONG !!!! I get XP1700+ Thr. core with 1466 MHz native speed.
And without any additional cooler i overclocked it to 2160 (180 * 12).
Yes.. that is 700 MORE MHz.
Mboard = ABIT NFS7 + PQI DDR 333 (up to 360 sync)
CPU Price = 80 Euros...
Now, that is speed of P4 2900-3000.
Can INTEL compare to that.
Any1 who buys Intel are ITIOTS.
They buy NAME not PERFOMANCE !!!!

Re:Overclocking

[Sj0]

heheehehehehheh

HAHAHAHAHAHAHAHAHHAHAHAHAHAH

BWAHAHAHAHAHAHAHAHAHAHAHAHAH!

too bad the SETI@home system is redundant enough to weed out a bad source like that. It would have been even funnier if it could actually happen. ^ ^

Besides -- what makes you think the hillbillies we have running the various countries of the world would *listen* to the answers to all the problems we have on earth? For the most part, I'd say they're a bunch of ignorant, closed-minded idiots who were fortunate enough to be born with a silver spoon in their mouths. They'd probably look at anyone presumptuous enough to give them advice with hostility.

Re:Overclocking

[svirre]

"Or you could run FreeBSD or Linux. Or is this not Slashdot?"

The software in question runs on linux. Windows versions do not exist, though generally solaris and HP-UX versions do.

This is EDA software, for which Red Hat Linux has become pretty much defacto standard.

Don't think that just because the OS is free that the applications are. What exist of free EDA software is pretty pitiful.

To elaborate on this: A lot of slashdotters think that free software is the Answer(tm) for everyone just because the applications they know exist in good free versions. In reality this is just true for either highly mainstream applications (which also sw developers need to use) or niche software targeted at software developers or systems management. This is because the users of this software have the ability to develop these apps. (or contain people who do)

For other niches the users generally do not have the ability (or inclination) to develop software thus no free apps appear.

Re:Overclocking

[Slack3r78]

Off with the other 333MHz FSB chips on the rich kids' playground ;) Seriously though, just because your MHz numbers match up with the 2600+ doesn't mean than it performs like one. Unless you're ALSO overclocking the poor chip's FSB as well... =)

Re:Overclocking

[kesuki]

Good points, however the end user doesn't actually know what the chip tested at, and also, the highest speed chips never have good overclocking, without over-voltaging and super cooling, which does wear down the processor a lot faster.
Also, with the way they're modifying processors nowadays it's really a pain to get a chip back up to where it tested at. Assuming you know it rated better.

Re:Overclocking

[Hoser McMoose]

Heat spreaders don't really help cool the processor, They do exactly what their name suggests, they spread out the heat so that the processor core is at a more even temperature.

The P4 and the Opteron both have rather large processor cores, so the differences in temp from one side of the core to the other can actually be quite large, which is not a good thing. Signal properties change with temperature, so large differences in temp can lead to errors, the end result being that companies like Intel and AMD can't clock their chips as high).

The Athlon, especially current Athlons, have very small processor cores, so the differences in temperature from one side of the chip to the other isn't much. As a result, a head spreader isn't nearly as important on the Athlon.

When it comes to actually getting rid of the heat, the heat spreader doesn't make much of a difference. There is a larger surface area to transfer the heat from the heat spreader to the heatsink, but you have an extra thermal junction in there to prevent heat transfer. What's more though is that the real thing holding back heat removal is getting the heat out of the heatsink itself and not so much getting the heat from the processor to the base of the heatsink.

You're definitely right about the last part though, the main reason why P4's run "cooler" than Athlons is due to the fact that they tend to use much larger heatsinks. P4's actually use more power (and therefore run hotter, all else being equal) than Athlons do.

Re:Overclocking

[Oakey]

I disagree here. I have an XP2100 Tbred b that I paid £70 for. It's currently running at 2.2GHz (13*170fsb). It's at 1.65v and Prime95 runs stable. I have an SLK800 with AS3 and a Thermaltake Smart Fan 2 (Adjust from 1300rpm -silent to 4800rpm - Jet Engine) I usually have the fan set to around 2000rpm. My temps are about 40c idle and 46c load. There are 4 other case fans but are all silent Papst fans and controlled by a fan controller. My Ti4400 has a Zalman Heatpipe cooler.

With all the fans off and the HSF at 2000rpm my machine is virtually silent with the exception of a fairly quiet fan in my Enermax PSU. Temps may reach 50c on a hot day which is where the fan controller comes into play, but even with the 4 Papst fans on, it's still a lot quiter than other coputers I come across.

If my motherboard was an nforce (instead of the crappy VIA KT333 I have) then I'd probably reach 2.4GHz.

As for safe, if the CPU temp exceeds 60c, the computer shuts down. If the fan fails, the computer shuts down.

So is overclocking dumb? Well I wouldn't say so, a 500MHz - 700MHz increase is definitely noticeable and more value for money compared to the more expensive AMD chips that barely come close to these speeds and don't overclock that well. You don't necessarily have a computer that sounds like a helicopter either, you just have to invest in some decent hardware. The Papst fans weren't too expensive, neither was the Zalman for the gfx, the SLK800 is around £40 but it's a great heatsink and I'm sure it'll last fine for a while yet.

Re:Overclocking

[Darren Winsper]

Eh? It's an 1800+ that does 166*12.5, which is what a 2600+ is. Ergo, it performs just like a 2600+.

Overclocking

[mrtroy]

Damnit why, everytime a new board comes out, overclocking is brought up.

First, overclocking works decent for a few people, but is not available to the masses for several reasons including technical difficulty and noise issues

Second, overclocking is kind of dumb (expecting 10000 evil replies for that, but listen first) because if the board really could safely go faster, the manufacturer would produce it that way, and sell it for more!

Third, maybe everyone doesnt want their computer to sound like a jet is going off from the cooling needed to overclock, especially since as computers are getting faster, and more "stuff" is being put in smaller and smaller spaces, heat is increasing as well. Thats why mobos are coming with bigger fans, graphics cards are coming with giant fans that take a whole slot, etc.

Now personally, I considered overclocking, fiddled with it, decided it wasnt for me, but I realize a small amount of people will do it. Cheers to them, but why can we not critically analyze a mobo without considering overclocking, which will benefit less than 1% of users! Lets look at the raw performance, and it should be sweet with this fat bus!

Benefit?

[NSParadox]

The Athlon chips have not been super-sensitive to changes in FSB. The performance impact of the Athlon XP moving from 2x133MHz to 2x166MHz was significantly less than the P4's gains going from 4x100MHz to 4x133MHz. The P4 gains have been incredible with the jump to 4x200MHz.

It seems that AMD is trying to squeeze every bit of performance out of an architecture that would be better squeezed by being optimized, i.e. Opteron. It's a shame that AMD's yields of Opteron have proven to be dismal, but if I was a motherboard manufacturer I'd be pretty mad at AMD right now. More motherboard manufacturers are going to have to qualify their boards and more chipset manufacturers will have to qualify their products as well, even if they can already meet 400MHz operation. Will the performance impact really justify the costs that all parties incur by moving to yet a new FSB in less than, what, 6 months?

Re:Benefit?

[Zathrus]

It's not a huge issue on the MB side - most boards have allowed for a 200 MHz FSB for awhile now, although only usable if you wanted to overclock.

Of course, now they'll need to make sure their boards can exceed 200 MHz, otherwise the overclockers will start whining about not getting an additional 0.5% performance.

Make up your own shit, you wanker

[Doctor Sbaitso]

From HardOCP [H]ardNews 6th Edition posted on Wednesday April 30th, 2003:

Athlon 3200+ Coming:
The Athlon 3200+ with a 400MHz FSB is on the way in the next two weeks, according to C|Net. It'll be interesting to see how well the processor overclocks at that high of a bus speed...it didn't seem to hamper the new 800MHz FSB Pentium 4, which many people were worried about too.

Re:Make up your own shit, you wanker

[mrtroy]

Wasnt that *gasp* YESTERDAY!

That my friend, is plagerism!

No wonder he failed out of school and works @ slashdot...he cuts and pastes like its going out of style!

Isnt plagerism wrong, immoral, illegal?!!!

Re:Make up your own shit, you wanker

[pacc]

Soo, like if Slashdot lives on original content...

Re:Make up your own shit, you wanker

[amembrane]

Yeah I noticed that immediately too. Way to give credit where credit is due.

Re:Make up your own shit, you wanker

[Nate Fox]

dont worry, he'll get what he deserves - his email was just posted on /. front page for every spam engine to harvest and rape :)

Re:Make up your own shit, you wanker

[Anonymous Coward]

(Sodalas) FYI not EVERYBODY reads the [H] and apparently alot of people did not know about the 3200+, 260+ replies. Not only that but Kyle didn't mind, I got an email.

Re:Weird characters

[Anonymous Coward]

Maybe because the poster made a straight copy/paste from HardOCP?

Damn plagiarists. They should all be sent to prison and molested up the arse with electric cattle prods. That'll teach them.

This thing is gonna be HOT

[checkyoulater]

I already have trouble cooling my XP1900 without having it sound like a jet engine. With a slow fan and decent heatsink, my CPU still sits around 48 degrees C. I'm afraid to think how hot this thing would be. How can anybody productively use a computer with a fan that is as loud as an engine idling?

Re:This thing is gonna be HOT

[chefbimbo]

The athlon is rated to be safely running at 85 C. Mine always runs at 60 because of low noise components (GOOD heatsink but still) and I couldn't care less for it.

Re:This thing is gonna be HOT

[CokoBWare]

I have the exact same processor, and I use a Volcano 9... variable rate fan that will speed up when the thing gets too hot... nice and quiet except when I play a high-poly game. I didn't know that an Athlon XP was rated for higher temps... I guess I shouldn't be so worried when it hits 49C... most of the time it runs at 41C idle!

Re:This thing is gonna be HOT

[Anonymous Coward]

Hell...I have a XP2200+, and it sits at about 60C idle. I wouldn't dream of overclocking this thing because I'm worried about heat damage, and you're worried about 48C? Man, I can't wait until my Vantec Aeroflow comes in...

Re:This thing is gonna be HOT

[mahdi13]

48C is pretty cool for an Athlon. My 1600+ runs around 53C, AMD specs them upto and past 65C...I wouldn't worry about it, just make sure the HS/Fan doesn't fall off =)

Re:This thing is gonna be HOT

[Anonymous Coward]

Don't worry about 48 degrees:

% sensors
...
CPU Temp: +57.3C
...
% uptime
...up 237 days...

Re:This thing is gonna be HOT

[drinkypoo]

With a normal-speed (~5500rpm) CPU fan my athlon XP 1700+ overclocked to 1.66GHz (Athlon XP 2000+ speed) my CPU temp is at 40 deg C even (104 deg F.) Perhaps you improperly applied your thermal compound, you need a better heat sink for that fan, or you have an older core which is less efficient? Or your room is just too hot :)

Re:This thing is gonna be HOT

[checkyoulater]

I believe the 85C rating is the die temperature. My crappy motherboard measures the temperature underneath the CPU. I believe you are supposed to add around 15-20 degrees to that temp to get your real temperature. That makes my 48 more like 68. Still well within acceptable temperatures, however.

