AMD Athlon XP 3200+ Released

SpinnerBait writes "AMD took the wraps off their next speed bump with the Barton core, the Athlon XP 3200+. This CPU runs with a 400MHz Front Side Bus at 2.2GHz and is targeted at competing toe to toe with Intel's latest P4. The benchmarks and review over at HotHardware, look pretty good but Intel's 3GHz/800MHz FSB P4 variant seems to squeak past it here and there. Regardless, more of that "yin" to compete with Intel's "yang" was released today by AMD and consumers will benefit again from the competition."

Comments

Still not faster..

[CausticWindow]

than my overclocked 2400 (in gigahertz that is).

Re:Still not faster..

[Anonymous Coward]

so what's correct mister monkey?

Re:Still not faster..

[SmileyByte]

2400 GHz??

WOW!!!

Re:Still not faster..

[Anonymous Coward]

2400 GHz??

WOW!!!

NEXT!

What if we really got 3.2GHz from AMD

[alta]

Wouldn't that be great! Damn fast! Come one AMD, drop the marketing and make it run at 3.2G!

Re:What if we really got 3.2GHz from AMD

[gilesjuk]

They use the marketing to show that their chips perform similar to Intel ones even though they're at a lower clock speed. Intel's P4 is designed to clock high and need to be clocked high as it can't do as much per cycle as the AMD.

Imagine the situation where car buyers only looked at the CC of the engine to determine how fast it would go, the AMD car would be a 2 litre but the Intel car would be 2.8 or something. So buyers would choose the Intel. Except the Intel engine has 4 cylinders and the AMD has 6 etc...

Re:What if we really got 3.2GHz from AMD

[AlgUSF]

No it would be like the P4 is a 3.4L 6-cylinder that puts out 140HP, and the AMD is like a 2.0L 4-cylinder that puts out 140HP. They both put out the same power, but the first one has more piston displacement .

Anyone who knows much about cars, knows piston displacement really doesn't mean shit, put it on a dyno and see how an engine really performs.

Re:What if we really got 3.2GHz from AMD

[fmaxwell]

Anyone who knows much about cars, knows piston displacement really doesn't mean shit, put it on a dyno and see how an engine really performs.

How many engines have you built? I've built a few and I know that greater displacement on normally aspirated engines usually leads to higher torque at low RPMs. Low displacement usually equates to lower torque and that the only way to make lots of horsepower from low-displacement is to design the engine for high RPMs -- because horsepower = (torque[lb.-ft.] * RPM)/5250. That's why a 1 liter motorcycle engine can produce upwards of 140 horsepower but would be completely unsuitable for powering a sedan that does fine with a 140 horsepower, large-displacement engine.

Re:What if we really got 3.2GHz from AMD

[AlgUSF]

A 1.0L motorcycle engine would power a sedan just fine. Just make sure you gear it down low enough.

Re:What if we really got 3.2GHz from AMD

[gilesjuk]

Well I think the P4 has less cylinders because it does less per CPU cycle so it needs a higher clock (it needs to rev higher). It's like a Honda engine, but the AMD is more like a higher cylinder engine where it revs less but puts out plenty of torque.

Re:What if we really got 3.2GHz from AMD

[Anonymous Coward]

They use the marketing to show that their chips perform similar to Intel ones even though they're at a lower clock speed.

I'm sorry but this is an urban legend. AMD's PR rating system for the Athlon XP is in relation to the Athlon Thunderbird, not the Pentium 4. An Athlon XP 1600 (1.4ghz I think) goes as fast as an Athlon Thunderbird would at 1600mhz; the PR system is to show that the XP is faster per mhz, especially considering early XPs had lower mhz ratings than the Thunderbirds also available at the time
If AMD really rated their Athlon XPs' PR rating system in relation to Pentium 4s, then an Athlon XP 1500+ would be a 1800, an 1800+ would be a 2200, a 2100+ would be a 2400, a 2400+ would be an Athlon XP 2700 and so on.

AMD's rating system has cost them alot of sales - even technical types underestimate how fast an XP system is in relation to a P4. Just because the P4 3ghz Xeon is fastest doesn't mean everything below it is faster than AMD's 'similar' offerings, despite the misleading PR rating. My next CPU will be an Athlon XP 2400+, which I shall overclock to 2.2ghz; P4 2.8ghz speed for about 98$ US is pretty good I think

Re:What if we really got 3.2GHz from AMD

[fmaxwell]

A 1.0L motorcycle engine would power a sedan just fine. Just make sure you gear it down low enough.

Untrue. The area under the torque curve very much influences the driveability of an automobile engine. If you have a peaky engine, such a 1-liter motorcycle engine, the powerband is insufficiently narrow to be used in a car, regardless of gearing.

Gearing the car down low will provide you adequate off-the-line performance, but what happens when you shift from first to second and your engine speed drops out of the peaky powerband? Answer: The engine bogs down.

Want another example? An OS model airplane engine that displaces only 9.95 cubic centimeters produces 1.8 horsepower at 16,000 RPM. That's with .2hp of the maximum allowed for mopeds. But do you think that the 3.3 inch long model airplane engine could power a moped? Of course not. It has too narrow of a powerband.

Re:What if we really got 3.2GHz from AMD

[macdaddy357]

It does as long as you have a liquid nitrogen cooling system. If not, it will melt into a blob of goo, and might cause a fire.

It's also the last 32 bit Athlon.

[Pop n' Fresh]

"AMD took the wraps off their next speed bump with the Barton core, the Athlon XP 3200+."

It's also going to be the LAST speed bump with the Barton core. AMD's next Athlon is going to be 64 bits:

http://news.com.com/2100-1006_3-1001106.html?tag =fd_lede1_hed

Re:It's also the last 32 bit Athlon.

[ryszards]

There will be other 32-bit Athlon's, 'Thorton' for one, a return to 256KB L2 when Athlon64 hits to turn the 32-bit range into the Athlon64's Duron equivalent.

They might not get any faster, but there will be more.

Oh man...winter just ended...

[MeanE]

and I already had bought a space heater.

Re:Oh man...winter just ended...

[Schezar]

You know, AMD procs don't have the heat problems they used to have. If they overheat your case, it's usually due to a poorly made case or some other factor. Like, say, a modern video card... -_-

Re:Oh man...winter just ended...

[Anonymous Coward]

he probably bought a p4, they run hotter than athlon

Re:Oh man...winter just ended...

[TheViffer]

Humm ....

Maybe you should read the original posters comments again. He did not say anything about "overheating" the proc. In fact he is dead on when he mentions space heater because of the massive heat exchange that would take place with two overclocked Athlons in the same case.

I run three-five boxes in my office at a time. In the winter the rest of my house is around 72 F. My office is a toasty 78F from the heat dispensed from the boxes. Great space heaters.

I am not saying what you mentioned to be untrue. In fact its dead on, Aluminum heat sinks (instead of copper), worn down fans, and underpowered power supplies will hurt you.

Re:Oh man...winter just ended...

[FroMan]

You are not in the midwest are you?

I had frost on the old windshield this morning.

Old man winter isn't quite ready to die, but if I find him...

Re:Oh man...winter just ended...

[archen]

And the sad thing is that it's been pretty chilly where I live and my landlord already cut the heat...
so I actually leave my computer on all the time which actually does heat my living room somewhat.

I sort of regret getting an LCD so soon too...

Re:Oh man...winter just ended...

[sketerpot]

I am in the midwest, and I wore shorts and short sleeves outside at 8:00 AM this morning.

What part of the midwest are we talking about here?

poor AMD

[MacOS_Rules]

...compete with Intel's "yang"...

Poor AMD, releasing a faster proc just because they feel...inadaquate...

I bet they drive SUVs too. A shame, really.

Athlon rating system over-rated?

[Imabug]

Based on what I've read of the 3200+ at Tom's Hardware and Tech Report, the CPU performance is good, just slightly better than the 3000+, but still gets trumped by the P4 3.0 and 3.06 CPUs. Both call into question the validity of AMD's CPU rating system, and judging from their benchmarks, rightly so. The 3200+ is also supposed to be more expensive than the P4s, which combined with the dissapointing performance may limit it's popularity.

It will be interesting to see how the 3200+ performs when overclocked.

Re:Athlon rating system over-rated?

[Psiren]

Both call into question the validity of AMD's CPU rating system

Unofficially perhaps, but officially the comparison isn't to Intel chips, but to AMD's older Thunderbird processors. A 3200+ is supposed to give about the same performance a tbird would, if it was clocked to 3.2GHz.

Re:Athlon rating system over-rated?

[Henry V .009]

Okay -- for the AMD fanboys out there -- what would you do if Intel tried that exact same line of reasoning on you? (Of course, they don't have to, since they've got the fastest chip on the block.)

Re:Athlon rating system over-rated?

[Troed]

AMD's rating system does not compare to Intel CPUs but to old Athlons. Thus - AMD's probably correct, and naive customers feel cheated.

Re:Athlon rating system over-rated?

[Anonymous Coward]

Well gee, there a useful comparision. I'll just imagine how fast my Thunderbird 750 could go if I overclock it by 2.5Ghz Mmmm, yeah thats pretty fast!

Completely useless marketing crap. Why not at least make it amusing, and set the base really really low. E.g. the Athlon 40,000,000+, where the baseline is an Intel 8086 running at 4Mhz. Probably the cost of the additional ink they'd need for the extra 0's..

Re:Athlon rating system over-rated?

[turgid]

Well, say intel used a performance rating to compare the P4 to a PIII, and took the 1GHz PIII as the baseline. Since the Pentium 4 does a lot less per clock cycle than the PIII, they'd probably have to call their P IV 3.0GHz a "P IV 2300+" or something.

Re:Athlon rating system over-rated?

