AMD K6 vs. P3

[alphaa10]

How to compare processor power between the AMD K6 @300mhz and Intel Pentium 3 @866mhz? Bpth the K6-2 and P3 have a 256k L2 cache.

Any rough guesses, that perhaps the P3 is about twice as fast as the K6? Are there any comparisons of processor power between the two CPUs at this site. (I have looked.)

 

[stephen1960]

How to compare processor power between the AMD K6 @300mhz and Intel Pentium 3 @866mhz? Bpth the K6-2 and P3 have a 256k L2 cache.

Any rough guesses, that perhaps the P3 is about twice as fast as the K6? Are there any comparisons of processor power between the two CPUs at this site. (I have looked.)

I have seen a Tom's Hardware Guide article that did show the performace of all, or nearly so, desktop processors in a historical scope, so I do think it exists but you will have to search for it in their archives.

The P3 might be compared to the K63 or perhaps the K62 but the P3 had a stronger floating point unit and was considered the superior processor, as I recall. And you should specify the models of the processors if you really wish an answer.

Here you go: http://www.tomshardware.com/2004/12/21/the_mother_of_all_cpu_charts_part_2/index.html

 

[jaguarskx]

I don't think the old AMD K6 had much performance compared to the P3. Back then I think, Intel was the company to stick with. That is until AMD unleashed the K7.

Try Google for a K6 review, I'm sure you find plenty of reviews.

 

[mr_fnord]

It really depends on the metric. The P3 might come close to 3x the performance in synthetic benchmarks and very processor intensive tasks like FAH, but in general system performance the P3 will be faster but not twice as fast. I had a K6-2 and bought a P3 667 and for general use they were fairly comparable, but the intel was slightly faster. I never had a 3d video card in the K6, and never did anything very intensive on it.

Intel designs generally gained more from cache, esp. the P4. I think that was even the case with the P3 too.

It doesn't really matter though, since both will do standard tasks like web browsing, Word, etc. and neither will play modern games or run intensive software.

 

[ms2005]

I had a K6-2 266 and P3 866, the copermine P3 was much faster, it's still running today while the K6 2 has been retired for years.

 

[gOJDO]

The P3 Coopermine 866MHz is in the worst sceanarioo at least 3 times faster than the K6 300MHz

 

[digitaldoc]

I used a 475 MHz K6-2 for years. It was roughly equivalent to the Pentium II family, and even half a notch behind that as I recall. The Pentium III architecture was definitely superior to any K6-2. I doubt you would want to use a K6-2 for anything in this day and age.

 

[eggsmckenzy]

The PIII is going to win, but that's not even a fair comparison, as the PIII has a 566Mhz advantage over it. Given the same clockspeeds, it would virtually be neck and neck between a K6 and Pentium III.

Now, the K6-III with its 256K on die full speed cache easily beat the Pentium III in most every task but ones that involved floating point, as the PIIIs FPU was pipelined, and the K6's wasn't.

 

[BustedSony]

The high-end P3s are still a capable chip. Any modern program (other than games) will run on them very well. The limit is that the associated chipsets only supported up to 512 Megs. I still use a P3-1000 (non-tualatin) daily as a Matrox video editor, though for major rendering jobs I put its Data Drive into the C2D, which renders more than ten times faster.

Any top-end K6s I've encountered on the other hand are just trash, they don't seem to run anyting acceptably. I've even "upgraded" a K6 to a P3 for a client without money.

My laptop is an IBM T23, 1.33 Ghz Tualatin. I just love that machine, seriously! It has a performance-heat-power consumption ratio to die for. That's why the Tualatin was one of the base designs for the Conroe.

 

[Crashman]

lol, K6-2 was a dog. It got a lot of good press, but was stomped on by a stock-speed Celeron in the most popular applications due to it's weak floating point. In fact, a PIII 450 was up to 2x the performance of a K63-450 in some 3D apps.

Here's the funniest part: A Via C3 1000 is a bit less powerfull than a K63-450 in similar applications where the K63 was already a failure, yet people think that the C3 is "efficient" where it's actually just weak.

Getting back to the K6-2 300, it's about 1/3 the sperformance of the PIII 866. That's an impressive win for the PIII, because processors do NOT scale perfectly with speed.

 

[evilr00t]

The P3 wins by WELL over 300%, it's not even fair:
K6 has outdated system bus-based cache, 66MHz FSB, and abysmal FPU
P3 has fast on-die cache, 133MHz FSB, and decent FPU

 

[qcmadness]

Pentium 3 should be much better than K6 series.

 

[NightlySputnik]

No contest even. K6-2 @ 300 mhz would have been crushed by any P3.

No offence, but it's like comparing an AthlonXP 2800+ with a Core2Duo E6600. The difference is probably alike.

AMD started dominating Intel (by more than 5%) on a frequency basis only when the P4 came out and it lasted until C2D replaced it. This is not to say that Intel never had performances crown from there until C2D came out. It's only that they had it when their cpu were at a much higher frequency, and never following the released of the A64. About P3 and K7, it was pretty much a tie, with a slight advantage to K7, but nothing to write home about.

P3 is superior to K6-2 with faster memory access and more bandwidth, better FPU (altough not quite as good as K7), SSE instructions that were already in widespread use, much faster cache access and bandwidth plus better out-of-order execution.

I have to say that at one point in my life I have been desperatly looking for a K6-2 on socket 7 to replace my the old Pentium MMX 233. Unfortunatly my mobo wasn't compatible

 

[Crashman]

Most incompatable motherboards worked fine. I set many Pentium MMX systems up with K6-2 400's by leaving the voltage at 2.8V and adding a larger cooler, and using the 2x multiplier (133MHz) which the higher-speed K6-2's interpreted as 6x. 400MHz from a system that didn't even know that 400MHz was possible.