I think I'm just paranoid, for some reason. I guess I spend too much time reading the arstechnica cooling forums.

Re:This thing is gonna be HOT

[checkyoulater]

With a normal-speed (~5500rpm) CPU fan my athlon XP 1700+ overclocked to 1.66GHz (Athlon XP 2000+ speed) my CPU temp is at 40 deg C even (104 deg F.)

Ah, yes but there lies my problem. I have Vantec stealth fan, undervolted to decrease noise. When I crank up the fans, I can get my temp down to the low 40's, at the expense of noise. That is something I will not accept. I'll deal with heat for a quiet room. Now if I could just do something about that damn GeForce fan...

Re:This thing is gonna be HOT

[Anonymous Coward]

Interesting.... I have a 2400+ and depending on the weather outside, which dictates the temperature in my house, and what time of year it is.. I have hit 60c with no problem.

I live in Phoenix, AZ. It DOES get awful hot here. I keep my air conditioner at 82 degrees in the summer.

I can get to 41c idleing only in the winter months : )

Re:This thing is gonna be HOT

[dboyles]

With a slow fan and decent heatsink, my CPU still sits around 48 degrees C.

I just installed a Thermalright SK-7 and Vantec Stealth fan with Arctic Silver 3 on a couple of machines running an XP 2000+. Idle temperature is in the low forties (Celsius) and the things are very quiet. Actually, the noise improvement over the retail heatsink/fan combo wasn't really noticeable, although temperatures did drop significantly.

We've also got a couple of machines running dual MP 1800+ CPUs, and they were *loud* with a cheapo heatsink and Global Win 60mm fan. Installing the Thermalright and Stealth lowered the temperature a few degrees, but cut noise by what seems like a factor of 3.

We bought all of this cooling gear from Best Byte and were pleased.

Re:This thing is gonna be HOT

[Guilly]

Temperature also depends a lot on case temperature... If you can cool your case correctly (without jet engines) then you can cool your cpu without jet engines. I have a 2000+ (1.67GHz) running at 1.80GHz, but with 5 silent thermaltake fan cases and a globalwin cpu fan I keep it around 35 C. My mobo is around 25 C.

To give you an idea of how much the fan cases matter, if I turn them all off, it takes 5 minutes for my system to crash under heavy load, as cpu temp. approaches 70C and mobo climbs to ~45. (I guess the overclocking is why it crashes at lower temperatures than it should)

It is very stable with the fans on.

Re:This thing is gonna be HOT

[Some Dumbass...]

I believe the 85C rating is the die temperature. My crappy motherboard measures the temperature underneath the CPU. I believe you are supposed to add around 15-20 degrees to that temp to get your real temperature.

You're supposed to add 15-20 degrees Fahrenheight! That's 6-8 degrees Centigrade.

Re:This thing is gonna be HOT

[Some Dumbass...]

Ah, yes but there lies my problem. I have Vantec stealth fan, undervolted to decrease noise.

You think a Vantec stealth fan sounds "like a jet engine"? It's not whisper quiet, but seriously, it's not very loud either.

Re:This thing is gonna be HOT

[checkyoulater]

You think a Vantec stealth fan sounds "like a jet engine"?

No, that is why I purchased it in the first place. Start with a quiet fan, and then undervolt it to make it that much quieter. You lose airflow, but you more than make up for it with lower decibels. The jet engine I was referring to is the stock fan that AMD puts on their cpu's, or those 5000+ rpm Delta screamers. Those are the fans you need to use to get the >2000 XP's into the low 40's.

Re:This thing is gonna be HOT

[Anonymous Coward]

ive got a kr7a-raid and palomino 2100+@2200, with a delta fan on it. its in a fat chenbro server so i dont care for the noise. it sits idle at 26C, 42C under load.

my new kd7-raid, barton 2500+(1.83g) and coolermaster hhc-001 (7300rpm) is noisy too, but the barton at the moment is running at 166x13(2.13g) with 40C under load, 27C idle. going to push the fsb up to 200 when i get some time.

all my systems are in my basement, so the noise factor i dont mind. dont have to hear them when i goto bed. and they warm up the basement on cold winter nights.

if i didnt want noise, id get an epia via fanless system as a headend upstairs and run all the big shit remotely.

Re:This thing is gonna be HOT

[Anonymous Coward]

WHAT??

Re:This thing is gonna be HOT

[jonbrewer]

I already have trouble cooling my XP1900 without having it sound like a jet engine. With a slow fan and decent heatsink, my CPU still sits around 48 degrees C.

I use this Zalman, and my Athlon XP1700 is at 49 C. The only thing I can hear from my case is an old 60GB Maxtor drive.

Anyway, the Zalman seems to be a perfect match for you, at least if you believe what you read on the Internet:

"it's designed to provide efficient cooling without sounding like a turbine engine..." (from the ThinkGeek page)

Re:This thing is gonna be HOT

[steveha]

As others have noted, if you currently have your CPU at 48 C, the new CPU probably will stay within spec using your existing cooling setup.

But depending on which CPU core your XP 1900+ has, the new CPU may not even be much hotter.

An XP 1900+ with a Palomino core dissipates about 60.7 Watts typical, 68 Watts max.

An XP 3000+ with a Barton core dissipates about 58.4 Watts typical, 74.3 Watts max. I should expect the new Athlon will be only slightly hotter than the Barton 3000+.

Oh by the way, the Thermal Design Power for a Pentium 4 at 3000 is 81.8 Watts. If we figure typical will be under 75% of that number, the typical heat dissipation of a Pentium 4 at 3000 will be similar to the AMD at 3000, and the worst-case slightly hotter.

These numbers from:
http://users.erols.com/chare/elec.htm

steveha

This is bad news.

[Krapangor]

Computer processors get faster and faster. You might think that this is a good thing, but I'll explain why it has only negative effects.

With faster processors we get less efficient programs. 10 Years ago you could do internet/email/word processing/spread sheets with just a 33MHz Intel 386 with 16 MB RAM. Today you need for the very same things a Pentium IV with 2 GHZ and 128 MB RAM. There are some niece applications which need a lots of CPU Power like Quake or Nurmerical Simulations, but must Joe Adverage apps don't really need it. The programs need it due to sloppy coding. And the faster CPUs gave rise to the OOP paradigm. While it primarily is a nice theoretical concept for safer and more secure program, it's used these days just for code-bloat and GUI overload. Inpedendent studies show that in fact 73 percent of all "OOP" code is just imperative with C++ class bloat added.
Further the higher compiler and debugger speeds introduced much more sloppy coding styles. In the 60/70ies the computers of the Apollo program hadn't a single computer crash, which is completely unthinkable these days. The reason why the NASA is keeping old 8080 Intels in their shuttles is that they won't get decent code quality form modern processors these days.

Personally I think that the whole CS community must rethink their position towards computers speeds. Instead of the todays faster-is-better point we need a paradigm change towards just-as-fast-as-necessary.

Re:This is bad news.

[Paddyish]

This is why open-source is so nice. You don't need much to run 'nix OS and apps. Folks aren't as fast to scrap useful hardware when effective software is written and widely available.

Re:This is bad news.

[dew-genen-ny]

I thought that NASA stays with the old 8080s (or whatever they have) because they have like a 8 micron spacing by comparison to new PIVs at .18 or .13 or whatever they've got to.

The amount of stray radiation plays havocs when the spacing is too close - electrons can get bumped up and jump.....

Or am I talking out of my arse.....

Re:This is bad news.

[pilybaby]

The reason why the NASA is keeping old 8080 Intels in their shuttles is that they won't get decent code quality form modern processors these days.

Employ decent programmers and use decent compliers is what that comes down to, nothing to do with the processor. NASA employ extreamly strict code review processes for their software although I tend to agree that for 'consumer' code sloppy coding has been the result, but better interfaces too.

I really think AMD has to get it's head on straight soon. They are being trounced by Intel and have gone back a few years to being the processor for budget buyers only. Hopefully the Athlon 64 will rock and hit intel where it hurts althoughw ith memory prices as they stand I can't see many people taking full use of it any time soon.

Re:This is bad news.

[Anonymous Coward]

10 Years ago you could do internet/email/word processing/spread sheets with just a 33MHz Intel 386 with 16 MB RAM. Today you need for the very same things a Pentium IV with 2 GHZ and 128 MB RAM.

why? I can still do everything you say today with that 33mhz intel 386.

I wasn't stupid and bought software just because it was there. when microsoft sells Office XP special edition 5 with beta-carotene does NOT mean you have to buy it.

Hell, I can do everything you mention on a 286 with 4 users using it at the same time PLUS serve as a firewall and HTTPD server.

just because someone is selling something does not mean you NEED it.

Re:This is bad news.

[turgid]

There's nothing from stopping you today from running the programs of 10 years ago on the hardware of 10 years ago.

Mensa member - beware the high IQ

Beware the pretentious idiot.

Re:This is bad news.

[tomcio.s]

"just-as-fast-as-necessary"
That almost begs to introduce nice asych arch into the mix. This way you can stay current for so much longer.
This and we would get rid of the whole 'penis envy' speed thing. Let's leave that to cars.

Re:This is bad news.

[MtViewGuy]

I have news for you.

Back in 1993, Internet access was mostly in text mode, so you really didn't need that much computer processing power to access it.

Today, Internet access is through web browsers reading highly-formatted graphical pages, Macromedia Flash/Shockwave graphics, and streaming audio/video, much of it through 384 Kbps and faster download speed broadband connections. This requires a LOT more computing power, to say the least.

Also, the needs for high-end games and multimedia processing has really bumped up the need for faster machines. Would you want to process your photos downloaded from a modern two-megapixel or higher-resolution digital still image camera on a 1993 vintage machine, let alone video movies downloaded via the IEEE-1394 connection from MiniDV and MicroDV camcorders? I didn't think so.

Take for example Adobe's new Photoshop Album program for digital still camera users. You really don't want to run this on a Celeron A 400 MHz CPU, since Photoshop Album--derived from the current professional Photoshop program--requires a good amount of CPU processing power to process images. The program would be more appropriate for CPU's that sport MMX, SSE, SSE2, 3DNow! and/or 3DNow! Professional CPU instruction extensions.

Re:This is bad news.

[sh4de]

There are some niece applications which need a lots of CPU Power...

And Bob's your uncle? Maybe you meant niche? Also, "a lots" is new to me.

Mensa member, beware of the high IQ

Anybody else being cut by the razor sharp irony in this? Or maybe I'm just bitter I didn't get accepted to Mensa with my puny IQ of 138.

Re:This is bad news.

[Anonymous Coward]

thats why X and the main windowmangers are bloated ram hogging crapware.

Re:This is bad news.

[athlon02]

You're absolutely right. NASA doesn't use Athlons or P4's because older processors stand up better against cosmic radiation and because older technology is proven to work after so many years of use, whereas Athlons and P4's have not proven to last that long (yet). The guy who posted the original article hasn't a clue on this issue it seems.