[Psiren]

Of course, they don't have to, since they've got the fastest chip on the block.

Yep, they have. But I find it interesting that if both chips were running the same true clockspeed, the AMD would be faster. It leads me to believe that Intel rely far more on being able to ramp up their clockspeed than they do on creating a better chip. For that reason alone I prefer AMD.

Re:Athlon rating system over-rated?

[ViGe]

Completely useless marketing crap. Why not at least make it amusing, and set the base really really low. E.g. the Athlon 40,000,000+, where the baseline is an Intel 8086 running at 4Mhz.

Have you ever really tried overclocking an Intel 8086 to 3Ghz? It actually is pretty fast, but as it does not have any protected modes as they arrived in the 386, the only OS it can run is DOS. The dir command however is pretty fast.

Re:Athlon rating system over-rated?

[Anonymous Coward]

yeah, then we can have the athlon xp 40MEGA+
finally!

Re:Athlon rating system over-rated?

[angle_slam]

It leads me to believe that Intel rely far more on being able to ramp up their clockspeed than they do on creating a better chip.

That's silly reasoning. You could also rationalize this by saying that AMD is incapable of making a chip that runs at 3 GHz and is therefore inferior.

Re:Athlon rating system over-rated?

[Psiren]

Okay, explain why the Intel isn't of similar performance at the same clockspeed? Especially considering Intel's vastly superior R&D and financial support.

Re:Athlon rating system over-rated?

[Anonymous Coward]

thats the biggest line of bs i've ever heard. amd ALWAYS introduces an XP XXXX+ that is exactly the mhz intel just released or what intel is expected to release next. 3200? perhaps in reference to the 3.2ghz intel is expected to release soon? give me a break. amd is getting pretty low and is starting to release more marketing than product.

Re:Athlon rating system over-rated?

[IWannaBeAnAC]

Because Intel found that the best way to improve overall performance is to increase the clock speed, even at the expense of making the per-clock performance worse.

They are right too: where has almost all of the increases in performance come from? Not from doubling the number of instructions processed per clock every 18 months, that is for sure.

Re:Athlon rating system over-rated?

[joto]

Have you ever really tried overclocking an Intel 8086 to 3Ghz? It actually is pretty fast, but as it does not have any protected modes as they arrived in the 386, the only OS it can run is DOS. The dir command however is pretty fast.

Yes, I put it in a closed system of liquid nitrogen, streaming over the processor at about 30mph. My only problem nowadays is to find a good CPU-slower utility to play all those old DOS games.

Re:Athlon rating system over-rated?

[mmol_6453]

It's the same reasoning that helps me choose OSS software over non-OSS. Are you going to call that silly, here on Slashdot?

Re:Athlon rating system over-rated?

[wbattestilli]

I have long suspected that the industry standard benchmarks have gotten a bit crazy in the past year. http://www.theinquirer.net/?article=9445 does a pretty good Job sumarizing my thoughts. The benchmarks don't add up. Last year's WinWhatever benchmarks give totally different results than this year's, even on new hardware. I actually think that AMD is *trying* to be genuine with their rating system, but I also think that special interests have corrupted mainstream benchmarks to make them an unusuable guide.

Re:Athlon rating system over-rated?

[psavo]

haha, Tom is Intel shop, and pretty lame at that one, too.
Just look at his arrangements and you'll easily see what is wrong with them.

Re:Athlon rating system over-rated?

[TheViffer]

"It leads me to believe that Intel rely far more on being able to ramp up their clockspeed than they do on creating a better chip."

Go get an AMD 2400 XP and a Pentium 2.4 and pull the heat sinks off while they are running.

Sound unlikely? Okay lets try again.

A wire gets stuck in your cpu fan.
Your CPU fan dies.
The epoxy connecting the cpu/sink drys out.

Tell me which one is still working (and motherboard) and then tell me who makes a more reliable "better" chip. The AMD will toast itself, more then likely the board, and whatever else it can get its pins onto. The Intel will underclock itself and save your system.

Running AMD's is like handling C4 .. if your not careful your going to get burned. I myself run them, but never feel as confident as I do with the Intels.

Re:Athlon rating system over-rated?

[sucresemoule123]

or maybe a P4 2300-

Re:Athlon rating system over-rated?

[Penguinoflight]

You should look at TCO, and I'm not just talking about support. RDRam is rediculous, and a complete p4 system ends costing a lot more than a comparable athlon. HotHardware.com did say it overclocked well, up to 2520mhz, a 13% increase, and 320mhz.

I'm waiting for a x86-64.

Re:Athlon rating system over-rated?

[drinkypoo]

The reason they doubled the clock speed rather than doubling the number of instructions executed is that OoO becomes more difficult and the complexity goes up. Hence the higher speeds. Unfortunately, aiming for higher speeds instead of more efficient operation in the first place meant a longer pipeline which meant you need better branch prediction and it apparently just didn't get it, or we're reaching some theoretical limit of branch prediction. Which I doubt, considering Intel has become the Microsoft of hardware. They can't innovate, so they buy innovators.

AMD increased the number of functional units AND increased the clock rate. Intel went for the clock rate. Both chips are about the same speed. AMD chips are vastly cheaper. What more do you need to say or know?

Re:Athlon rating system over-rated?

[Dun Malg]

Tell me which one is still working (and motherboard) and then tell me who makes a more reliable "better" chip. The AMD will toast itself, more then likely the board, and whatever else it can get its pins onto. The Intel will underclock itself and save your system.

You sound like a weirdo friend of mine who drives an old Volvo. He tries to justify the money he spends by going off on its safety features. A thermistor on the CPU does not in itself make Intel's product a better performer.

Re:Athlon rating system over-rated?

[gm-7]

Well I think a Thunderbird would burst into flames at 3200.... ;) Is that the result they were going for?

Re:Athlon rating system over-rated?

[Anonymous Coward]

Well.. You probably should know that the AMD 3200+ is slower than the Intel P4 3.0GHz, and more expensive (costs 9% more than the 3.0GHz and 14% more than the older 3.06GHz which also seems to be faster in tests.) Of course, prizes may vary from place to place, but I would not say vastly cheaper :)

Re:Athlon rating system over-rated?

[IWannaBeAnAC]

Well, until a year or two ago the Intel approach was undeniably the winner. That is how they managed to get to be king of the hill in the first place. But for sure, branch prediction has huge theoretical limits. Consider:

while (1) { if (rand() % 2) foo(); else bar(); }

How do you branch-predict that? Beyond simple for-loops (predict correctly N times in a row, predict wrong exactly once at the end of the loop) it gets very difficult.

But branch-prediction is only one part of the total pipeline performance, and you are arguing the wrong way anyway. Overall, it is the slower clock speed processors where pipelining is more important. In these processors you need to make sure the pipeline is full to get more operations per clock, to compensate for the slower clock speed. At 3GHz, it doesn't matter much if your code doesn't pipeline very well. At 500MHz, it is critical. That is why pentiums are such good general purpose processors, and why Itanium sucks.

I agree with you that they now play the thug to stay on top (burying alpha, doing a huge con job with Itanium etc etc).

Re:Athlon rating system over-rated?

[IWannaBeAnAC]

Small correction, I didn't really mean to say "year or two ago", I really mean up until the time around the K7 came out.

Re:Athlon rating system over-rated?

[angle_slam]

If you choose OSS because you don't approve of MS's business practices, that's fine. If you choose AMD over Intel because it's cheaper or because you don't approve of Intel's practices, that's also good. (Personally, my home computer runs a 1 GHz Athlon). But to say that Intel is inferior because of a design choice IS silly. Sure, at the same clock rate, AMD is faster. But that misses a very important point: Intel runs at a faster clock rate. One could very easily say the failure of AMD to run processors at 2.5+ GHz, which Intel has been doing for nearly a year, makes AMD "inferior". Or the fact that AMD chips require more cooling makes AMD "inferior".

You could say you are choosing AMD over Intel because it is better engineered. Or you could simply say you are choosing AMD because it's cheaper or because you hate Intel. I think the latter two choices are more honest reasons.

Re:Athlon rating system over-rated?

[ZeLonewolf]

I actually think that AMD is *trying* to be genuine with their rating system, but I also think that

Consider:

Athlon 2600XP+ - 2133MHz
Athlon 2800XP+ - 2166MHz

You call that genuine? A 1.5% increase in clockspeed somehow equals 7.5% perfomance increase??

Re:Athlon rating system over-rated?

[be-fan]

Who uses RDRAM anymore? The newest Canterwoods are all dual channel DDR.

Re:Athlon rating system over-rated?

[pod]

It leads me to believe that Intel rely far more on being able to ramp up their clockspeed than they do on creating a better chip. For that reason alone I prefer AMD.

Why? And if AMDs are so underclocked compared to the P4, why do they run so much hotter? Why would you prefer a slower hotter chip over a faster cooler one? Intel's architecture can be clocked higher and higher as processes improve. How far can AMD go? Not far, seeing as the architecture is dead and being replaced with a 64 bit one, so you can bet the ratings will be even more skewed and fixed at whatever AMD feels like.

Re:Athlon rating system over-rated?

[Anonymous Coward]

AMD chips are NOT vastly cheaper. A few dollars difference at the high end doesn't qualify as 'vast'. The low end has always been AMD's domain anyways, they'll slash prices there till it hurts just to maintain share.

Re:Athlon rating system over-rated?

[Anonymous Coward]

they don't run hotter, the p4 @ 3ghz is the hottest chip around (in the consumer market, at least)

Re:Athlon rating system over-rated?

[Jeff DeMaagd]

A 3200+ is supposed to give about the same performance a tbird would, if it was clocked to 3.2GHz.