 

[evilr00t]

Weren't the K6-2 series 2.2/2.4V? Sounds like you'd be a-toastin' those K6-2's at 2.8V

 

[Crashman]

Indeed, voltage range from mobiles to factory-overclocked trash was something like 1.80V to 2.25V, depending mostly on core quality. But all those processors lasted several years, probably due in part to the oversized coolers.

I had a cheap souce for the big black Pentium III (socket 370) coolers, and that's mostly what I used.

 

[DrBlofeld]

The PIII is going to win, but that's not even a fair comparison, as the PIII has a 566Mhz advantage over it. Given the same clockspeeds, it would virtually be neck and neck between a K6 and Pentium III.

Now, the K6-III with its 256K on die full speed cache easily beat the Pentium III in most every task but ones that involved floating point, as the PIIIs FPU was pipelined, and the K6's wasn't.

:lol: :lol: :lol: Wow, this guy completely missed the point and went down the AMD fan-boy super-highway.

 

[NightlySputnik]

Most incompatable motherboards worked fine. I set many Pentium MMX systems up with K6-2 400's by leaving the voltage at 2.8V and adding a larger cooler, and using the 2x multiplier (133MHz) which the higher-speed K6-2's interpreted as 6x. 400MHz from a system that didn't even know that 400MHz was possible.

That might be very possible. It's only that I wasn't even close to know all I do now, so I didn't want to take a chance. Anyway, that machine was replace with a P4 1.8 Northwood about 2 years later. A computer that then cost me more than 2000$ (P4 1.8, Radeon 8500 with 128 MB(!!!!!!!!!), 512 MB memory, a "whooping" 40GB HDD and a SB 5.1). And that was a "bomb" back then. I could play Ghost Recon at 1280*1024 with setting maxed out. Oh... was I proud. Of all that, only my case and the 40GB HDD are still in use by me. Everything elses was replaced.

I guess it shows purchasing a computer isn't an investment, but a spent

 

[NightlySputnik]

:lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol:

So true!

 

[m25]

How to compare processor power between the AMD K6 @300mhz and Intel Pentium 3 @866mhz? Bpth the K6-2 and P3 have a 256k L2 cache.

Any rough guesses, that perhaps the P3 is about twice as fast as the K6? Are there any comparisons of processor power between the two CPUs at this site. (I have looked.)

Hey, slow down, you're not comparing a K8 with a Prescott (even in this case a prescott with 2x the frequency of a K8 will soundly beat it). The answer is simply YOU'D BETTER NOT COMPARE THEM!
K6 vs P3 is much like P2 vs P3 and the P2 was total failure, just a brief broken step to the glorious P3. Up to near the end of the P3 era, Intel's and AMD's architectures were very near and the same was for clock per clock performance; the P3 held an edge on the multimedia due to SSE and Athlon's/Durons had better floating point.

 

[Crashman]

Hahah, do you remember when AMD got sued by a group of investors over the development cost of the Athlon? The investor group said they didn't need the Athlon, that it cost too much to develop and was a mismanagement of funds?

Wow, that K6 series was turds compared to PIII's, and even Celerons in some apps (except for the early cacheless Celerons). Had AMD NOT spent all that money on development, their technology might have progressed as far as VIA's Eden by now, but only in the extremely unlikely event that they'd survived this long

 

[gOJDO]

The PIII is going to win, but that's not even a fair comparison, as the PIII has a 566Mhz advantage over it. Given the same clockspeeds, it would virtually be neck and neck between a K6 and Pentium III.

Now, the K6-III with its 256K on die full speed cache easily beat the Pentium III in most every task but ones that involved floating point, as the PIIIs FPU was pipelined, and the K6's wasn't.

Please share with us, whatever you are on!

The K6 performs similar as same clocked Pentium MMX or Pentium Pro.
http://www.tomshardware.com/1997/04/06/intel/page3.html

The Pentium III Katmai wipes the floor with same clocked K6 in every benchmark known to mankind. Even AMD are claiming that the best K6-III with extra 512kB of L3 cache is underperforming same clocked P3-Katmai.
http://www.amd.com/gb-uk/0,,3715_13530_1260_1288^952,00.html

This bencmark is bottlenecked by the graphics card, so we can not conclude by how much the K6-III is outperfromed by P3-Katmai.
Let's check out the CPU charts by the THG.
http://www.tomshardware.com/2004/12/21/the_mother_of_all_cpu_charts_part_2/page19.html

For real-life apps, the K6 is outperformed by lower clocked Celeron:

The Celeron 500 is 125% faster than the K6-2+ 550 for Lame MP3 encoding.

For WinRar 3.4, the Celeron is 61% faster

For DivX 5.2 encoding, the Celeron is 70% faster

Can you realise how brutaly the best P3(Tualatin-S with 512kB of L2 cache, 256bit 8-way ATC) core wipes the floor with the best K6 at the same freqfency?

 

[dasickninja]

Why is it that AMD consistently has better floating point performance?

 

[m25]

Yeah, they say that history repeats itself, but CPU history is still young to do it, that's why it's still so exciting and thank God AMD is still alive to make it a race rather than a walk. VIA is just a fossil, a dinosaur and they'd better make some decent chipsets or they'll soon be only on tech history books.

 

[eggsmckenzy]

If you must know what I am smoking, then here: http://www.tomshardware.com/1999/02/23/new_cpus_from_amd_and_intel/page5.html

The rest of the review pretty much shows that in non-floating point tasks, the K6-3 is equal or faster than a Pentium III Katmai.

It's weird though. I read somewhere that the Pentium III was not to be the successor to the Pentium II, but the Athlon was trouncing everything so badly, they had to do it to save face.