Re:This is bad news.

[Hagakure]

No, he meant niece. The rest of us won't fully understand that statement for another 10 years.

Re:This is bad news.

[The Lynxpro]

Amen... I can remember the days in Atari ST land when some German hackers got mad at the performance of the TOS operating system and they decided to rewrite it line by line into 68000 machine language... I can't remember if TOS (C/PM 68K) was written in assembly or C (probably C)... Can you imagine rewriting Windows in x86 machine language versus C++?

Re:This is bad news.

[javahacker]

NASA doesn't use Intel 8080 processors in the shuttles. The computers they use were developed for the Apollo program, way before the Intel 8080 existed. They use them because they are simple enough to have provably correct operation, something not true for most processors. This is a quality that must be designed for in the processor, and is more difficult to achieve as the processor becomes more complex. Code quality has nothing to do with the decision. Their code is all assembler by the way, so your code quality is very high, and very expensive.

Inpedendent studies show that in fact 73 percent of all "OOP" code is just imperative with C++ class bloat added.
You mean it was crappy, non object oriented code, written by bad programmers! What a shocking notion! Anyone can write bad code in any language, it hardly takes any skill at all, which is the problem, lack of skill.

And the faster CPUs gave rise to the OOP paradigm.
OOP is simply a codification of what programmers were already doing, it is neither a magic bullet, or a terrible evil.

Re:This is bad news.

[Anonymous Coward]

The reason why the NASA is keeping old 8080 Intels in their shuttles
Yes, Columbia had one too. But they overclocked it little too much.

PS: Moderators, please moderate me down. This is very bad one, not funy at all...

Re:This is bad news.

[Malc]

Give it a rest. Who cares if programmes require more horsepower? Only elitist old timers who are trying to show up the historical ignorance of newbies.

I welcome more powerful CPUs. This allows me to spend my time writing code faster and more bug free. In most situations I encounter, this is more important than wasting time needless optimising for speed and space. Optimise when necessary, otherwise concentrate on robustness and reliability, and delivering in a reasonable length of time. Personally I'm glad I don't have to spend my time optimising and debugging assembly or badly-written cryptic C code that uses the most obscure methods allowed by the language to obtain questionable performance benefits.

As for the CS community, most of the software you're complaining about have little to do with computer science values. They're catering more to business values, and making money. Idealistic computer scientists are a pain in the arse until they learn that their employer's goals are more important their own pet projects - if they don't, they end up being those older pain-in-the-arse prima donna coders who have to make things difficult just for their own job security.

If don't agree with what I'm saying... well there's always open source, but a lot of the stuff that isn't hobbyist quality probably won't fit your needs anyway. Look at the size ("bloat") of KDE... but at least it works, and works very well.

Re:This is bad news.

[ipjohnson]

I think his basic point was that rather than design something correctly alot of people have the notion of "just throw more horsepower at it" I know many many programmers that the solution to poor design is add more memory and a second processor (especially java). I think that just seems silly. I'd rather see a nice tight design that takes a little more time than a quick sloppy one that can work if you have a monster of a machine (the reason being its easier to scale up when your not bloated).

Now I'm not saying it has to be this way I just I have seen more abuse of this idea from OO programmers than the hardcore C fans (I'm biased being a C nut :).

Just my 2 cents.

Re:This is bad news.

[SpamJunkie]

An IQ of 138 is easily high enough to get into Mensa, but that's not very interesting. My limitted experience with Mensa suggests that it is not full of intelligent people but stupid people that do well on IQ tests, which is really unfortunate because the next best alternative for finding other smart people might be slashdot which is very, very sad. Either that or I'm just generalizing and being melodramatic. Hard to tell, I'm pretty hung over.

So ya, Mensa members tend to be more navel gazing people of average intelligence than smart people with lofty goals. But hey, at least they aren't quite as evil as the scientologists.

Re:This is bad news.

[jsse]

I wish my argument wouldn't hurt your Mensa membership status and high IQ. Just for the sake of discussion. :)

A 33Mhz Intel 386 16MB RAM is still usable to lot of us. I've helped a couple of companies making use of these as gateways and firewall with Linux 2.0.x kernel. This might be a striking fact for you that these functions don't require Checkpoint for Windows on Pentium IV.

GUI is not bloated, it improves productivities by straightening the learning curve. Neither OOP paradigm is bloated, if you've compared no. of compare computer projects done today and 10 years ago you'll probably see how change of paradigm actually helps in development cycles.

It's too risky to jump into conclusion with the slight fact displayed on the tip of an iceberg. The reason why NASA keeping 8080 for few of the systems in space shuttles is the issues with heat dissipating in space. It's not like 8080 is too good to replace. Check the fact and avoid making too much assumptions, before jumping into your own conclusion.

And it's not true that old coding style makes computer more reliable. I'm not going to elaborate on this, as it's too damn obvious to the rest of us. Hint: think in the line of complexity of computer systems in the past and present.

Actually I really understand what you rant about. *Some* companies push their customers to upgrade their products by making things unnessarily complex. It's not like the fault of computer industry in general. Some people are still doing things as simple and useful as possible.

Re:This is bad news.

[Malc]

I don't think OO is to blame. I think OO was the predominant technology in use and one of the main buzzwords when the big IT boom of the 90's attracted a lot of people who were after money and not satisfaction. Those people tend to be of a lower calibre anyway. If the boom had occurred earlier, I think we would have a seen a lot more shocking C code and there would be more complaints about the structured programming paradigm ;)

But then I'm a biased former C++ nut (it's still my primary language) who wants to get away from the problems of that language and its predecessor.

Re:This is bad news.

[theanorak]

I can't decide whether this is trolling or not - if it is, its pretty clever - appeal to the functional luddites.

Faster processors = less efficient programs - as someone already said, so? Granted there's some really dog rough code lurking about, but...? For the most part, its not relevant.

10 years ago I was doing wordprocessing and spreadsheets on an 8086 - an Amstrad PC1512 (probably a UK-only thing). It worked, but I would't dream of going back to it. No way. I can say with utter certainty that the wordprocessors and spreadsheets I use today are VASTLY superior. Vastly.

You don't need a P4 with 128Mb?? RAM. Actually, if you were using XP, it would run like crap anyway. Most of the researchers in our office are using something between PIII-600 and PIII-1Ghz, with multiple browser windows, multiple documents and proprietary databases open. It works for them.

"Niche" applications are the only ones that need power? Er, that's crap. Talk to some "average" computer users - I think you'll find that the marketing campaigns have worked. They don't just want to send email and write letters. Why?
- Digital cameras
- Digital video
- Gaming
- Music
- Home entertainment

All of these use serious computing power. You don't have to be a Photoshop pro to need power for processing the 100 3megapixel images from your consumer-level digicam.

You don't have to work for Dreamworks to need computing power to edit together your cam footage from the summer barbecue, or your 5 year old's birthday, before burning a copy for the in-laws.

And I haven't even touched on ease of use. The fact that my I've been able to persuade my mum to use email, and show her its not that complicated - that's worth the admission price alone. It would never have happened with a CLI email client. I'm not saying there's no use for them, but my mum would never have bothered. Would your gran? Your parent? So if a few processor cycles are used making it look pretty, so someone who's not comfortable with technology is happy to give it a try, I say use 'em.

Re:This is bad news.

[frooyo]

OOP is simply a codification of what programmers were already doing, it is neither a magic bullet, or a terrible evil.

I think you meant to say 'Silver' bullet, not magic bullet. And yes, nothing to this day has been stamped as a Silver bullet to programming.

Re:This is bad news.

[Sj0]

+1 evil

+1 subversive

Finally,

+1 not every tragedy is so terrible we can't joke about it.

All in all, good post. :P

Re:This is bad news.

[Anonymous Coward]

There are some niece applications which need a lots of CPU Power like Quake or Nurmerical Simulations, but must Joe Adverage apps don't really need it.

....

Mensa member, beware of the high IQ

And who's this Joe Adverage he's talking about?

Re:This is bad news.

[Slack3r78]

althoughw ith memory prices as they stand I can't see many people taking full use of it any time soon.

Am I missing something? DDR memory is the cheapest I've seen it yet right now. If you mean most people aren't going to have >4GB of RAM, you're right - they don't need it. At the moment, 512 is as much as 90% of the people out there are gonna need. >4 gig memory are for server and science/engineering people, for the average consumer, the improved x86 performance is going to be the selling point.

Re:This is bad news.

[Beliskner]

And the faster CPUs gave rise to the OOP paradigm. While it primarily is a nice theoretical concept for safer and more secure program, it's used these days just for code-bloat and GUI overload. Inpedendent studies show that in fact 73 percent of all "OOP" code is just imperative with C++ class bloat added

GUI bloat can be eradicated by removing the User Interface. If your spreadsheet runs as a service with no User Interface then naturally 99% of the code becomes redundant and can be deleted.

I never thought the Osprey (latest military fly-by-wire plane) would need a reset button, and that this reset button would kill people, but it does!

Press Releases

[rwiedower]

Analysts say it's too early to know how the new chips will rate against each other, with testing not yet complete.

Yes, I know it's too early to know how the new chips will rate. Everyone should know this. It used to be that a PR blitz was timed for the launch of a product. Now it comes out well in advance. This, in turn, means that delays that could affect the delivery date have to be factored in. Next thing you know, we'll have helpful stories over a month in advance of launch with more helpful statements about how the chips haven't been tested yet.

Yes, if the chips have already been produced and are filtering into distributors, this point is moot. I just wish more was made when the products emerged and less when it was all pie-in-the-sky hyperbole.

Re:Press Releases

[sigep_ohio]

Me too. I get so annoyed when I read stories about the "launch" of a new processor or graphics card, only to find myself holding mr happy for 2 months while I wait for the damned thing to actually show up in stores. Companies shouldn't be allowed to say something is "launched" or for sale unless it is actually in the stores. Not while some OEM is getting test samples.

Judging by the response...

[Paddyish]

I'm beginning to think CPUs might be effective kitchen appliances.

Re:Judging by the response...

[Anonymous Coward]

I'm beginning to think CPUs might be effective kitchen appliances.

Yes! Who needs a stove when you got an Athlon? Somebody has already fried an egg on a XP 1700+. Imagine what you can do on a 3200+ !

Hmmmmmmmm, hmmmmm, hmmmmmmm.

Re:Judging by the response...

[Sj0]

I've said it before, and I'll say it again -- You're talking about a processor with a thousand times the processing power of the entire world 50 years ago -- That's why it needs a heatsink and fan. Why anyone would expect such a powerful processor to automatically keep the same thermal specs as an old 386 is beyond me!

so what?

[gregmac]

Unless they shrunk the Athlon core, I don't see a lot fo room for overclocking. The 3000+ isn't an overclocking dream, so simply moving to a faster bus ain't gonna make the 3200+ any better.

Ok. So who cares? Why do you need to overclock it? We've already gone from 1ghz to 3ghz in the last year, is this not fast enough?