If this is really true, then the AMD engineering rep that I talked to got it wrong. He sayd that the rating was in comparison with the most directly equivalent MHz/GHz Intel chip. He said _nothing_ about the older AMD processors, and I would think such a comparison wouldn't be easy to make or relevant. It might also show that a 3.2GHz P4 might not have any gains over a 3GHz version. Or the series of benchmarks used internally by AMD to determine the "+" number is somewhat different than that of what the review sites use.

Re:Athlon rating system over-rated?

[Master Bait]

You need an old Athlon to pull off that FUD stunt. You also need to be deaf, so that you can ignore the wailing beep-beep-beep as the high CPU temp trips the warning.

Re:Athlon rating system over-rated?

[ponos]

>special interests have corrupted
>mainstream benchmarks to make them an
> unusuable guide.

This is true, to a certain extent!
People used to benchmark with, say,
Office scripts or other "office" jobs.

Most modern benchmarks are almost
100% multimedia (Internet content
creation, Divx, MP3 photoshop etc
come to mind). This is very convenient
for Intel because P4 is a multimedia
design (long pipes, high MHz, small
cache, fast FSB, SSE2) designed for
serial operations (small loops, no
branches) with huge data sets (fast
memory is good here).

The Athlon is a completely different
design and it cannot compete. I am
really wondering how it performs e.g.
with kernel compiles or other
system scripts.

Anyway, this reminds me of the situation
with K5 and Quake. Until the era of
Quake (and Pentium Classic) all
competing processors (AMD K5, K6,
Cyrix) had very very slow FPUs.

Quake was the first widespread game
to use FPUs (it relied on the FPU to
compute 1/z with fdiv) and even though
the K5/K6 were VERY fast with integer
code the Pentium Classic had vastly
superior performance under Quake
(popular benchmark at the time!).

Anyway, this whole situation is not
necessarily the fault of the benchmark
designers. Frankly, even though the
AMD may be faster at e.g. Excel I don't
intend to use huge multi-megabyte
spreadsheets (most of us don't) but
I do intend to use MP3/MPEG2 etc
algorithms for which (like Internet
Content Creation) the P4 may be
a better choice for those that can
afford it.

Actually I do own an Athlon 2200+
and I am very pleased with its
performance mostly (and most
importantly) because it is CHEAP.

(AMD always has better MIPS/FLOPS
per $, that's why I buy AMD)

Re:Athlon rating system over-rated?

[Anonymous Coward]

I thought AMD has the thermister being on board instead of on-die like intel, such that the latency was the killer.

Re:Athlon rating system over-rated?

[_avs_007]

You are assuming that you are around when it happens. Your server may be in the basement, and you may be in the bedroom.

Re:Athlon rating system over-rated?

[Anonymous Coward]

seeing as rand() just multiplies by some big number, it's not very hard to predict what it's going to be.

Re:Athlon rating system over-rated?

[dnoyeb]

ACtually increasing clock speed requires no input on the part of the development community. But building a better architecture can require some compiler optimizations to take full advantage. Intel knew the lazy man theory would win out. But only temporarily as compilers are upgraded eventually.

Re:Athlon rating system over-rated?

[fault0]

Uh, if you pull the heatsinks off of a Athlon XP 2400+ and p4 2.4, both of them should power down.

I guess you haven't heard that AMD has put in on-die thermal protection on all of their chips for a while now. All Athlon XP's have it, afaik.

Re:Athlon rating system over-rated?

[fault0]

Uh, difference in clocks speeds DO NOT SCALE LINEARLY in terms of performance. This is true in some instances more than others, of course.

Re:Athlon rating system over-rated?

[darc]

That's FUD. The latest P4s, or actually, most P4s, run hotter, or as hot as AMDs. So do some research before blindly bashing AMD.

Re:Athlon rating system over-rated?

[Slack3r78]

And while AMD was focusing purely on ramping clockspeed, they were beating Intel in that field. Who was first to 1 GHz? Here's a hint - not Intel. What's happened in the last year or two with AMD is actually twofold:

1. AMD realized that they couldn't continue to ramp up clockspeed indefinitely. They shifted from working on increasing pure speed to increasing efficiency per clock cycle. If you want an analogy - what good would a 300hp car be if you only had 100lb-ft of tourque. (I realize that it'd be damn near impossible to get a disparity that far fetched out of a car, but it's just an example)
2. It's also worth keeping in mind that AthlonXP development has lagged for the past year because AMD has, from what I understand, had the majority of their engineers assigned to work on Hammer - the home version of which is 6 months behind schedule at the moment. In short, AMD was planning on the Athlon64's being available much sooner than the reality will be. If I'm wrong with this information, please correct me, but as gung-ho as AMD has been about Hammer, the reports I read saying as such just made sense.

I think AMD's approach is proving to be successful so far. It's easier to get discouraged when a 3000+ doesn't really match up point for point with Intel's 3.06 GHz offering... until you realize that the true clock speed of the Intel is 36% (!!) higher than the AMD. Now, I'm taking a cursory glance through a few benchmarks here, comparing Intel's 2.2 GHz chip with their 3.06 GHz chip - it generally lags behind by about 20% in the benchmarks I'm looking at. When you take that into account, suddenly AMD's little 2.2 GHz "3200+" being able to perform within +- 10% most of the time seems far more impressive - and validates AMD's claims that efficiency is just as important as pure clock speed. Remember, Moore's law says computing power will double every 18 months - not raw computing "speed."

Re:Athlon rating system over-rated?

[Slack3r78]

Again, you're wrong. The AMD's have a safeguard as well - it's just different from the P4's. Instead of underclocking, when the CPU hits a critical temperature OR if the fan stops running the machine makes the "beep beep beep" the original poster mentioned... and turns itself off. Since the Athlon is supposed to be a home chip, not a heavy duty server, this is perfectly acceptable behavior, and won't leave the user wondering why their machine suddenly got slower.

This has been a feature AMD has required in ALL AMD Approved mobos for almost a year now (I believe it went into effect like June/July 2002). I can tell you first hand, that if I pulled the side cover off the Duron machine I'm using right now, and put my finger on the CPU to stop it, the computer would beep angrily at me, and then safely turn itself before any damage is done. Personally I like that approach better - you KNOW there's something wrong with the hardware and what to check, rather than running in circles chasing a "software performance" degradation wild goose.

Re:Athlon rating system over-rated?

[Slack3r78]

You really need to get a clue, man. Yeah, the old Thunderbird Athlons ran hot as hell - but that was FIVE CPU cores ago by my count. A P4 will almost always run hotter than a comparable Athlon.

Athlon's are not "underclocked." AMD made the decision about a year ago to focus on processor efficiency rather than raw speed. The fact that a chip 36% slower in raw speed puts up similar, and sometimes better, performance numbers to its Intel counterpart speaks volumes. (3200+ vs 3.06GHz)

As far as AMD's architechture being dead, funny that Barton core thing that they just upped the FSB on twice in the past 6 months. Yes, Athlon64 is on the way. Yes, it is a Good Thing(TM). But the 32 bit Athlon architechture is far from a dead end. AMD just has something better to move on to.

I don't have a problem with people preferring Intel over AMD - there are plenty of legitimate reasons, depending on your application. For example, if you were going to be doing mass volumes of video encoding, buying a P4 would make sense. But it just annoys the hell out of me when people base their preference on "reasons" that are complete BS.

Re:Athlon rating system over-rated?

[Slack3r78]

If that's the case does that mean the next chip core's gonna be code named "Pheonix"? ;)

Re:Athlon rating system over-rated?

[_avs_007]

Sure, but isn't the thermistor not on die on the AMD's? Didn't Tom's hardware say that the latency caused by this, is what caused the processor to fail?

I would rather the processor slow, than stop, because then at least productivity won't drop to nil. Again, like you said tho, for home use its probably fine. And I'm pretty sure, if this happened on a P4, you won't be going on a wild goose chase, it should be pretty easy to track down.

But anyways, back to your regularly scheduled program....

Re:Athlon rating system over-rated?

[Slack3r78]

The Tom's Hardware article was the reason for the change to the system they're using now. Again, when I said wild goose chase, I was referring to your normal, off-the-streets user, not a geek. A geek probably could track the problem down fairly easily, but your average user won't. The machine just not working is more likely to get their attention and get the problem fixed than the chip simply underclocking itself is. Like I said, it would suck on a server, but the Athlons aren't for running servers. Hence why I find their behavior preferable for the every day user than I do the P4's.

Re: Athlon rating system over-rated?

[Black Parrot]

while (1) { if (rand() % 2) foo(); else bar(); }

> How do you branch-predict that?

It has been a while since I took my architecture class, but I thought I remembered that they were working on designs that would execute along both branches simultaneously, and then throw one branch of the execution away a few cycles later when it became clear which branch should commit.

Re:Athlon rating system over-rated?

[Elwood P Dowd]

Remember, Moore's law says computing power will double every 18 months - not raw computing "speed."

No, no it doesn't. Go look it up.

Re:Athlon rating system over-rated?

[angle_slam]

Who was first to 1 GHz? Here's a hint - not Intel.

Who was first to 2.4 GHz? Oh, that's right, Intel reached that over a year ago while AMD still hasn't reached it.

The bottom line is simple: AMD is generally less expensive to operate at the same benchmark speed. But Intel is faster at the high-end, both in clock speed and in benchmark speed.

Re:Overkill

[afidel]

There are plenty of uses for fast cpu's that don't involve bloated software, hardware synthesis (ask Xylinx about it), media encoding and creation (software music synths are cpu hogs), high detail 3D visualization, genetic algorthims, etc. These are all apps I have used either professionally or recreationally (or both) and they all will have no problem scaling to any cpu based on moving electrons.

Quake 3 benchmarks again

[BenjyD]

No! They did benchmarks using Quake3 again. Who cares about Quake3 performance?