I suppose if theres some urge that makes people feel they absolutely have to overclock, then they can just go buy one of the older models, for a bit less money. Then spend probably as much money beefing up their cooling. Not to mention the time in getting it to work. Now they have a processor that runs as fast as this one, that cost just as much money, but is a whole lot noisier and has no warrenty.

Re:so what?

[sigep_ohio]

well, if I was even half as smart as most /.'ers I would have quoted the original artical in my post. I was merely responding to the statement about overclocking, which by the way I have no interest in doing. My view on overclocking is similar to yours: whats the point?

Re:so what?

[shamilton]

Hm... so really, why do anything at all?

I don't think many overclockers do it because they feel they will gain much from it. They do it because it can be fun, can be stimulating, can be challenging. Like many things, it is something to do between sleeping, something to do to pass the time before dying.

Re:so what?

[Anonymous Coward]

If that's the case, you have your pick of hobbies. Why not try one that's actually interesting? How much do you really learn from overclocking a prepackaged processor, anyway?

Re:so what?

[Sj0]

That's wierd. The last two chips I overclocked (A p166 clocked to 200Mhz and my Geforce 4 to some frequency I can't even remember, but 600Mhz sounds familiar...), it's because it meant increasing the power of my machine without dropping anything on a new component.

I guess people who overclock "...because it can be fun, can be stimulating, can be challenging" are the same people who enjoy russian roulette.

Whatever your tastes, I guess.

AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800 MHz

[Forge]

Hey. Marketing people love to trumpet all kinds of fantasy based figures when they talk about CPU spead.

The troth is that the only CPU mesure that matters is how long dose it take to rip and encode a DVD to DivX (One of the few tasks that still taks hours.) or whatever application YOU run which YOU feal is too slow on whatever system you have now.

And for comparison, Athlon 3200+ vs iNTEL 3.2 GHz is not what matters. What matters is iNTEL's $500 CPU vs AMD's $500 (or $100 CPU).

Enough of 32 bits! Give me 64!

[turgid]

Please, AMD, just get on with it and give us the Athlon 64. Consign the 32-bit chips to the bargain basement. The workstation and server market has been 64-bit for nearly a decade. It's time we caught up. I'm off to ebay to buy a second-hand alpha workstation...

Re:Enough of 32 bits! Give me 64!

[Jungle guy]

It will probably be the last Athlon XP to be released. After that, ADM will release Atlhon 64. You won't have to wait too long.

Re:Enough of 32 bits! Give me 64!

[turgid]

I've been waiting since 2002 already... and I've been putting off upgrading my old K6-2/500.

Re:Enough of 32 bits! Give me 64!

[Anita Coney]

Why do you NEED clothes, a house/apt, a car, a computer? Why do we NEED anything but food and air?!

Re:Enough of 32 bits! Give me 64!

[Anonymous Coward]

and water.....

Re:Enough of 32 bits! Give me 64!

[jandrese]

Wow, when did Archer Daniels Midland get into the Processer biz? Pretty big jump for an agricultural company.

Re:Enough of 32 bits! Give me 64!

[svirre]

Because it would get awfully crowded in the tropics if all us hairless monkeys were to move south for the winter.

Re:Enough of 32 bits! Give me 64!

[ProtonMotiveForce]

Pfft. Right.

Deranged idiot: "64 > 32. Must get 64 bit processor".

What the hell do you think you need 64 bits for in a desktop? I assure you it won't help your porn, your games (regardless of what that AMD schill Sweeney claims), or your spreadsheets.

Re:Enough of 32 bits! Give me 64!

[turgid]

64 bits is the future. It would help me to develop future-proof software if I had a 64-bit machine at home, since I can't really use the machines at work for my own personal frobbings. Being a deranged idiot is fun. In fact I'm so deranged I refuse to spend $20k+ on a workstation for home.

is it just me?

[wirefarm]

or did others stop caring a lot about speed somewhere around 1Ghz?

Re:is it just me?

[Anita Coney]

I agree. My first computer had a 133MHz Pentium. I then upgraded to a 233PII. Then to a 550III. Even though my current system has a 1700+ XP, the 550 isn't exactly slow. In fact I doubt if my wife could tell he difference.

I had planned on upgrading to a 2600+ XP when the prices go down, but the only reason I see to do it is for Doom III. I certainly won't be doing it to speed up day-to-day tasks as my 1700+ XP is fast enough for me.

Re:is it just me?

[pezpunk]

it is totally and completely just you.

things my AthlonXP 2000+ can do significantly better / faster than your 1Ghz processor:

- encoding mp3 / divx / etc
- high quality divx playback
- games games games. War3 at 1280x1024 is beautiful =)
- Windows XP =/

Re:is it just me?

[Anonymous Coward]

I stopped caring so much after 1GHz because the performance bottleneck is no longer the CPU, it is now the FSB.

Announcements about faster busses and putting more devices (e.g. gigabit NICs) on the FSB get me excited now.

Of course, once bus speed catches up, then the CPU clock speed wars will get more interesting again.

Re:is it just me?

[shamilton]

Indeed. I had the grandparent's mentality for three years, then recently used some extra cash to upgrade my 500 to an 1800 to play UT2003.

The improvement in general system responsiveness was far greater than I had expected. Windows open faster, it's less evident that things are "drawn" instead of just appearing, much faster boot, etc.

Also pleasant was reduced time waiting for compiles when making small code tweaks, waiting for photoshop filters, etc.

You will never hit any sort of upper limit for CPU usage and just stop benefiting.

Re:is it just me?

[utahjazz]

or did others stop caring a lot about speed somewhere around 1Ghz?

Try running Cinelerra on a 1Ghz. Unusable. Your 3D graphics card will not help, you need raw CPU, and I'll bet with an AMD 3200+ you'd still crave more.

Re:is it just me?

[Some Dumbass...]

Try running Cinelerra [heroinewarrior.com] on a 1Ghz.

Anyone ever heard of this "Cinelerra" thing? No? Not even hardcore geeks have heard of it, much less the average joe? That's not much of an argument for faster CPUs.

Re:is it just me?

[mcgroarty]

I pretty much agree with you. I stopped caring somewhere around 1MHz as well.

Running my 128 in double clock mode broke virtually all my C64 titles. Even GEOS didn't benefit properly without buying all new software. :(

Re:is it just me?

[Anonymous Coward]

Windows is so slow that you can see windows being drawn at 500MHz? And people complain that MacOS X is slow?

Re:is it just me?

[Linknoid]

Curious, did you also upgrade your video card at the same time? I had a Celeron 900 with onboard video, and I upgraded to an Athlon 1100 underclocked to 900 MHz with a Radeon 8500LE, and using the same PC133 DIMMs, and it felt a whole lot faster than the difference between an Athlon and a Celeron should make. I really think the video card has more to do with how fast screen draws feel than the CPU.

Re:is it just me?

[Klaruz]

Anyone ever heard of this "iMovie" thing? No? Not even hardcore geeks have heard of it, much less the average joe?

Jeeze man, video editing exists on all platforms, just because you haven't heard of a certain linux editor, doesn't mean nobody edits movies.

Re:is it just me?

[falser]

The main improvement is programs crash much faster now. I never have to wait for XP to hang for 20 seconds before killing an application anymore.

Re:is it just me?

[shamilton]

When I indicate screen draws, I mean when you open "My Computer" it chugs away doing whatever and tries to draw the window at the same time, so the bottleneck isn't the video card, but the system being busy (in the case I describe.)

Re:is it just me?

[addaon]

Haven't decided yet, haven't broken 600MHz yet. (And I've bought three computers in the last year.)

Re:is it just me?

[Some Dumbass...]

Jeeze man, video editing exists on all platforms, just because you haven't heard of a certain linux editor, doesn't mean nobody edits movies.

Okay then, how many computer users edit movies?

My point is still valid. If you're going to argue that people should care about 1Ghz+ CPUs, then you really need to come up with a task which more than about 1% of the population does. Big CPUs are great for MATLAB, SPSS and other scientific software as well, but that doesn't mean that 99% of the populace cares.

The fact that you came up with an obscure movie editor as an example is just icing on the cake. Perhaps iMovie doesn't need a 1Ghz CPU?

Re:is it just me?

[the_real_tigga]

oh yes indeed.

now if they only would build a decent laptop with around 700-1000MHz, along with the newer technology added, especially battery-wise, and sell it real cheap.

Re:is it just me?

[Hoser McMoose]

No, everyone stopped caring after the 386 came out... at least for a few years, until they realized how slow everything was on an old 386. Then they stopped caring about anything beyond a 486, for a few years until that became slow. Than the Pentium was all that anyone would ever need, until that was too slow. But once we got to 200MHz, well really that was fast enough to run all the programs out there.. for a few years until that started to seem too slow. Now that we've got 1GHz chips though! That's for sure all we'll ever need! At least for another year or two until 1GHz chips seem really slow.

I can absolutely guarantee you that in a few years time, people will be saying that their old 1GHz chips are way too slow for their regular every-day software, and other people will be talking about how clock speed doesn't matter beyond 5GHz. This cycle has been repeating itself for the past 15 years, and doesn't show any signs of letting up.

Re:is it just me?

[intermodal]

thats windows. The big kicker there is probably ram speed and ATA speed. I think it's your motherboard that made the difference. I've got a 300mhz that significantly outperforms another box someone gave me that runs at 400mhz. Why? the 400 has a crappy motherboard.

What price power?

[AccUser]

I work from home, and have a network of 5 PCs of various specifications. You can tell when they are on because the floorboards in the hall vibrate, never mind the noise they make.

I have recently invested in a VIA EPIA-M10000 motherboard. It is very, very quiet.

Sure, it isn't as fast as the latest P4 or Athlon, but it plays DVD (with hardware support), DivX, and MP3 media without any problems. Quake 3 runs well.

More importantly, I can run all my business applications without any noticable loss in speed.

I'm going to ditch my other boxes and buy some more of these EPIA systems. It's the quiet life for me.

Re:What price power?

[Anonymous Coward]

Well said.

I have enjoyed my 9 EPIA 800MHz systems for a while now. 1 is used as a media player (DVDs, Divx, MP3s, etc.) and the other 8 are configured in a Beowulf cluster.

Awesome little boards, and you can't even tell that they are running. Beats the hell out of the other 7 beige boxes I have running on my network.

My Pentium and AMD machines are quite effective as air cleaners, though (WHIRRRRRRRRR!).

Re:What price power?

[Zathrus]

I recently checked into the VIA boards... I'm setting up a home media server and wouldn't mind it using less power and making less noise (not as much of an issue since it'll be relegated to a closet eventually). But the Via chips don't cut it - they may be fine at office tasks, but their MP3 encoding times are abysmal (as is anything else requiring floating point).

Q3 may play on it just fine (for you... looking at the benchmarks I sure wouldn't call it acceptable), but Q3 is 3.5 years old now. The Via processors are not going to be up to running Doom3 - as stated they're really not up to running Q3.

No doubt, Via is great for playback only HTPCs, office workstations, and a lot of other tasks... but if you need to do audio or video processing (as in creation or editing) or play games, forget it. I'm not saying you need the top of the line Athlon or P4 either, but there is a happy (and inexpensive) medium.