How about useful benchmarks - can I have a 10 plane dogfight over the front line in a Falcon 4SP campaign at 40fps+ with all the settings maxed out? My guess is no.

Re:Quake 3 benchmarks again

[afidel]

Yeah I was overjoyed by the Anandtech article on the new generation of cards from Ati and Nvidia because they had a page dedicated to minimum framerate, and what it showed was that even the not yet released $500 dollar Nvidia card could muster more than 23fps at the slowest part of the game, that's what you will notice, not the faster than your monitor can refresh 200+fps.

Re:Quake 3 benchmarks again

[Cthefuture]

Duh, because Quake3 is one of the very few engines that scales well with faster processors. It pretty much runs as fast as your CPU and graphics can handle because it's written down at the bare metal level.

It's a more "raw" benchmark than the rest of the games out there. Other games often have many strange and bizarre bottlenecks due to the way there were written.

The Q3 benchmark is not about showing how fast the game is. It's about showing how fast the processor is compared to older versions.

Re:Quake 3 benchmarks again

[Arethan]

Technically, FPS is likely to be more limited by your GPU than the processor. In these benchmarks, the CPU is going to spend more time compressing textures, shifting data around, and trying to push data down the AGP bus than anything else. I'll be more impressed when they start using software rendering when they do these benchmarks.

Realistically, almost all of these game based benchmarks are crap. If you want real statistics, have it render CGI movie frames. THEN see how it compares to other CPUs. Until you do that, you aren't really looking at anything other than the CPU's ability to feed the graphics card.

Re:Quake 3 benchmarks again

[Anonymous Coward]

yep, the quake 3 engine is a masterpiece, no doubt about that (and doom 3 is soon out *drool*).
remember that it took around 2 years from the game was released until the hardware availible could run it at the speed the hardcore gamers wanted (well, i don't think it still can if we're talking about 1600x1200).
competetive quake 3 gaming needs HIGH frame rates, and it needs to be STABLE at the maxfps you cap it to.

i wonder how long it'll take before we have hardware that'll let you run doom 3 with all features on and _never_ drop below 120 fps (in 1024...i don't even dare to think about 1600x1200)

Re:Quake 3 benchmarks again

[Slack3r78]

Finally, someone who agrees with me. I can understand using flight sims as they tend to have hardcore physics engines, which will take advantage of the differences in CPU power... if they would turn all the graphics options all the way down/off. Unfortunately, as scientific as these benchmarking sites like to make themselves look, I've never felt like they've done all that great a job of REALLY finding the differences between CPUs.

Better benchmarks....

[Anonymous Coward]

Here

System Recomendation

[Bob McCown]

With all the new hardware/mobo's/etc coming out lately, and me just about ready to upgrade, WWSCD (What would Slashdot Crowd Do) for a decent priced, fast combination? Id like onboard LAN and sound, and a decent graphics card. HELP!

-=Bob

Re:System Recomendation

[Surak]

Gigabyte makes a nice 400 FSB board, the G7VAXP has on board 10/100 LAN and awesome audio with SPDIF support. Abit makes a similar board with gigabyte ethernet, I believe.

Personally, I never buy a new chip when it first comes out. I always stay behind a generation or too, that's the sweet spot in terms of bang/$.

Re:System Recomendation

[Quila]

nVidia has an nForce 2 chipset for this, including GeForce 4 MX graphics, LAN, USB, Firewire, 5.1 audio, etc.

Re:System Recomendation

[ilsie]

Um, no. The Gigabyte G7VAXP may have a VIA KT400 northbridge on it, but it can not actually do 400 FSB with sync memory.

The only motherboard currently on the market that officially (as in, official support) does 400 FSB is the Asus A7N8X rev 2.0, with Nvidia's NForce2 chipset.

Re:System Recomendation

[Surak]

Thanks for the info. I didn't think any motherboards 'officially' supported it, but then I don't read Tom's Hardware as often as I should.

Re:System Recomendation

[afidel]

I'd get either an NForce 2 based mobo with non-integrated graphics or an Sis 746 based board like the L7S7A2 ($55 at newegg, a steal) and add a Geforce 4 Ti4200 128MB, total with 512MB DDR333 ram, an Athlon 2100+, cheap CDRW, a decent case and a mid sized hdd should come out around $550 shipped. One hell of a PC that can run just about every game out there at more than acceptable speed. Sure its not nearly as fast as the top of the line box but its really speedy for the price.

Re:System Recomendation

[CodeHog]

Go to Overclockers for more information on mobos. IMNSHO stay far away from onboard graphics. It's just a pain in the arse if you want more graphic power later. You can get away with onboard sound if you don't want it for much other than listening to music while you work. If you do more than that, like recording or mucho gaming, spend the money to get a decent soundboard. Good luck and have fun!

Re:System Recomendation

[slaker]

To the best of my knowledge, the GA7VAX(P/P-Ultra) does not support a 200MHz FSB. The 1.1 rev of the PCB has a DIP switch setting for 166MHz, and the 1.2 rev (current AFAIK) has a setting for 100MHz or "other", basically switching to 133/166MHz operation. Gigabyte is not terribly overclocker friendly, although the 1.1 rev of that board does just fine with a Tbred "B" at 166.

My two knocks on that board are the northbridge fan, which is a piece of crap that really doesn't need to be there (the GA-7VRX didn't have one), and the onboard sound, which loses its ability to handle line-in when you enable the SPDIF.

Also, the KT400A will be available Real Soon Now(tm). It'll offer better sound hardware than the ACL650 on the that gigabyte board and according to most of the stuff I've read, a nice bump in performance, too, particularly in memory bandwidth.

If you need something decent right now, and stability is a top concern, the GA-7VAX is a great choice. For budget users I usually suggest the current Shuttle offering, the AK32A. Performance junkies get nforce2s, but there's enough variation among those to need a personal choice.

The sweet spot CPU right now is probably the XP2000+. I'd pair that with a Speeze (Spire)5F263B1M3 80mm (and damn near silent) HSF and, if it were my machine, flip it to 166MHz operation. That'd get you a "2600+" for about $60.

Re:System Recomendation

[Hoser McMoose]

If you're reading Tom's Hardware once every few months (for a quick laugh), then you're reading it at least as much as you should! :>

On the other hand, if you haven't already done so, you should read Tom'z Hardware Bible!

Re:System Recomendation

[Hoser McMoose]

Well, as is often the case, what you should buy depends largely on what your planning on doing with the system. Your best bet is to start by figuring out what applications you spend the most time using and/or those applications that are running the slowest now and you want to run a bit speedier.

As a general system though, I'd start with either an AMD AthlonXP 2600 or 2800 (166/333MHz DDR bus speed), or a P4 2.4C or 2.6C (200/800MHz DDR bus speed) if Intel ever gets around to actually shipping those chips. The performance difference between the two chips is relatively small, and reall you're unlikely to notice the difference, though depending on what applications you run most one or the other may be somewhat faster.

For motherboard chipsets, with the Intel it's only a question of the i865 or i875, with the latter being a bit faster but somewhat more expensive. For the AMD chips, I'd definitely go for a board based around the nVidia nForce2 chipset. While VIA and SiS also make chipsets that would work for either of these systems, I've found the level of driver support to be better on Intel chipsets and better still on nVidia chipsets.

As for video cards, that depends again on what you're going to be doing with the system. The integrated video in the nForce2 chipset is very good and more than sufficient for most people, even if you enjoy the odd game every now and then. However if gaming is the main reason for buying the system, then you'll want something better. Here it comes down to a choice of ATI and nVidia, with each offering a pretty wide range of cards at a range of price points. I tend to favour nVidia again here, again because of the quality of their drivers, though ATI has improved significantly in this regard in the past little while.

After you get past that stuff though, you get to the important things, like memory and hard drive. For most people, these are the two areas that are going to have the biggest effect on how fast the computer "feels" when you're actually using it. Given the price of memory, I'd figure on 512MB for the majority of users, while some power users might want to go for 1GB of memory.

Hard drives are a tough one to follow, but get yourself a FAST one if at all possible. If you're like me though, you'll also want a QUIET hard drive, which is why I'd probably get either one of the new Seagate or IBM drives. www.storagereview.com is a great site for info about hard drives, and they're recommendations are probably more up to date/better then any info I could offer.

Re:System Recomendation

[macrom]

Gigabyte's board is cool, but search Google for the 7VRXP. That board is nothing but defective and Gigabyte has basically ignored all of the problems. As someone who wasted money on one of these boards, I would say proceed with caution. Maybe they have reconciled their ills with the AXP. The whole Dual BIOS thing is cool, though.

And while gigabyte ethernet would be infinitely cool (until terabyte ethernet came out), you probably mean gigabit. :^)

Re:System Recomendation

[Surak]

And while gigabyte ethernet would be infinitely cool (until terabyte ethernet came out), you probably mean gigabit. :^)

Yes that was a typo. ;)

Finally!

[The-Bus]

Excellent, now I will finally be able to upgrade the chip on my homemade iLoo.

Re:Finally!

[Schezar]

" Excellent, now I will finally be able to upgrade the chip on my homemade iLoo."

I had a cow chip. I was happy with it before, but I don't think it will cut it anymore when Doom ]|[ hits the shelves.

I was thinking of going with the new Intel "Dungbeetle" core, but I've heard it has heat issues.

Re:Finally!

[m1chael]

and the toilet seat is your heatsink.

Re:Finally!

[Repugnant_Shit]

Wooh! Hot seat!

Re:Overkill

[AvitarX]

you don't need 3 GHz

You need a 2.2 Ghz AMD

duh.

Tom's Hardware: Paper tiger

[geschild]

Tom's Hardware isn't so positive in their review. Quote from the conclusion:

"XP 2800+ would have been a more realistic label for the processor, which wouldn't have been a problem for anyone, if AMD still wants to go toe-to-toe with Intel's P4."

Oh well, the old adagium for benchmarks/statistics aplies I guess.

Re:Overkill

[Anonymous Coward]

Fact that you don't use any applications needing processor power such as 3D-rendering, encoding or mathematic software, doesn't mean no one use them.

Re:Overkill

[Pros_n_Cons]

Well, I'm just now reading about Gentoo and thinking about installing it, but my PC is too slow.(233) Id sure love a 3GHz to compile all that stuff. BTW if X is so bloated how come my 233, 64MB ram only uses next to noting on cpu cycles and about 15 megs of RAM, granted i use fluxbox but contrary to popular dis-information X is a damn well built peice of software.

And it is brown!

[leuk_he]

HEXUS reckons a 200MHz front side bus can't hurt. here. There's a picture of a brown semiconductor, also known as the "brains of a computer".

TOM'S HARDWARE has a controversial conclusion about the 3200+ and describes it as a "spineless paper tiger". It thinks the 3200+ is "much too aggressive" and it should be an XP2800+.


SUDHIAN Some crazy looking geek at Sudhian (hi Joel), reckons that AMD is being a little coy with clock speeds while its PR speeds have rocketed skywards.


FIRING SQUAD says AMD's odyssey for the performance crown has been a little more treacherous than Her Indoors, but welcomes the introduction of the 3200+ and the 400MHz bus.


TECH REPORT says there's not much new to report about the 2.2GHz chip apart from the fact that it runs on a 400MHz front side bus. But it reckons that the release is timely. There's a picture of a brown semiconductor which appears to resemble the brains of a computer.


LOST CIRCUITS contrasts the real brown brains of a computer with the hypothetical 3200+ brains of a computer it previewed a month or two ago.


BIT-TECH reckons that AMD's finally released the processor that the 3000+ should have been, denies the site's too pro-Intel, and puts it through its paces. There's a picture of a brown chip which appears to be the brains of a computer.

I stop whoring now, more to be found at amdzone

Re:And it is brown!

[Anonymous Coward]

maybe you should consider linking to the page you took it from... The Inquirer

Re:And it is brown!

[leuk_he]

And the inquirer took ik from amdzone, at least i linked that one.

Re:Overkill

[Lumpy]

well if you want comparisons....

we just bought some Dell 2.2ghz P-4 laptops here to replace some aging and damaged laptops.. what they are replacing are P-III 800 laptops.

All of them run W2K and the users are NOT feeling a speed difference. Yes some of the processor intensive apps are fast. the winstone tests show it's faster.. but word processing and internet does not get a speed increase.

So in conclusion of my findings I also reccomend to EVERYONE to not upgrade their computer unless they absolutely have to. If you own a P-III that is 800mhz or higher, you will not see any difference unless you are a power user or a gamer.

It's just silly to spend money for the sake of spending it. as soon as we get a magnitude of speed change that will be very noticable (read that as SCSI like hard drive speeds... IDE is too damn slow) it is a waste of time and money to upgrade like we did 2 years ago and earlier. there are no real performance and quality changes (except for downgrade in quality)

And they are only doing it to themselves...

[somethingwicked]

There was always this "Well, the AMD doesn't ACTUALLY run that fast, they just label it that way" thing going on.

HOWEVER...

Used to be, when you put an Intel head to head with the Equally-Labeled AMD, normally the AMD was up to or faster than the Intel chip.

So, no problem. If the electric car runs as fast as a 3.2 liter combustion engine, you won't get any protest from me in labeling it as a "3.2 engine"

But they are screwing up here. When you start over selling things, you make it difficult to trust you are being upfront all around.

And when you already have one * built in to you processor name, you don't want to call attention to it.

Before, I have had no problem feeling that I was getting a far comparison from AMD (my current processor is an AMD)

But I now I will be hesitant at buying their hype. Is this chip really as fast as the Intel?

Re:And they are only doing it to themselves...

[geschild]

In answer to your question: the benchmarks say it isn't as fast as the Intel part they've put it up against so I'd have to agree with you.

The surprise is in the price AMD is asking for this part. It's much lower than anticipated by the market. This could mean that AMD is well aware of the fact that this processor isn't up to snuff.

The question remains why they would want to shoot themselves in the foot so badly then and 'over-lable' the part. Tom's HW is legitimatly (imo) asking AMD to relable the part but I don't think that will happen.

I guess the marketing department has had something (too much?) to do with it... ;-)

Re:Overkill

[angle_slam]

I'm not just saying this because most of us are out of work right now and barely getting by financially, but I honestly don't see why someone would need a three Gigahertz processor to run today's popular software.

An unemployment rate of 6% means "most of us" are out of work?

Re:Overkill

[Anonymous Coward]

high detail 3D visualization

Care to support that?

Realtime physics

[mmol_6453]

One game: Software rendered Doom III. :)

Well, that may not be practical, but consider this: Game interactivity could gain massive boost of any sort of fluid model was used; Watch all the windows in a building get blown out when some idiot throws a grenade. Another example would be mass-spring systems for solids. Ever want to take out a load-bearing wall in a building map? Or cause a cave-in?

Re:Realtime physics

[EpsCylonB]

Actually the area of gaming that could benefit most from faster processors is AI. The AI in games these days is more of a hack than a serious atempt to replicate human behaviour. Serious AI simply takes up too much processor time. Beacuse AI is so difficult it is usually given whatever spare power is left after everything else in the game is run.

Re:Realtime physics

[Anonymous Coward]

Take a look at the paper "Graphical Modeling and Animation of Brittle Fracture". This technique is pretty much state-of-the-art, and runs at several thousand times slower than real-time. CPUs are going to have to get a lot faster for real-time physics, although the GPU companies will probably set their sights on hardware accelerated ODE solvers.

Multimedia apps need lots of CPU power

[MtViewGuy]

I think people are starting to find out that multimedia applications such as still-image processing, audio editing and video editing does require serious amounts of CPU power if you want anything done reasonably quickly.

Take for example Adobe Photoshop. The Photoshop LE edition that comes with some software CD-ROM discs included with your new digital camera may not have all the doodads of the full version of the program, but it still uses a lot of CPU power to do things like creating special effects for your pictures or to correct things like removing red eye, removing power lines, sharpening the clarity of background objects, etc.

Video editing is another program that really uses a lot of CPU power. After downloading your home videos from your MiniDV and MicroDV cameras, the editing process is quite complex and takes a lot of CPU power to create a final edited home video that you can burn onto a recordable CD or DVD disc.

Re:Overkill

[Micro$will]

I agree with you totally, especially with laptops, but have you looked into other factors, like some of the useless junk OEMs like to put in the disk images? The first thing I do with a new computer is fdisk/format/reinstall and eliminate all the cruft.

Another issue is all the power saving features on Intel *-m processors tends to lag it down a lot. Just uncompressing the Linux kernel at boot takes a lot longer on my PIII-m 933 Thinkpad than it does on my old PIII 600E home machine.

Re:Overkill

[mmol_6453]

Considering the tech sector was the hardest hit? And tech people without jobs are more likely to spend time on Slashdot?

Re:Overkill

[afidel]

Yeah SolidWorks or any of the other CAD programs my engineers run.

IDE slow?

[mmol_6453]

I think you're mistaking peak throughput with average throughput. Let me know the first time you read enough physically consecutive data to reach the peak throughput of IDE or any SCSI variant.

Solid-state mass storage would all but eliminate latency, but it's just not cheap enough for consumer use yet.

From the article...

[fok]

The AMD Athlon(TM) XP processor with performance-enhancing cache memory features 64K instruction and 64K data caches for a total of 128K L1 cache and 512K of integrated, on-chip L2 cache for a total of 640K of full-speed, on-chip cache.

So... will I be able to run MS-DOS programs directly from the processor cache? ;D

Re:From the article...

[makapuf]

yup, 640k ought to be ... *whack*

Re:From the article...

[Cam Wheeler]

Yes, you will never require more processor cache than 640k.

Re:From the article...

[nusuth]

You almost can. To completely map the DOS era PC 1Mb should be enough for everyone.

The BS known as benchmarking

[grondu]

For a couple of looks at benchmarking, read these two articles:

Did AMD's Athlon XP 3000+ earn its rating?

AMD XP 3200+ benchmarks disable Intel hyperthreading

Mesa!

[mmol_6453]

Software-mode Mesa, baby! :P

Releasing another speedbump?

[someguy]

Wouldn't having a processor referred to as a speed bump be a 'bad thing'?

Re:Overkill

[Anonymous Coward]

I have a corperate image that we install on all new machines thus blowing away all the crap that the computer companies force on their unsuspecting users.

so yes, I have done that... but laptops have horribly slow hard drives. and it is really slowing the machines down so that the apparent speed to the users is the same as the older machines. and from what they used them for... it is a waste to upgrade them.

Re:Overkill

[10Ghz]

So.... you are saying that over 50% of tech-people are unemployed?

Double your RAM instead

[Neil Watson]

Often you'll get a better performance boost out of doubling your RAM than getting a bigger CPU or other hardware. The cost is a lot less aswell so you can buy your self some beer to celebrate your cleverness.

The source of the problem

[mmol_6453]

AMD released a marketing schedule back just before they changed their labeling system. AFAIK, they haven't had any slips in their release dates. They may just be sacrificing performance for market presence, which is a Really Good Idea.

If you lose market presence, you lose the game. Permanently. The only way back in is to come up with something so revolutionary, you'd attract VC in this economy.

Biased benchmarks

[edxwelch]

I've just been reading this: http://www.theinquirer.net/?article=9445 Very interesting claims about the validity of certain benchmarks. According to the inquirer the PC World bencharks are the only ones to be trusted. Also quite interesting what they are saying about how consumers have been ripped off for buying Celerons.