Re:What price power?

[mrtroy]

Good points, good sir, but look at what the mini itx boards are aimed towards. They are not for the heavy gamers. They are not for video encoders. They are for the majority of people who want a quiet, decently fast, quality mobo.

Re:What price power?

[horster]

> They are not for the heavy gamers. They are not for video encoders.

seriously - the above was describing a heavy duty workstation, not a general purpose pc. for that, take out the lawn mower. i am seriously considering via platform for my next pc, the noise factor on my home built athlon is a big minus.

Re:What price power?

[Anonymous Coward]

I work from home, and have a network of 5 PCs of various specifications. You can tell when they are on because the floorboards in the hall vibrate, never mind the noise they make.

That's the builder's fault, not your PCs... my concrete floor never vibrates!!

Re:What price power?

[Zathrus]

I agree... but I don't even want one for video encoders or gaming. But the MP3 encoding times are so bad that even my Athlon 750 (current server system) will be faster for that.

Like I said, if I was building just an office PC or general usage PC it'd be a cinch - they're perfect for just web surfing.

Re:What price power?

[drinkypoo]

The thing about that is that video editing is becoming a more and more common task for home PCs. What with the easy availability of DV camcorders with IEEE1394 output, digital video is getting cheaper and easier all the time. Ditto for DVD burners. Hence, more people are going to want to do MPEG2 encoding.

plagarism

[synthe]

When I read the article text submitted by SoDaLaS, I realized I had read it before. Look for [H]ardOCP's news about the Athlon XP 3200+ posted yesterday at 11:50am.

"3200+"? What's the real clock speed?

[ArmorFiend]

What's the real clock speed of this beggar?

The last Athlon XP

[Jungle guy]

Acording to my sources, this should be the last Athlon XP to be released by AMD. Their next desktop processor will probably be Athlon 64, wich will need a different motherboard.

So my advice is for not buying a computer with Atlhon XP 3200, as your upgrade roadmap will be locked. It is better to buy a computer with a slower (and cheaper) Athlon, and wait untill the price drop to buy an Atlon XP 3200. Or wait for the release of Athlon 64 - it will be an excelent computer for video edition, 3D rendering and games like Unreal Tournament 2003 or Doom III.

Re:The last Athlon XP

[Anita Coney]

My god, what's up your butt? Jungle Guy's advice may have been a little obvious, but it is still utterly valid. What's with all the anger? I hate to imagine if he posted something totally stupid.

Re:The last Athlon XP

[swv3752]

I don't know about you, but every processor upgrade I have gone through has always resulted in a new Mainboard. Almost half have been to replace Mainboards, but I would not have upgraded if I did not have to replace something.

Mmmm....

[foxtrot]

The best thing about a 400 MHz FSB being available on an Athlon chip?

The 333 MHz FSB chips will drop in price!

Re:Mmmm....

[Joehonkie]

Now that's what I'm talking about!

I don't do it as much with hardware anymore, but I do love getting stuff that is still totally kickass and usable just because the "Next Gen" superceded it. Like how the newer 333MHz XPs pushed down the price of the older ones which are faster but have less cache. And a few benchmarks I saw had the older ones coming out on top!

IT DOESN'T MATTER

[w42w42]

It's the naming scheme. It is a 3200+ because that is what it's performance is relative to - a 3.2GHz Intel piece. The 400MHz FSB just allows AMD to take what would have been a slower part, and mark it as such.

Re:IT DOESN'T MATTER

[Anonymous Coward]

Actually it wasn't meant for comparison to Intel but a comparison of the Thunderbird Athlon running at that speed. However the XPs with the model number still do seem to be the Intels with that Clock Rating.

Today vs Yesterday

[nuggz]

10 Years ago you could do internet/email/word processing/spread sheets with just a 33MHz Intel 386 with 16 MB RAM. Today you need for the very same things a Pentium IV with 2 GHZ and 128 MB RAM.

I still use my p133 for many tasks, irc, email and personal server.

Web browsing on a 386/33, never did it, I had a 386/40. It was VGA (640x480 w16 colours), It was slow, the pages were simple. It was the only thing I could do at the time.

Now I browse with many windows, 24bit colour at higher resolutions (rarely anything as pathetic as 1024x768).
I can play mp3's without skipping a beat, along with movies. I was glad to get a .mod playing on my 386 without skipping.

We've come a long way, we do have overkill for many applications, but it isn't all waste. I think too many people who complain aobut how excessive it is today forget how relatively wimpy it was before it became mainstream.

Does anyone else remember how cool it was to have a 486 that would dir a directory listing faster then you could read it?

Re:Today vs Yesterday

[johny_qst]

Whoa... I wasn't the only one. What was really awesome was realizing that i could then | more so that I could peruse the output at will. Thanks... you took me back there for a couple seconds :)

Re:Today vs Yesterday

[RealAlaskan]

Does anyone else remember how cool it was to have a 486 that would dir a directory listing faster then you could read it?

I remember overclocking a Kaypro four by putting in a faster crystal and a 6MHz Z-80. It was so much faster than the IBM PC (a measly 4.77MHz clock) that the difference was astounding. SBASIC and Wordstar just flew!

By the time that the 486's rolled around, there just wasn't much novelty left in it for me.

Re:Today vs Yesterday

[MoOsEb0y]

I had no problems playing mods on my 286 with an 8bit homemade DAC. Albeit they were 4 channel, but I loved the music. In truth, many mods back then were better music than what is made professionally today. Anyone remember stevespam.mod ? perhaps 12thwar.mod ? I still listen to mods even today.

Re:Today vs Yesterday

[Anonymous Coward]

"Does anyone else remember how cool it was to have a 486 that would dir a directory listing faster then you could read it?"

Yeah, I miss my old Sinclair ZX-80. Now *that* was a computer!

man!

[0x12d3]

Have I really put off the upgrades for that long? There are FSB's faster than my processor? The funny thing is there is so much latency loading a lot of the modern software you really don't appreciate it! Outlook XP takes just as long to load as OE did back in 97. The xfer speeds have definitely improved (a _lot_) but rendering websites takes as long if not longer than it did 10 years ago (unless yer a fellow dillo user!). I should quit complaining; I'm as thrilled about bus speeds improving as the next guy (I do a good amount of hw irq intensive stuff) but jeez, It would be nice if avg. joe could see the improvements too and not just those of us compiling kernels on the weekend. So much of the hype causes aunt Ethel to upgrade every year with no appreciable speed improvements. But then I guess auth Ethel's never heard of a front side bus...

Re:man!

[Austerity Empowers]

I just upgraded my three PC's from pentium 2's to Athlon XPs. Putting in the slowest 333MHz FSB XP chip, ran about as well as the fastest (I could lay my hands on, 2800+). The difference over my previous setup of pentium 2's (@400MHz? can't remember) was night and day. I had a pentium 3 1GHz at work, wasn't much of an improvement, but DUAL DDR and the faster front side bus is amazing.

Re:man!

[RogerWilco]

The trick when upgrading is not upgrading your software (at least not
at the same time), then you really appreciate the increase. :-)

That's nice, but I'm sticking with Intel

[Anonymous Coward]

Sorry but too many negative experiences with AMD based systems. That's not to say AMD is at fault but their recommended chipsets are a little suspect *cough* VIA. I've had 4 Athlon based systems with massive failures in the mainboard (2 VIA based, 1 AMD760 and 1 SiS). Not to mention friends who have all had their own problems with these system. The other thing is that they run so damn hot, even with top-rated HSFs. I still have 2 up an running (nForce), it just seems to be hit or miss with these chipsets.

Anyway I have an Intel P4 2.4 with an Intel Mobo and the thing is quiet, cool, fast, extremely stable in both Win2K and Linux.

Re:That's nice, but I'm sticking with Intel

[Anita Coney]

I keep hearing from people like you who have trouble with VIA/AMD systems, and I'm just perplexed.

For myself, friends, and family I've built 7 VIA/AMD systems and each have been rock solid. Absolutely no problems. And with the money I saved buying the VIA/AMD combo versus an Intel product, I was able to buy better graphic cards, more ram, and larger hard drives.

Re:That's nice, but I'm sticking with Intel

[NSParadox]

I swore off AMD as well due to chipset problems.

The computer I built for my girlfriend, a first-generation nForce 1, has had absolutely no problems whatsoever.

The first AMD computer I built, an 800Mhz Athlon Thunderbird with an ECS mobo using a VIA KT133 chipset crashed every 12 hours to 3 days, which sucks since I was using it as a web and database server. Thinking it was a heat issue, I bought a big ThermalTake heatsink and tried to install it, but lo and behold I cracked the core because AMD refuses to update their packaging (who the hell still uses ceramic packaging anyway). No, I'm not a heatsink-installing newbie. This was probably the 10th computer I built. I was forced to turn my P3-450 desktop into a webserver to replace the system, and it worked flawlessly.

I then bought an Abit KT133A-based mobo and an Athlon XP 1.2GHz. The motherboard chipset fan started making sounds and died after only a few days, so I had to replace that. Then, when I upgraded my video card from a GF2 GTS to a GF4 Ti4200, I developed crashing problems in all kinds of settings. After doing a little bit of reasearch, it seems to relate to VIA's drivers (or perhaps the hardware itself). Flashing the BIOS made the problem worse, not better, and Abit has stopped posting BIOSes for my motherboard.

Then, I built a dual AMD Athlon MP 1.2GHz with a Tiger S2460 mobo (uses the AMD760MP chip), to be used as a game server. The machine worked great for 6 months, and then I started getting kernel panics. Turns out that the motherboard had corrupted various portions of my 4 sticks of 256MB Crucial registered ECC DDR. I was pissed.

I don't think the culprit is the processor. I think it's pretty clearly the chipsets. I will never buy another AMD or VIA chipset again after these experiences, which for the most part leaves me with NVIDIA and Intel. Hyperthreading looks really, really cool, so I think it's bye-bye AMD for me and hello to a company that's willing to support their processors with rock-solid chipsets and drivers.

Re:That's nice, but I'm sticking with Intel

[pecosdave]

I found this to be true on the socket 7 systems, K6-2 or higher. You're right, I agree, in those cases the chipset was liable. The Aladin 5 was the biggest piece of shit ever released, however with the K6 systems I had good luck. My first socket 7 system was a K6 166, I upgraded to a K6 233 and my mom is still using it. The K6 166 is still being used in my Netware file server with the only chipset that was worse, I can't even remember the name of it, but it seems like IBM was guilty somewhere along the lines.

However the only problem I've run into with all the Athlon systems I've built was cooling nForce chips. I've added lots of airflow to compensate, other wise very few problems from the slotted 750 up to the multi GHz rated range.

Re:That's nice, but I'm sticking with Intel

[BJH]

Hmmm... I seem to recall that I saw this exact same post after the Opteron article.

Can you say A S T R O T U R F E R ? I knew you could!