Tunnel Vision???

[Anonymous Coward]

This guy has a serious case of it!

6% unemployment != "most of us are out of work right now"

Maybe it's higher in some tech sectors, but many of us are doing quite well.

As to why somebody needs faster processors ... some of us do more with our hardware then play solitare. I have a 2.6GHz Xeon Beowulf, and my model still takes a couple days to run. (GA based optimization of astrodynamics problems)

Re:Tunnel Vision???

[linuxelf]

Actually, with 6% unemployment, that's 94% employment, so, yeah, most would be at work, no?

Re:Tunnel Vision???

[Anonymous Coward]

I have a 2.6GHz Xeon Beowulf

Imagine a Beowulf cluster of those!

ENOUGH WITH THE CAR ANALOGIES!!!

[Anonymous Coward]

Everytime there is an AMD vs Intel debate the inevitable car analogy is made. There is absolutely no parallel between car engine performance and CPUs.

Wow

[Wordsmith]

"This CPU runs with a 400MHz Front Side Bus at 2.2GHz ..."

Now -that's- overclocking.

Re:Overkill

[mmol_6453]

No, just that there will be a disproportionately larger number of tech people to boring people showing up here on Slashdot.

Put another way, people on Slashdot are more likely to be tech people, who are more likely to be unemployed than people of other fields.

Granted, that doesn't justify his use of the phrase "most of us," but it does give you an idea of part of the formula that could be used to reach that conclusion.

gratis feature?

[MoreDruid]

will the seat be heated to the point that you can't sit down anymore? Or will you use "flush cooling"?
Inquiring minds want to know :)

AI

[mmol_6453]

I agree that AI is somewhat neglected. However, anything more automated than state machines and fuzzy logic is notoriously difficult to write properly.

If you have the (human) time, use a state machine. You can put all the intelligence and complexity of human behaviour you want into it.

Re:Overkill

[Anonymous Coward]

vs windows nt 4, which would be wiping the remains of X all over the street.

MOD Parent up (+5 funny)

[Anonymous Coward]

nt

Valid comparisons

[shdragon]

I'm sorry but I'm calling bullshit. I am sick and tired of some marketing dept. deciding that today they're going to re-write what commonly accepted words/comparisons mean.

It's called deceptive advertising.

Plain & simple. Intel makes processors. Their competitor is AMD. A reasonable person would expect that as they are competitiors any "marketing friendly" term used to describe speed is comparing apples to apples. The layman is not a computing expert. It is reasonable for them to assume that a AMD XP 3200+ would be comparable to an Intel 3.2 GHz machine. This is clearly not the case. I would encourage AMD to re-consider it's naming scheme if it is not offering a valid comparison. Anything less brings squalor to their company.

And for the record, I am a HUGE AMD fanboy. All of my PC's run Athlon chips and I only build systems with AMD chips as the cost/performance has been there. If AMD is unable to keep it's worth, I have no qualms with switching sides.

Re:Not even close to reality

[Trespass]

Maybe the problem is your vendor. Then again, you've thrown around a lot of (for a technical site) vague accusations so the problem may lie elsewhere again.

Re: Paper tiger

[hendridm]

You know, I can't help but think AMD is shooting themselves in the toe by not making the labeling numbers not as close to their P4 counterparts as possible. If they were truely representative, you might hear people purchase them based on their marketing label, ie "Give me a 2800+" for a system that is to be about as fast as a P4 2.8GHz processor.

Since the number don't really seem to match their true performance, people will discount them as innacurate and simple ask, "How many GHz is it?" (ie, "Should I buy the 2.2GHz Athlon or the 2.8GHz Intel?")

I am not a big fan of AMD processors, but I value the competition and it's a shame the MHz myth cannot be shed so easily.

No, what it REALLY means is...

[Anonymous Coward]

... that they're LYING about it's true speed by a full GIGAHERTZ now, calling it a "3200" and misleading the public, when it only runs at 2200MHz.

Re:No, what it REALLY means is...

[Anonymous Coward]

no, they're calling it "xp 3200+", and they're also telling you that it's running at 2.2GHz.
if the average lamer is too stupid to understand that the model name does not equal speed, don't bother to understand what they're buying, and that the clock speed alone won't tell you shit about real performance, then it's their fucking problem.
you would do a little research if you were to buy, say, a car, wouldn't you?

does the Porsche 911 have 911 hp?

Re:No, what it REALLY means is...

[gl4ss]

" ... that they're LYING about it's true speed by a full GIGAHERTZ now, calling it a "3200" and misleading the public, when it only runs at 2200MHz."

you do realise that all the other manufacturers are doing essentially same, and even worse? the intels mhz marketing is essentially even bigger lie than the pr-system of amd (which, IIRC, is speed compared to a tbird cored athlon).

does 2200mhz celeron compare straight to 2200mhz p4 not to mention 2200mhz barton? no. the celerons 'speed rating' (plain mhz) is the biggest 'lie' of the bunch, castrated cpu running seemingly high mhzs.

knowing just the mhz doesn't help you at all when comparing the performances.. not that it ever did. like plain cubic inches don't translate automatically to performance..

Re:No, what it REALLY means is...

[Anonymous Coward]

if the average lamer is too stupid to understand

So, deception is OK if you're just deceiving "the average lamer." Thanks for clarifying that.

does the Porsche 911 have 911 hp?