Re:That's nice, but I'm sticking with Intel

[DarkBlackFox]

I wouldn't blame the chipsets right away... Most problems as you've described are caused by the motherboard itself. It's a well known fact in the IT world that ECS motherboards are crap, and are oftentimes expected to crash.. i.e. it's an anomoly to have one run stable for more than 3 days. Same deal with Abit, Tiger, and to an extent Asus. The latter manufacturers put more of an emphasis on performance than stability. For the past 3 years I've worked as a system builder and technician for a small company, and have worked with virtualy every motherboard manufacturer on the planet. I've seen a number of very stable motherboards, mainly from MSI, Biostar, DFI, and a few others, with VIA chipsets, and not one of them had any sort of hardware problems, once all drivers are properly installed. I have seen and heard of many issues with SiS chipsets, but VIA is rock solid when integrated properly by a manufacturer.

Re:That's nice, but I'm sticking with Intel

[Anonymous Coward]

It's not as though they die immediatly, but 3-6-8 months (24x7) down the road the system will shutdown as if the plug were pulled. You go to reboot and get a beep code or maybe some power to the fans and nothing else (they're never reliable after that). I'm using good PSUs, HSFs, memory, even the CPUs are intact.

Newsgroups are filled with these stories so there is either a lot of incompetent builders or some engineering/quality control issues. One thing I have noticed (in the last 14 years of building my own systems) is that it seems the quality of components have definetly gone down in the last 4 years. Especially the stuff coming from Malaysia.

Re:That's nice, but I'm sticking with Intel

[Anita Coney]

Oh I agree that people are having troubles. I've read about it numerous times. I just never experienced it firsthand nor do I know anyone who experienced such problems firsthand. Maybe I and everyone I know are just lucky.

Re:That's nice, but I'm sticking with Intel

[SophtwareSlump]

My first (actually my second, I had a 486 DX4@120) experience with AMD processors was very bad. It turned out to be the craptastic Abit KT7 RAID Board I bought. All sorts of problems with it. Random lockups, HAL errors, etc. I swapped out every piece of hardware and it would still crash and hard lock randomly. I wrote it off as a demonic board and gave it to my friend. He now suffers with it, convinced he's only step away from getting it to run stable.

Since then, I've stuck to buying Asus and Shuttle boards and have had ZERO problems. Maybe that Abit board had a flaky chipset, who knows. I have an XP 1600 and a Thunderbird 1.33 running at home and they run fine. Never had a problem. They run just as stable as my PII@300.

I just can't resist the price/performance ratio of AMD chips when I go to upgrade my machine.

Re:That's nice, but I'm sticking with Intel

[Anonymous Coward]

The other thing is that they run so damn hot, even with top-rated HSFs.

Is this really true? I don't have two chips (AMD & Intel) of the same speed to compare but according to the published docs, the newer AMD chips dissipate less power than the Intel chips. Of course, I'm only looking at the 1-2 GHz range - am I missing something or does "AMD run hotter than Intel" just a myth?

Re:That's nice, but I'm sticking with Intel

[Dr. Manhattan]

The Aladin 5 was the biggest piece of shit ever released...

Amen, brother. I was lucky, mine ran stable with a TNT card, but a GF2 just killed it; I never got more than a few days without a crash. An old ATI Rage 128 flat refused to run until I basically disabled everything in the BIOS.

Re:That's nice, but I'm sticking with Intel

[Qbertino]

...All sorts of problems with it. Random lockups, HAL errors, etc. ...

Would those 'HAL errors' you mention be him not opening the pod-bay doors, chucking your comrades into solar orbit or is he just going on your nerves with nursery rymes whenever he's got a memory leak?

Re:That's nice, but I'm sticking with Intel

[gokulpod]

OS Name: Microsoft Windows XP Professional
OS Version: 5.1.2600 Service Pack 1 Build 2600
System Up Time: 11 Days, 2 Hours, 7 Minutes, 35 Seconds
System Manufacturer: ECS
System Model: K7S5A

Re:That's nice, but I'm sticking with Intel

[NSParadox]

Actually, Tiger boards are primarily used in high end servers and workstations.

The problems with the VIA KT133A are very well documented. It's not a board issue. MSI, Soyo, Abit, Asus, etc. all suffer problems due to the KT133A chipset.

Re:That's nice, but I'm sticking with Intel

[Hadlock]

my powerspec 400 mhz celeron running win 98 would happily run for a month or more at a time inbetween reboots... and that was a shitty ass computer. 11 days is nothing. my tibook has been running continiously for 18 months now (with brief reboots for OS updates & critical security patches) with 6 crashes, 4 of them software.

Re:That's nice, but I'm sticking with Intel

[Hoser McMoose]

VIA chipsets can work well, and they do for millions of users out there, but they can also be problematic. There are a lot of small issues with virtually every VIA chipset I've ever encountered. Fortunately each of these problems are very rare, and for most users you can get by without encountering any of them, or at least without any serious issues resulting from these bugs. However, every now and then one will crop up and bite you on the ass, and then it can be painful to fix.

Now, that being said, anyone you thinks that Intel is immuned to the sort sort of issue is a complete idiot. I've also had all kinds of problems with Intel chipsets. Their first bus mastering drivers for the PIIX4 southbridge (i430TX and i440LX chipsets) were TERRIBLE! The first set of drivers for their i8xx series of chipsets also sucked ass. They had the hugely publicized problems with the i820 chipset, and even some of their high-end workstation and server chipsets have had their fair share of problems.

However, in general, I've found VIA to be one of the most problematic chipset vendors (along with ALi). SiS and AMD have been a bit safer in my experience, and Intel is safer still. That being said, the best chipsets I've encountered have been nVidia's nForce chipsets. These have been the easiest to setup and most problem-free chipsets I've used.

As for the heat thing, Intel P4 chips use more power than AMD's Athlons, so all else being equal, Intel chips run hotter. The reason why people say that AMD chips run hot is because they use small heatsinks on their Athlons and huge monster heatsinks on P4s. Put a 80x80mm heatsink and a slow-spinning fan on an Athlon and it will run nice and cool while being virtually silent.

MHz is already meaningless

[TexVex]

AMD CPUs outperform Intel CPUs at similar frequencies. That's why AMD stopped marketing their processors based on their frequency. In some benchmarks, an Itanium running at 900 MHz outperforms 3 GHz Pentium IVs. Once upon a time, before clock multiplying, MHz meant more than it does now. But even in the 8-bit days, a 6502 running at 1 MHz would perform similarly to an 8086 running at 4.3 MHz.

Re:MHz is already meaningless

[Farce Pest]

I think you mean 8080 or Z-80, not 8086... The 8086 is a 16-bit processor with a 20-bit address space (IIRC). The 8088 was an 8086 with an 8-bit data bus. The 8080 (and faster clone, the Zilog Z-80) was an 8-bit processor.

The 6502 (1 MHz) compared so well against the higher-clocked 8080/Z-80 (usually 4-5 MHz) because of better pipelining and shorter instruction cycle times (IIRC).

Feeding the Troll

[drinkypoo]

First, overclocking works decent for a few people, but is not available to the masses for several reasons including technical difficulty and noise issues

Anyone can learn to overclock by reading one book. I suggest the motherboard manual. For instance, I was able to overclock my Athlon XP 1700+ to 1.66 GHz by simply increasing the multiplier from bios, and doing nothing else.

Second, overclocking is kind of dumb (expecting 10000 evil replies for that, but listen first) because if the board really could safely go faster, the manufacturer would produce it that way, and sell it for more!

This sentence actually led me to believe you might not be a troll and that you believe what you are saying. Either you are more clever than your post would suggest (a classic strategy of trolls) or you are one of those people who is technically unable to overclock, because anyone capable of it should know that the board is not the problem, as you say, but the processor.

Third, maybe everyone doesnt want their computer to sound like a jet is going off from the cooling needed to overclock, especially since as computers are getting faster, and more "stuff" is being put in smaller and smaller spaces, heat is increasing as well. Thats why mobos are coming with bigger fans, graphics cards are coming with giant fans that take a whole slot, etc.

Cooling needed to overclock? My modest but significant overclocking job did not require any additional cooling, or hardware of any kind.

Motherboards don't come with fans, except on the chipset, and those fans are small. Again you either show your ignorance or deploy a smokescreen. Processors sometimes come with fans, and it's true that their heat sinks are getting bigger on both those and other coolers in general, but that's not because it's going into a smaller space, it's because the newer CPUs have tended to dissipate more power in the form of heat. Athlon XPs are known to dissipate about 70W that way; The Athlon 64 reputedly only dissipates 40W, so your argument is fairly limp here also, even if we correct your obvious errors.

Now personally, I considered overclocking, fiddled with it, decided it wasnt for me, but I realize a small amount of people will do it. Cheers to them, but why can we not critically analyze a mobo without considering overclocking, which will benefit less than 1% of users! Lets look at the raw performance, and it should be sweet with this fat bus!

Now comes the obligatory "/. users are better people" blurb: Those of us who are here are more likely to be overclockers. Hence, a site with news for nerds definitely should include information on overclocking, since to many of us, that is stuff that matters.

This bus still has half the theoretical bandwidth of intel's latest offerings. Whether intel is using that additional bandwidth currently is another issue outside the scope of this comment/flame. The point is, this is not so much revolutionary or even evolutionary, but it's simply "about time".

Re:Feeding the Troll

[Slack3r78]

The Athlon 64 reputedly only dissipates 40W

Any good links about this? This is the first I've heard about that, so it piqued my interest. If it's true, Athlon 64 just keeps getting better as far as I'm concerned.

Re:Feeding the Troll

[drinkypoo]

I don't remember where I saw it, I think I commented on it in an earlier (but still recent) /. article, perhaps "what it will take for athlon 64 to succeed" or whatever that was called.

Incidentally my Athlon XP only dissipates 46W so I dunno where that 70W shit came from in the first place. I bet my GF4Ti4200 uses more power than that :)

I care again

[DeBaas]

Actually, I had stopped caring at 400 mhz. My laptop does (did) everything I want nicely at 400 Mhz. (did go to 192 mb memory right after buying it)

But recently I started creating DVDs on my desktop (Duron 800). And if you want quality mpeg2 files, that's very time consuming on that machine.
In the Pentium II 400 days, creating those at home was hardly an option. If I now get an Pentium IV, I can do this stuff a lot faster. That would be nice if start to backup my collection of VHS tapes.

So it's just a matter of what you do with your machine. We now can do more stuff at home, that we couldn't before.

It is all about percentages my friends...

[Anonymous Coward]

But now how does an o.cer look at percentage increase on AMD cpus: MHZ or PR rating? Ohhhhh yes, now I remember, it is based syntetic/tweaked benchies!!!!

Die AMD, die, for turning beautiful, exact, scientific concepts (MHZ), into artsy-fartsy-public-relations b.s. (PR).

Re:It is all about percentages my friends...

[Sj0]

You first.

The new Sledgehammer runs at 1.8Mhz. In a dual-processor combination, it can slaughter the Xeon 3.06Ghz in some applications, and holds it's own in most others.