No, but nearly every car that preceded it didn't use the horsepower in its model number. With processors and clock speed, it has been (and with first-tier vendors like Intel, still is) the case that clock speed in MHz has been specified as part of the model number (e.g. Pentium 100 as opposed to Cyrix 5x86 PR100.

~~~

Re: Nope.... Inclusion

[uarch]

Everything that is present in each of the L1 caches is also present in the L2 cache.

Then again, as long as you run a "small" program you could run it all from 512Kb of cache with DOS. ;)

Other alternatives than just doubling your memory

[aksansai]

This is an old philosophy that I'm sure is running out of steam.

Myth: Doubling your RAM leads to significant benefits.

Fact: RAM has been cheap for quite a long time. In fact, it is not uncommon to find machines with 512M or more of system memory - not only that, but extremely fast memory as well. It would not be surprising to find that most individuals have at least 256M of RAM. And for 90% of the regular consumer - 256M is more than sufficient (as of current).

In yester-years - most people struggled with large bloated applications with very little memory. We all remember the days of 8M to 32M (wow!) and 64M to 128M (big wow!). Windows 95 did one thing for the industry that was absolutely positive: it forced people to stop struggling along with 16M or 32M of RAM and move up to the big top - 64 to 128M. (Unfortunately, Microsoft never stopped - 256M seems to be a nice sweet spot for Windows 2000 - will we be seeing a 1G sweet spot anytime soon?)

But this is a new era of computing: software still runs well on hardware that is a few years old. Prices for computer components have drastically fallen in the last decade. Components are now extremely modular and flexible - and can accomodate new and improved hardware. You can still find the ubiquitous i440BX chipset (introduced in the early P2 era) hosting P3-500+ processors.

Since people have been able to stock their computers with more memory (more than enough for mundane tasks; good enough for large behemoth games), RAM doubling may not help that much. Sure, an OS's page cache can always benefit from "wasted" memory by loading everything under the sun into memory. I have three computers that I regularly use: one with 768M, one with 384M, and one with... 32M (a paltry laptop).

Granted, the laptop would see a huge performance benefit with another 64M module. But the 384M machine always has free memory, and the 768M workstation (although loaded down pretty well) will manages to avoid swap without a problem.

My 768M stocked workstation is what I call ideal for me: a 10000RPM 80G hard drive, P4-2.4GHz, and with an ATI Radeon 7500. It's for browsing, compiling, and editing - no gaming.

My 384M stocked workstation is what I call the bare minimum: 5 hard drives (for a total of 145G), a dual P2-400MHz, and with an nVidia GeForce 2. It's for moderate gaming, browsing, and compiling.

My 32M stocked laptop is what I call "Jurassic": 1 hard drive (for a total of 1.6G), a P-166MHz/MMX, and a I-don't-know-what-the-hell-it-is video chipset. It's for board games, card games, and light browsing.

I'll focus on my 384M stocked box. I could get rid of the 5 drives and invest in 2 80G or 120G 7200/10000RPM drives. Hard drives are the biggest bottleneck in a system. The higher the density and the faster the spindle speed is a theoretical higher output from the drive. Other factors weigh in considerably - but this is a mere example!

I could also upgrade my processors to the maximum limit of the motherboard. If I flash the BIOS, I'm supposedly able to move to dual P3-800M. According to Pricewatch, I can get two of these bad boys at a small price of $156 (total).

Since gaming is not really a big focus for me, I can deal with this aged video board. But ideally, I would probably go with a relatively cheap dual-Athlon board, get a couple Duron 1.3GHz processors, move my 384M RAM over and be done with it.

Re: Paper tiger

[geschild]

I am not a big fan of AMD processors, but I value the competition and it's a shame the MHz myth cannot be shed so easily.

And there you hit the nail on the head. The MHz myth is what hurting the cpu market more than anything else. This myth, much to the detriment of PPC, Sparc, hell even the venerable alpha keeps on having a bigger impact than anything on competition.

The only solution seems to be to find a better way to compare apples and oranges in the processor world. Unfortunatly, great minds have tried and failed so unless somebody comes up with a 'eureka' moment it'll be bussiness as usual for the foreseeable future.

Disclaimer: I happen to be a big fan of AMD, but that's more to do with the fact that they seem to screw up less than performance.

I guess no analogy is simple enough for you...

[fmaxwell]

There is absolutely no parallel between car engine performance and CPUs.

Yes, there is and the analogy that the original poster made is a good one. A high-performance engine design can have greater horsepower with a lower displacement. Much as an AMD Athlon can have greater computing horsepower with a lower clock speed.

Re: Nope.... Inclusion

[lars_stefan_axelsson]

Everything that is present in each of the L1 caches is also present in the L2 cache.

Not on the AMD, which has an exclusive cache.

So it's the MS-DOS and application executing completely on chip. Someone post benchmarks please. ;-)

Re:Overkill

[kardar]

I upgraded a 750mhz Duron to a XP 1700+ ; same everything, just a different jumper for the front side bus.

I could tell the difference, but mainly in things like how fast it does a seti work unit; with browsing and word processing, I agree, it's a little more difficult to tell. Big software applications open up a little faster, things compile faster, but web surfing and word processing, yes, it is hard to notice an obvious difference. It's nice to have a little extra power if you need it, though.

Re:Overkill

[Graemee]

Laptops seem to have the slowest HDs of any PC. 5400rpm is a fast HD on most laptops. Combine this with speedstep they can be slower than a desktop of lesser CPU speed. My 933 PIII was faster than the 1.2G PIII laptop. Speedbrake(step) on the P4 is worse it runs the PC at a really slow Mhz unless it decides that you need more and steps up. No cycles for you.

sweet spot CPU

[raygundan]

I would highly recommend an XP2500+ rather than the XP2000+. $125 gets you a Barton core with twice the cache, and a 333MHz FSB rather than the 266 the 2000+ has. It's the only Barton that's not up at the high-price end of the AMD scale.

Of course, that's the route I'd take. I'm a bit hesitant to overclock to the degree the parent poster has.

Re:sweet spot CPU

[slaker]

An overclocked Tbred is actually a bit cooler than a barton at default FSB. The barton core, at low speeds, is really a fairly uncompelling purchase.

A $45 Tbred 1700+ can VERY easily slide in to barton territory if it's both unlocked and on a 166MHz FSB. I've even managed to get them up over 2 -actual- GHz (again, 2600-equivalent).

Will someone *please* tell me

[Anonymous Coward]

why they don't use MIPS/bogomips for chip benchmarks???

HotHardware is too sloppy to be quoted on /.

[dasboy]

I realize that this is an aside, but, HotHardware's article is a mess and should never been cited here. The raw Sandra benchmarks are all screwed up (the P4 3.0Mhz MM B/M has an AMD memory B/M in its stead; the numbers in the CPU section for P4 3.0/200 vs. P4 3.06/133 look wrong unless one has HT turned off). This is just sloppy. Go to some of the other hardware sites that have been posted above.

Don't sounds like a big fanboy to me

[Steveftoth]

if you're willing to switch sides if intel had the better price/performance ratio and faster chips overall. A real fanboy would stick with AMD chips even if they got half the frames on Doom 3 and blamed it on Carmack optimizing for intel instead of 3-d now! ;)

I agree with you but real fanboys stick with their horse long after it has lost the race. Just look at anyone in the Apple camp who says the Altivec is the bomb.

Re:Don't sounds like a big fanboy to me

[shdragon]

Haha.. point taken. I guess I'm a fanboy so long as I don't end up looking stupid because I'm wrong. :b

Re:Overkill

[deadgoon42]

I agree totally. I work for a computer manufacturer and people constantly ask me for recommendations on new computers. The first thing I want to know is what they will be using the computer for. If it is word processing and internet surfing is what they are going to use it for, I recommend a used P-II or P-III machine or a K6-2 500. My max would be a 1 GHz t-bird (the chip is about $30 now). I alway recommend that people spend more money on memory, storage, and other features they think they might need and not on processor power that is just overkill. And if you're just using it for light web surfing, watching television, or checking email, I have a Powermac 5400 I can sell you.

Re:Overkill

[Anonymous Coward]

All of them run W2K and the users are NOT feeling a speed difference

Well that sounds like a pretty unoffical benchmark. I upgraded from an Athlon 1300 to a 2100 and I couldn't feel much of a difference either. I did notice a big increase in encoding ogg vorbis files. You should never upgrade a computer unless you really need to anyway. I mean if you're happy with what you got then why screw with it?

As an IT Manager, I'm typically going with Athlon 2200s in new computers. CPU's have been "fast enough" since they came out with the 1400 IMHO. And AMD has the best price/performance ratio.

Well...

[gilesjuk]

In my original post I was referring more to the markings on cars.

People do look at engine sizes when really they need to be looking at power to weight ratio.

Re:Well...

[fmaxwell]

I was referring to the poster who disagreed with what you wrote. I think that your analogy was fine.

Bajillion Ultra Hertz!

[nick_davison]

So, the logic behind AMD's naming structure was that, while their cores ran slower, their names gave you an idea of what speed a comparable Intel chip would run at. That meant a 2000+, while running significantly slower, produced about the same output as a 2GHz Pentium, etc.

Now they have the 3200+ which runs... well a bit slower than the 3Ghz Pentium. It also runs at a 2.2Ghz core. So nowhere in any of its design does it really justify a 3200+ moniker. 3000+ is closer with 2950+ possibly being the most accurate, according to the benchmarks.

If the n+ meaning has nothing to do with real world speeds, core speeds, relative speeds or any other kind of speed anymore, why don't they stop worry and just get on with calling it the "Bajillion Ultra Hertz!" model (note the important exclamation marks)?

You know, I need to get myself a V8+ sticker for a Pinto.

Re:Bajillion Ultra Hertz!

[ratfynk]

As with Intel ignore the fluff. Consumer electronics have become an ongoing junkware war! The advertising of the specs for processors is about as reliable as an infomercial. Anybody who pays attention to the fluff and does not look at tech deserves what they get. 3GHz is meaningless to the average computer user they just think they gota have it. In this respect Intel is winning the minds of the masses.
Give AMD credit for producing anything at all when monopoly has become the norm! I do not think AMD is an Intel wantabee I just think they wantabee in buisiness.

Re:Bajillion Ultra Hertz!

[Anonymous Coward]

The AMD performance numbers don't compare to Intel speeds. They're supposed to show how fast the chip is if it was a Thunderbird running at that clock. So 3200+ is just TBird @ 3200. That the difference re Intel speed/performance is so visible now and is pissing people off only shows that those people didn't know what the numbers meant in the first place.

Re:Bajillion Ultra Hertz!

[Anonymous Coward]

Your argument is based on the incorrect premise

" So, the logic behind AMD's naming structure was that, while their cores ran slower, their names gave you an idea of what speed a comparable Intel chip would run at"

AMD's naming structure is based on a comparison to a T-bird Athlon, and is calculated using third-party written and audited benchmarks. AMD have a FAQ entry for this.

Re:Overkill

[ekilfeather]

>I honestly don't see why someone would need a three Gigahertz processor to run today's popular software.
>... like VisiCalc back in '79

Dammit he's right Doom 4 will NEVER be as popular as VisiCalc

If you're gonna use car analogy...

[splerdu]

...at least get it right!

MHz = operational speed, aka RPM. Not CC
Since Intel's P4 is does less work per cycle, it's like a small displacement engine working at high rpm. AMD's Barton is like a large engine working at lazier, lower revs.

Re:If you're gonna use car analogy...

[gilesjuk]

Except that cars are marked with their engine CC not their rev limit.

Re:Overkill

[Anonymous Coward]

I got 1.8Ghz and I'm happy.

Re:Overkill

[luisdom]

In fact I allways felt "comforted" by this when my hardware was getting older. You know, I'm not a scientific, nor a die hard gamer...