If you think that AMD should honestly be trying to convince the public that 1.8Mhz the 1.8Ghz model of this is faster than the 3.06Ghz model of that, you've got a screw loose. Personally, I think their current system is accurate enough when comparing two processors of the same or different brands that it is a service to the customer. It's certainly easier than trying to figure out from Intel literature whether a 1.2Ghz Pentium 3 is actually slower than the same clocked Pentium 4.

Re:It is all about percentages my friends...

[Anonymous Coward]

Isn't SledgeHammer 64Bit? Is Xeon 64bit?

Re:It is all about percentages my friends...

[juhaz]

It's quite the other way around, AMD saved meaningful, well-defined scientific concept from being transformed into meaningless performance marketing crap (which it never could tell, and never wont).

Re:It is all about percentages my friends...

[Anonymous Coward]

So please do explain in a nice scientific way the meaning of XP rating. An equation would be nice.

Re:It is all about percentages my friends...

[mlyle]

The new Sledgehammer runs at 1.8Mhz. In a dual-processor combination, it can slaughter the Xeon 3.06Ghz in some applications, and holds it's own in most others.


DAMN that's some instruction level parallelism, if a 1.8MHz part can slaughter a 3.06GHz part. Overcoming a 1700x speed difference!

Re:It is all about percentages my friends...

[Sj0]

Well, you know -- AMD knows their shit. ^ ^

(in situations like this, I like to blame it on "being too oldskool". YEah. That's it, I've been using computers with 1->200 Mhz so long, a Ghz seems kooky and strange. ^ ^)

Use different architecture

[dmelomed]

MISC

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Anonymous Coward]

What matters is iNTEL's $500 CPU vs AMD's $500 (or $100 CPU).

That comparison is kind of difficult because when you compare the most expensive AMD to a similarly priced Intel chip the AMD is only a marginally better value, but as the prices go down the AMD power/dollar advantage gets larger.

Example: ~$320 will get you an AMD Athlon XP 3000, the most similarly priced Intel P4 is their 2.8GHz for ~$300, you're only getting a slight price advantage with AMD. However for ~$90 you can get a 1.6GHz Intel P4 or an AMD Athlon XP 2400.

Blatantly misleading.

[ProtonMotiveForce]

First, why do people insist on the "it's only 200Mhz, quad pumped" comments? Hint: 200 x 4 is 800.

Second, the speed rating is becoming a joke. The 3000+ couldn't even beat the 2.8GHz P4 in a lot of benchmarks. I guarantee you the 3200+ will lose in damn near every modern benchmark to the P4 3.06GHz.

So much for their credibility.

Re:Blatantly misleading.

[Phiro]

AMD's speed rating system is excellent. On average, AMD has _undercut_ the speed rating by 100-200mhz. Yes, you can probably pick out a few benchmarks that rely heavily on SSE2 so your precious P4 can "whomp" on an AthlonXP with the equivelent speed rating. (realize, whomp usually = beat by 5-10%) I can also pick out ten standard benchmarks for each one of yours that show the speed rating equiv. AthlonXP whomping your P4. ---- Don't get me wrong, I'm not an AMD booster, nor do I own _any_ of their stock. I'm just sensitive to peeps ripping on AMD's speed rating system - they did the best they could and they even put in plenty of margin for cranky people like you.

Hey Where are the adverts for the opteron

[eadint]

I keep hearing about this opteron proccesor, that will once again make the world safe for capitalisim.
where is it at what is all this flack about chinsy 32 bit proccesors.

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[bryanp]

The troth is that the only CPU mesure that matters is how long dose it take to rip and encode a DVD to DivX (One of the few tasks that still taks hours.) or whatever application YOU run which YOU feal is too slow on whatever system you have now.

With enough processing power and memory maybe more people would run spell checkers.

(yes, I'm an evil bastard who can't ignore the chance to take a cheap shot)

Underclocking

[Anonymous Coward]

My current cpu is underclocked, by 350 mhz, to save on heat and make it last longer. It is a thunderbird, 1.4.

And don't get me started on fans, my system has 9 fans total, and a "amd approved" fan that came with the processor. But hell, If I sty to play starcraft with my proc @ 1.4, it bakes and shuts down. So I run it at 1050, it stays cool, uses less power, and my system actually runs smoother.

Eat that you uber geek overclockers. I was once you, but I guess I mostly play PS2 games now, so the need for a dual overclocked 8 ghz machine to play unreal or C&C or something, is not what I desire anymore.

Is a large market for the high end desktops pushed by gamers? There is nothing I can not do on my 1.4 -> 1.05 ghz machine, compile the kernel, write code, use gimp, photoshop or premire, office, internet, some games, watch tv, ect. So you 3 gigers can do it 33% faster, and it takes me an extra 10 minutes for some tasks, I don't care, I am not in that big of a hurry.

Don't get me wrong, the urge to build a dual amd 3000+ machine with 4 gig of ram, 1Tb of hard-drive space, and 2 of the new gforce cards is still there, I just cannot justify spending the cash to replace something that works well.

But anyway, to each their own, keep pushing the limits of hardware, and keep pushing AMD and Intel for faster products. Its all good.

Re:Underclocking

[Sj0]

It sounds like there might be something wrong with your system, if it won't even run at 1.4Ghz. Personally, I'd take out some of those fans. Try to get a more efficient airflow going, and you'll be able to keep your machine twice as cool with half the fans. I installed a chimney fan in my case, and removed 4 or 5 other fans in the process. My system temp dropped from the mid 50s to the low 30s. (under load, I run at 37'. Without load, I'm usually closer to 34). The only fans I have that aren't directly attached to a processor are my chimney fan, my PSU fan, and an intake fan in the front of the case which may or may not actually do anything. Remember: less is more. Good airflow is infinitely more important than lots of fans. Just think of those Dells which don't even have a fan on the heatsink, and just use a shroud and a large fan to cool the entire system!

Re:Underclocking

[juhaz]

"Even at 1.4GHz"? 1.4G Thunderbird is still one of the hottest AMD chips out there, newer cores run quite a lot cooler. XP 2800+ and faster may catch it, but that's about it..

Re:Underclocking

[Sj0]

Yes, but there's nothing paticularly exceptional about the heat. It's hot, but it's not OC'er hot. In that case, because of that, a well thought out cooling system should be able to keep the chip running within operating specs.

MOST OVERCLOCKERS ARE IDIOTS!

[Anonymous Coward]

Especially the one above who loves noise, and probably doesn't know the meaning of stability!

Now, if they'd just pull their heads out....

[NerveGas]

... and raise the FSB of the AthlonMP's, then we'd have a *really* nice setup.

Of course, they don't want to risk hurting sales of the Hammer, but it would still be nice to have more than one option, for crying out loud.

steve

Re:Now, if they'd just pull their heads out....

[Phiro]

Consider the AthlonMP line dead. Sorry if you feel screwed, but it was never very effective. It makes no sense for AMD to put another dollar towards the AthlonMP line. AMD has has a very weak MP lineup for as long as I know, so releasing the MP Hammer (Opteron) first made alot of sense, shoring up where they are weakest first.

It also makes alot of sense for them to stop the AthlonXP line with the 3200 or so (no more major jumps).

The Athlon64/Opteron line is the entire farm, and AMD is betting it all. So far it looks really, really good. Twelve months from now, we'll probably know if AMD is going go bankrupt or get their stock back up to the $40 range. (currently trading around $8.50, fyi).

Re:Now, if they'd just pull their heads out....

[NerveGas]

The AthlonMP line was never effective?

Perhaps from a marketting perspective, but certainly not from a technological perspective. We took a $13,000 quad P3-Xeon machine, replaced it with a $3,000 dual AthlonMP, and guess what - the loads dropped in *half*.

They were (and are) very good performers. Their only limitation was a memory bandwidth limit. AMD went to all of the trouble to give each AthlonMP it's own independent bus, but they never took the time to mate that with a dual-channel memory controller, so that each processer could actually *utilize* the entirety of it's bus. Even so, they were (and still are) very capable machines.

steve

Re:Now, if they'd just pull their heads out....

[mikefoley]

Actually, it was DEC that gave the AthlonMP its bus design. It was based on the EV6 Alpha bus.

In the Intel solutions, the CPU's share the bus. In the Alpha/Athlon MP bus, each one is dedicated.

I agree that another memory controller on the Athlon MP Northbridge would have been stellar.

Re:Now, if they'd just pull their heads out....

[alienmole]

We took a $13,000 quad P3-Xeon machine, replaced it with a $3,000 dual AthlonMP, and guess what - the loads dropped in *half*.

I'm a big Athlon fan myself, but the above claim doesn't give enough info: what CPU speed and memory bandwidths did the Xeons have? There are some old slow Xeons out there, e.g. down to 500 MHz or so. If you're saying a dual 2GHz outperformed a quad 500MHz, that's not much of a surprise. Also, the ratios of CPU bandwidth to memory bandwidth has improved in recent generations from both manufacturers, which has made a big difference.

Re:Now, if they'd just pull their heads out....

[NerveGas]

We went from a 4xP3/700 Xeon (1MB cache) to a 2xAthlonMP 2000+. The Xeon system, with 4-way memory interleaving, gave a theoretical bandwidth of 400 MB/second, with the Athlons having to fight for the 266 MHz memory bandwidth.

Overall, we went from 2800 MHz of P3 Xeon cycles to 3332 MHz of AthlonMP cycles, and lost most of the cache while we were at it. That alone certainly doesn't account for *twice* the performance - although increased memory bandwidth and lower SMP overhead (2 vs. 4 processers) does help explain some of it. Getting twice the performance on an RDBMS application with only a FOURTH of the cache is pretty impressive! I can't wait to get my hands on some Opterons, with 1MB cache and oodles of memory bandwidth.

Supposedly, the ServerWorks chipsets in the 4xP3 Xeon has a crossbar-type switch in front of the processers, so that each one can get 100 MHz to/from the memory at the same time. I haven't dug deep enough to verify that, but I'm not entirely convinced that it's true.

steve

Re:Now, if they'd just pull their heads out....

[alienmole]

Thanks for the details. It seems to me the increased memory bandwidth plus CPU bandwidth would pretty much account for the difference. I don't know anything about that ServerWorks 4-way architecture, but no matter how you slice it on the Xeon side, your Athlons would have 2133MB/s between them, or 1066MB/s each. That can certainly compensate for a system with lower bandwidth but bigger cache. The cache can be thought of as simulating higher memory bandwidth (imperfectly), but at some point, the lower memory bandwidth just can't compete.

I agree the Athlons are impressive, though. AMD provides a nice antidote to the Intel approach of making the CPU ever more complex to achieve performance, whereas AMD seems to take a more straightforward (but smart) approach, and still performs better MHz for MHz.

And another thing...

[turgid]

Back in the day when 32-bit PCs were coming out, I was the one telling people that there was no point in buying a 386, stick to the 286, no software needs the 386 etc. Believe it or not, a 20MHz 286 ran 16-bit code faster than a 20MHz 386. Speed isn't everything. Those were the days. Anyway, what do I know? I'm just a deranged idiot.

Price Drop

[Wheaty18]

I've been looking into aquiring a 2800+ or a 3000+ recently. Hopefully this will bump the prices down a bit.

SOAP: Soon, all subroutine calls will use XML!

[Animats]

I still think XML is stupid. It's just a way to represent trees, and an inefficient one at that. Some agreed-upon binary representation would have been better. But parsing and text generation are great ways to keep all these new CPUs busy. One wonders if it's planned that way.

echo problems

[Anonymous Coward]

The first time I considered AMD was when Intel was pushing back with the Coppermine. It usually lost to AMD, but I was heavy into compiling where the cache architecture gave it a big advantage. The next time I got a machine for my parents and I did get an AMD processor, with a VIA chipset.

It was a disaster. I really wish I had spent the extra money and got an Intel machine. It went down constantly. Replacing the Creative Labs soundcard with a generic piece of junk, and the generic tulip-compatible ethernet card with a brand name one, plus lots of VIA 4-in-1 driver updates and a couple BIOS upgrades later it is a lot better. It still goes down sometimes for no explicable reason, though.

Since then I've stuck to P4 class machines. I would definitely consider an NForce2 chipset with an Athlon XP right now, but my most recent upgrade happened between price drops and the similarly performing Athlon was more expensive. That is no longer the case, but I'm no longer looking for an upgrade.

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Anonymous Coward]

Rip & encoding DVD to DIVX takes you hours? Well, with MPlayer and my old 1.8G CPU it works in real-time!

Ich can watch while encoding. Can you dig it? Don't ask which OS I use...

Athlon 64

[kg4czo]

I'm waiting for the Athlon 64. To get near 2.8Ghz performance out of a 1.6Ghz chip is just amazing. The heat factors should greatly reduce too. Nice.... :-)

Is anything above 150Mhz FSB for real?

[Qbertino]

I followed a diskussion recently in which a guy who should know (ITK Engineer) said that all this FSB hype is bogus since the real clock speed can't go over 150Mhz. We all now that true higher speed is achieved by a kind of 'frequenzy modulation' of the real lower clock speeds but I really would like to know if a 2000Mhz P4 with 400 Mhz FSB really is 4 times faster than a Celeron 500 Mhz with 100Mhz FSB.
What's the reality behing all this system clock craze?

Sounds canadian to me.

[Lispy]

Or maybe he got confused in this discussion?

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Anonymous Coward]

Funny thing is he got the phonetics right. My native language is not English, so I've always wondered why English language doesn't use phonetics for spelling.

I mean why is:

• Knights pronounced as "Nights" and not as "KA-NI-GHITS"
• Stephen King is pronounced as "STEVEN King" and not as "STE-PHAN King".
• (list is endless)

Why? Why? Why I ask you.

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Darren Winsper]

Simple; modern English is essentially a mixture of many languages, most from Europe. Back in ye olde days, the English language wasn't that sophisticated, so the solution was to hack bits of foreign language onto English, the end result being the somewhat "interesting" modern English.

Re:"3200+"? What's the real clock speed?

[Luminous Coward]

Allow me to speculate.

Barton (Model 10) comes in three flavors: 3000+ (2.167 GHz), 2800+ (2.083 GHz) and 2500+ (1.83 GHz). All other things equal, the 3200+ should run at 2.25 GHz, same as the 2800+ Thoroughbred (Model 8).

However, if AMD were to increase the FSB speed, you can expect the CPU frequency to be slightly lower. I would guess between 2.083 and 2.167 Ghz.

AMD keeps a definitive list up to date.

No noise athlons

[Nazmun]

Just buy the right low heat core athlon 1700+ not overclocked. Make sure you buy a nice fat heatsink and place a large but quiet fan on it. I have two athlon machies, one is very noisy and the other's fan is almost unhearable with the case open.

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Anonymous Coward]

"Troth" and "dose" are valid words and probably would have passed muster. What is needed is some way of verifying that a sentence simply makes sense.

I miss Grammatik because it would also test the reading level of a document.

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Anonymous Coward]

yes, I'm an evil bastard

Any relation to Fat Bastard? Or would that be Dr. Evil?

Does 1/2 Dr. Evil + 1/2 Fat Bastard = Evil Bastard? That seems to be an unlikely pairing, unless you are an evil clone.

If you are a clone, are you a mini-e.....vil bastard?

My mind reels at the possibilities. That is to say I am mentally staggering, not that my brain is not dancing an archaic fast dance.

I hope this is funny. Ha ha funny that is, not the other kind.

Whoa, look at the time....

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Anonymous Coward]

No, comparing encoding speed is NOT a good test of a CPU. Encoding involves next to no forks, which is great for the P4's ungodly large pipeline. But many REAL world apps have tons of forks in the code and can cause the P4 to have to flush the pipeline, run NOOPs, or throw out parts of the work it did because it is no longer valid. The intel POV is that if they can just make the frequency high enough, those wasted cycles won't matter. The way to test CPUs is in real world apps, NOT encoding!! If all you do it encode stuff, then you will always want to buy intel. If you like to play games then MANY times you want to buy AMD (it sways back and forth on that, intel and AMD keep managing to beat each other out). If you want a good CPU and you don't like getting ripped off, you always want AMD.

If that post didn't make sense to you then STFU cause you don't know what you are talking about.

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Forge]

Actually you agree with my point.

I.e. If Gamming is what matters to you then find out which CPU gives the most bang per buck in the games you like. I don't play many "modern" 3D games. Pure game play fun matters more than graphics to me so a lowly Celeron 300 MHz is fast enough.

However I do rip an occasional DVD (For personal use since dialup is too slow for sharing :). For that I needed a faster chip. Performance in that application is what mattered most to me.

What is computer science???

[willtsmith]

Ummm,

In case you missed it Computer Science is mostly concerned with software and largely couldn't give a rip about the hardware. The hardware is incidental incidental. That's what comilers are for.

Secondly, if you knew anything about code optimization, you would also know that most programs display the 90/10 rule. That is 90% of execution is performed inside of 10% of the code.

The key is:
1) making sure the code is correct and getting it built. Otherwise #2 doesn't matter.
2) Identifying the relevant 10% and optimizing by either using efficient data structures and algorithms or careful inspection and elimination of redundant/unecessary steps.

Furthermore, the actual SIZE of code doesn't really affect CPU performance. Larger code makes more demand on primary and secondary memory. It's the OS's job to sort out which bits of code to keep closest to the CPU (in L2 cache, primary RAM or swap-file).

I think you need to educate yourself as to what Computer Science IS. OR, re-direct your comments towards the Computer Electronics Engineers community. These are the guys who develop CPUs and other computer components, not CS guys.

For a final exercise ... I invite you to write a web browser straight in x86 assembly in the efficient manner as you proscribe. Then ... port it to PowerPC and personally witness the beauty of high-level languages like C++, Java and C# (VB excluded it is "high", but not in the right way :-).

400Mhz strikes a great chord with DDR400

[willtsmith]

So far DDR400 has sucked largely because it does not have good harmonics with 266Mhz front side bus speeds. Bad harmonics between memory speed and FSB speed leads to wasted time on both sides of the connection as the clock speeds just don't sync well.

A 400Mhz FSB on athlon will make a DDR400 useful and effective on Athlon based machines.

Re:400Mhz strikes a great chord with DDR400

[Slack3r78]

Well, if you try to use PC2700 in a pre-2600+ system, yeah, you're going to take a slight performance hit. When the FSB got bumped to 333MHz, that made PC2700 the way to go. Now with a 400MHz FSB, it actually makes sense to use PC3200. Really, it's not a problem as long as you have the sense to match your RAM speed to your CPU's FSB capabillity. PC2100 for 266MHz chips, PC2700 for 333's, and PC3200 for the 400's. What's so complicated about that?

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Captain Pedantic]

Ahem, even in realtime, a three hour DVD is still going to take "hours" to rip and encode.

Try Linux

[wirefarm]

Funny thing is, I just bought a barebones PC (18,000 yen ~= $130) and the slowest chip I could find for it was an AMD 2200 (~$60). Put it together with 512MB and RedHat8.
It doesn't really seem any faster than my 900Mhz. (Which had died.)
My server (www.wirefarm.com) is a 300Mhz that performs beautifully, even when it's encoding MP3s (it really does this) and serving up web pages.

- encoding mp3 / divx / etc
BladeEnc. Lame. Works like a charm on the 300. Learn the command line a bit and you'll be surprised what your hardware can really do. If you're encoding a DiVx movie, do you really need to *see* it process each frame? Leave out the eye candy and let it do its work.

- high quality divx playback
MPlayer or Movix2. Fullscreen with no jumps, stutters on the 900. (Plus, they happily ignore annoyances like region codes and MacroVision. Use your mouse wheel to fast forward and reverse as you like. Skip around the way you like...)

- games games games. War3 at 1280x1024 is beautiful =)
Wouldn't know. The only game I play is VI.
I think if I liked games, I'd probably buy a Playstation.

- Windows XP =/
Well, there you go...
Go download Knoppix and give it a try. I was using it this morning on a 233Mhz machine and it was fine for the web, email and Gimp. The Gimp was a wee bit slow, but did what I needed.

I guess I've just become an old fart about hardware - I used to crave faster hardware, but then again, I was using Windows...

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[dr.badass]

Marketing people love to trumpet all kinds of fantasy based figures when they talk about CPU spead.

The troth is that the only CPU mesure that matters is how long dose it take to rip and encode a DVD to DivX

Well, you know, it's probably not a good idea for marketing people to show DVD-ripping times, given the dubious legality.

Since we're joking about it...

[BrokenHalo]

We're heading into winter here in Australia. It's nice to know I don't have to buy a heater: all I have to do is upgrade my computer :-)

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Anonymous Coward]

ahh.. spare me the crap details..

Re:"3200+"? What's the real clock speed?

[odie_q]

It will most likely run at a multiple of 200 MHz, as it must sync against a 400MHz FSB. Perhaps 2.2 GHz?

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Arbogast_II]

Dang, someone dealing in real world!! Two Thumbs Up

Re:AMD $500 CPU vs iNTEL $500 CPU Re:400 MHz, 800

[Duds]

Well then use "Video encoding" and show someone doing a nice home movie of their wife and a stranger on a table with a ...........

Sorry, fantasy took over again.

Underclocking is MUCH cooler

[MarcQuadra]

I prefer UNDERCLOCKING. I have a barton running at 1200MHz, it's much cooler than it would otherwise be, and anything over 800MHz seems to 'float my boat' just fine. I'd rather have a quiet, cool, and reasonably paced system than one that heats my room and costs $40 in electricity every month.

I don't understand why you would want to overclock and risk data loss. I guess there are people out there who don't really care, they just want an extra 3 Frames/sec in their games. I strive for uptime and productivity with my systems, overclocking just seems childish and wasteful.

TX-PRO

[pecosdave]

thats the one I was thinking of thats probably worse than Aladin 5. Yech, I have a TX-Pro board that I actually use, with a K6-166, but I use it as a Netware server. It flat can't run Windows or X, and leaving a monitor on it will fry the monitor. But Netware doesn't REALLY need a monitor and isn't to demanding on it. Runs stable as long as I don't ever have to turn it off.