But then I started doing my master thesis and god, how I miss having a 2GHz+ brand new machine... when the f***ing "ETA" says 5h to go, and the oracle thingie sucks 350Mb of RAM and the hard disk sounds like a car starting and the deadline is lurking me, like a gollum to a ring... ... and then I think on all the kiddies with their 2000$ new machines for word and GTA3 and suddenly robbery doesn't seems so bad.
Af, 11 hours working, shouldn't be posting nonsense
:wq

Is it just me...

[oaf357]

Does Intel seem to be losing the number game (big numbers in front of MHz or GHz = more sales)? Or is AMD beating Intel at their own game?

You do need new hardware in some cases.

[MtViewGuy]

There are a couple of instances where new hardware is a good idea:

1. Multimedia processing. Image-editing programs and video-editing programs are MAJOR hogs of CPU power, and you definitely want a decently fast CPU to edit pictures downloaded from your digital still camera or videos downloaded from your MiniDV/MicroDV camcorder. In my opinion, you probably want at least a Pentium 4 2.0 GHz (Northwood-core version) or an Athlon XP 1900+ CPU if you want to do multimedia processing decently fast.

2. The latest games. Let's face it: by the end of 2003 we'll be seeing plethora of games that will require the use of DirectX 9.0/9.0a features. You want to get a graphics card that supports DX9 features in hardware; cards that use ATI's R300/R350 GPU and nVidia's NV30/NV35 GPU do this and will allow the latest games to run smoothly even on highly-complex backgrounds (e.g., Doom III).

Yin and Yang

[Esion Modnar]

OK, I just could not let this go.

How exactly is AMD "yin" to Intel's "yang" ?

Yin and yang can mean a variety of things, for instance,

Yang:
light
male/masculine
aggressive
forceful
powerful

Yin:
dark/shadowed
female/feminine
passive
yielding

So, in light of this definition, how would AMD fit as yin anything? The popular notion of yin and yang is simply that of opposites, which loosely gets applied to any competitors, which is incorrect.

Re:Yin and Yang

[Anonymous Coward]

Actually yin and yang are supposed to represent balance. Whether or not this is true for this situation ...

What about Price?

[jonr]

How much does the XP 3200 and the P4 3000 cost?

Re:What about Price?

[ozzmosis]

Currentlly from pricewatch.com the lowest price of the Athlon 3200+ is $466 while the P4 3.06GHz is $368, but the Athlon 3000+ is only $310.

Re:What about Price?

[Anonymous Coward]

Intel P4 3.0 costs $417 US. AMD 3200+ costs $464 US.

Re:Overkill

[faraway]

Did you expect word or the internet to be faster if you upgraded your processor? I think you need to upgrade the user and the upstream connection, not the processor for that.

Re: Paper tiger

[be-fan]

You know what would help PPC and SPARC even more than getting rid of the MHz Myth? Actually releasing processors that outperformed highly-clocked P4's in real benchmarks. MHz myth is one thing, but if your processors aren't keeping up in SPEC, that's something else.

PS> I'm not comparing a SPARC *machine* to a P4, you Sun fanboys! High end SPARC's have systems architectures that blow away most PC's. But the CPU itself is a rather weak performer.

Amd's FRONTSIDE BUS sucks !!!! God AMD !

[zymano]

Where is your leadership. Tom's Hardware is taking you to task ! You guys should have quaded that bus up the same way intel does.

AMD is going downhill according to this article at TOMS.

Er, very wrong.

[Anonymous Coward]

There really should be a "-1 wrong" moderation.

There's no way an Athlon XP 1533MHz (model number 1800+) is supposed to be 30% faster than an Athlon 1400... a simple look at the benchmarks shows that's not the case. The whole reason they introducted the model number is the Thunderbird 1400 kicked a Pentium 4 1700's ass completely.

AMD, of course, doesn't officially say what the model numbers mean other than some vague stuff about "relative performance of Athlon *XP* processors" so they won't get sued. But the model numbers are obviously way closer the Pentium 4 MHz speeds than to the relative performance vs. THunderbird - if they tried to claim a 1533MHz Athlon XP 1800+ was supposed to be as fast as a 1800MHz Thunderbird, they would've gotten sued as a simple overclocking of the Thunderbird would show how wrong they are.

Sorry.

[Anonymous Coward]

Oops, I accidentially mismoderated your post down. Hopefully this will back out my moderation.

Re:Overkill

[Anonymous Coward]

no he didn't but the CEO who sits in his office and has an IQ of around 80 when it comes to anything but money does.

This is the delimma... rich idiots think that a newer faster computer fixes everything.... it diesnt.. and that is why we call the rich ... IDIOTS...

Just curious

[nusuth]

What about using a CVT with such an engine?

Re:Just curious

[fmaxwell]

What about using a CVT with such an engine?

You're welcome to try, but most CVTs have a very limited range of ratios. The problem is finding a CVT that can adjust ratios such that the moped could go anywhere from stopped to 30mph. It takes one heck of a ratio to let the CVT turn 15,000RPM when leaving the line and not exceed 16,000RPM at 30mph.

why this analogy sucks

[_avs_007]

This is a bad example in relation to CPU's.

GM's LS1 V8 is 5.7 litres, has 8 cylinders, one camshaft and 16 overhead valves.

Nissan's VQ engine is 3.5 litres, 6 cylinders, 4 camshafts and 24 valves.

The LS1 engine produces 345 HP, the VQ 260. For arguments sake, lets stick a supercharger on the VQ, and it should match output.

Are you saying the VQ will be a better solution, because it has less cubic inches, so it does more? Its all relative. Its smaller in terms of cubic inches only. In terms of physical size, the LS1 is actually smaller/lighter. You think those extra moving parts, like the 3 extra camshafts and 8 extra valves, etc etc have no mass or take up no space? (or the blower?) How much low-end torque is on that VQ again?

I remember reading an article a long time ago, stating that GM's 3800-Series II V6, had something like 100 less parts than Toyota's 3.0 Litre DOHC V6.

While the VQ engine is a really fine engine, (one of wards ten best), with a blower on it, there is really not much else you can do. Turning this motor into a reliable 600+hp engine will be difficult. Turning the LS1 V8 into a reliable 600hp engine is actually considerably easier.

These things may/may not be true of the Intel/AMD race, so this is why the car anology does not apply here. There is more to an engine than just cubic inches.

Example: my L67 gets better fuel economy AND blows the doors off the wife's 3.2 litre VTEC V6.

Re:why this analogy sucks

[fmaxwell]

How much low-end torque is on that VQ again?

Quite a bit with a blower.

In terms of physical size, the LS1 is actually smaller/lighter.

I read that the LS1 weighs 497 lbs. while the Nissan 3.5l VQ weighs 460 lbs.

You think those extra moving parts, like the 3 extra camshafts and 8 extra valves, etc etc have no mass or take up no space?

They aren't "extra". They are there to improve performance (no, performance is not measured solely by peak horsepower on a dyno).

There is more to an engine than just cubic inches.

And there's more to a CPU than just clock speed. That's the point of the analogy. You can't just measure one thing to determine performance potential. Just as you cannot just measure cubic inches to determine the performance of an engine, you can't just measure clock speed to determine the performance of a CPU.

Re:why this analogy sucks

[_avs_007]

it may have been the older VQ for the weight comparison. When I was talking about extra parts, I didn't mean to infer they didn't do anything. I meant more parts = more stuff to fail = more weight, etc etc.

Otherwise I agree with you, thats the whole point I was trying to make, and point out why this car analogy is worthless, because its so completey different.

Yes, numbers alone mean jack, but that's my point. It depends what you are looking for, in order to determine which engine is better for you. Just because VQ does more with less cubic inches does not mean its a better engine, just like just because AMD does more with less cycles doesn't mean its a better processor either.

I'm not saying they suck either, I'm just saying each has its place.

Yes

[jelle]

You probably missed that the FSB was increased by 20%.

100% compatible Linux Box (fast and cheap)

[MarcQuadra]

Get an EpOX ep-8kha+, it's only DDR266, but it's PLENTY fast for pretty much anything.

Put 1GB PC-2xxx or better RAM into it. I suggest Kingmax TinyBGA parts because they've never caused me ANY trouble. Only pay for PC3200 RAM if you intend to move it into your next machine. This board maxes out at PC2100, but 2700 and above work fine.

Then put a Barton core in it and UNDERCLOCK. The mobo won't do 333MHz, it will underclock your CPU for you a bit, but you'll generate MUCH less heat and you'll get the faster 512K barton cache. Your chip will run at 80% the 'native' clock speed if you build this system.

For a hard drive you should get something with ATA-100 or better and 8MB cache. I suggest a Maxtor MaxLine II Plus or a WD Special Edition. Don't get top-of-the-line capacity unless you like to overspend.

Get an ATI card from the RADEON family. Spend more if you game. Get a 32MB DDR Radeon 7500 if you don't game much, it's good enough to game occasionally but not at all pricey. ATI has native linux drivers in xfree86.

Get a SoundBlaster Live 5.1 card. They're cheap and the digital out is kewl if you are into hi-fi audio. Don't spend more than $50 on a sound card, please.

[OPTIONAL] Get an adaptec-2940u2w and an ultra2wide 9GB SCSI drive for your / partition if you want to run super-fast. Even though it looks slow on paper, you'll see a MAJOR improvement in launching apps and other intensive random disk activity. Put your /home on the IDE drive.

Follow this advice and you will get an incredibly fast machine for very little money. I've had a similar setup for two years and it still kicks butt, feels faster than my friends' 2.4Ghz PIV boxen. The board has onboard sound, but I prefer the SBLive's digital-out. Also, almost ALL onboard LAN chipsets are trash, put an Intel or 3com NIC into the thing for MUCH improved performance when pumping data over ethernet.

What about a generator/electric motor combo

[Anonymous Coward]

like a freight train? shouldn't that work or is it just too heavy or not effieicnt enough?

Re:What about a generator/electric motor combo

[fmaxwell]

Not even close to efficient enough. A good electric motor is about 50% efficient. That means that you would have to put in 4 hp worth of energy (about 1500 watts) to get 2hp of rotational energy out of it. Then there is the inefficiency of the generator.

Freight trains use diesel-electric power because they need incredibly high starting torque and the electric motor gives that. There's a saying about trains: A diesel (electric) locomotive can start a load that it can't pull and a steam locomotive can pull a load that it can't start. That's the difference between torque and horsepower.

Okay, I'll feed the troll here:

[Anonymous Coward]

The real "killer app" that the industry types keep crying about (the one that will push the need for faster CPU's and start the upgrade cycle all over again) is already here, but most people don't recognize it yet: multi-tasking.
I play Windows games in Linux with Winex, but the sound leaves a bit to be desired, so I like to play MP3 files and turn down the game sound. Combine this with burning an MP3 and IM and even alot of newer PC's start to be stressed. I think the reason this kind of multi-tasking hasn't taken off is that the older PC's and previous versions of MS Windows did a lousy job of it. As the hardware and mainstream software becomes capable of doing it, the demand will start to ramp up.

Re: Paper tiger

[geschild]

There's more to real world performance than SPEC. As usual there's also more to performance than a brand. And to top it all off, my comment was about making it easier to compare oranges and apples, which is exactly why I don't mind that you keep on treading the old path. You don't know any better.

You and I have no better way to offer so lets please hold our comments on what out-performs what to ourselfs since they are based on a fiction called a benchmark (as I pointed out before).

Re: Paper tiger

[be-fan]

We're not talking about system performance here. We're talking about processor performance. And for measuring processor performance, SPEC is a pretty good indicator.

Re: Paper tiger

[geschild]

If you want to believe in benchmarks, that is your right. I happen to think that synthetic benchmarking has the same value as using games as benchmarks. Actually, I think games are a fairer benchmark sinch they are real-world loads.

As long as there is no straightforward way of determining processor power (and I don't believe in SPEC) that everybody agrees on... though shit.