VOGONS


P2/P3 VS K6-3+ - A Great Battle Commences

Topic actions

Reply 20 of 90, by d1stortion

User metadata
Rank Oldbie
Rank
Oldbie
F2bnp wrote:

Not really interested in emulators. I can do this on my main PC too 😁

You can most probably do any of the games you were talking about on your main PC as well... 😀

@GeorgeMan: Your problem might be the video plugin. Get a 3dfx card in there and use Lewpy's Glide plugin.

Not that I would use this on a regular basis (or emulators at all), but I do find it fairly amazing what 1999 high end was capable of.

Reply 21 of 90, by feipoa

User metadata
Rank l33t++
Rank
l33t++

If your ALI board has 512 KB of L2 cache, I believe the cacheable memory range is 512 MB.

For a fair comparison, you really should use the same hardware for both the slot 1 and super7 motherboards, otherwise the results will be open to scrutiny.

I suspect the results of the K6-III-550 will approximate the PII/III-450. The averaged FPU results for the K6-III-550 fall between that of the PIII-500 and PII-400. Regretibly, the PII/PIII-450 are the two chips not included in my comparison, though I now have these, recently donated, chips in my possession. When I get time... more like "if"...

Regardless of what/how you decide to test, it will certainly be nice to see a 3rd party comparison of these systems.

Plan your life wisely, you'll be dead before you know it.

Reply 22 of 90, by Forevermore

User metadata
Rank Member
Rank
Member
feipoa wrote:
If your ALI board has 512 KB of L2 cache, I believe the cacheable memory range is 512 MB. […]
Show full quote

If your ALI board has 512 KB of L2 cache, I believe the cacheable memory range is 512 MB.

For a fair comparison, you really should use the same hardware for both the slot 1 and super7 motherboards, otherwise the results will be open to scrutiny.

I suspect the results of the K6-III-550 will approximate the PII/III-450. The averaged FPU results for the K6-III-550 fall between that of the PIII-500 and PII-400. Regretibly, the PII/PIII-450 are the two chips not included in my comparison, though I now have these, recently donated, chips in my possession. When I get time... more like "if"...

Regardless of what/how you decide to test, it will certainly be nice to see a 3rd party comparison of these systems.

In terms of 3d games, does the K6-3 respond well with the right graphics hardware when compared to it's P2/P3 counterpart?

So many combinations to make, so few cases to put them in.

Reply 23 of 90, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

There are actually reviews of this CPU on the Internet 😀

Here is one article I could find:

http://www.anandtech.com/show/211/4

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 24 of 90, by feipoa

User metadata
Rank l33t++
Rank
l33t++

AnandTech’s tests illustrated a 4% increase in performance when using 3DNow! accelerated Quake 2 OpenGL drivers as opposed to using the standard OpenGL drivers under Quake 2 with the K6-3 test system’s Riva TNT video card.

Ouch, only 4%

Plan your life wisely, you'll be dead before you know it.

Reply 25 of 90, by GeorgeMan

User metadata
Rank Oldbie
Rank
Oldbie
d1stortion wrote:

@GeorgeMan: Your problem might be the video plugin. Get a 3dfx card in there and use Lewpy's Glide plugin.

Not that I would use this on a regular basis (or emulators at all), but I do find it fairly amazing what 1999 high end was capable of.

Tried it already, not impressed at all.
But it's impressive that one can play games of a console on a PC of the same era via emulation, at full speed and with good graphics! 😊
Stopping the offtopic right here. 😀

1. Athlon XP 3200+ | ASUS A7V600 | Radeon 9500 @ Pro | SB Audigy 2 ZS | 80GB IDE, 500GB SSD IDE2Sata, 2x1TB HDDs | Win 98SE, XP, Vista
2. Pentium MMX 266| Qdi Titanium IIIB | Hercules graphics & Amber monitor | 1 + 10GB HDDs | DOS 6.22, Win 3.1, 95C

Reply 26 of 90, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Some more:

http://www.anandtech.com/show/160/10

http://www.xbitlabs.com/articles/cpu/display/k6-iii.html

http://www.maximumpc.com/article/features/old … amd-k6-2_3d_cpu

http://www.tomshardware.com/reviews/600-mhz-s … et-7,210-6.html

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 27 of 90, by Darkman

User metadata
Rank Oldbie
Rank
Oldbie

very similar to the upgrade I did, went from a 450Mhz K6-III to a 450 Mhz PIII , both using a Voodoo 3 3000 , same sound card too (I did go from 128 to 256MB RAM , but really that wouldn't affect things much)

noticed a pretty big difference, for instance in 3Dmark2000 the score went up from 1486 to 2475, likewise most games showed a 10-20 fps increase, including strangely Quake 2.

Reply 28 of 90, by mwdmeyer

User metadata
Rank Oldbie
Rank
Oldbie
feipoa wrote:

AnandTech’s tests illustrated a 4% increase in performance when using 3DNow! accelerated Quake 2 OpenGL drivers as opposed to using the standard OpenGL drivers under Quake 2 with the K6-3 test system’s Riva TNT video card.

Ouch, only 4%

I believe the main issue was that the TNT drivers didn't really have any special 3DNow support built in. This is why 3dfx + K6 was recommended. I believe this was improved around the time of the TNT2, so the newer drivers show a larger improvement.

Vogons Wiki - http://vogonswiki.com

Reply 29 of 90, by swaaye

User metadata
Rank l33t++
Rank
l33t++

I remembered that AcesHardware ran some nice tests for K6 video cards around 2000.
http://web.archive.org/web/20040601081548/htt … ?article_id=103

3dfx looks strong and Glide is always the best choice. NV looks OK though. Geforce is certainly bottlenecked terribly by K6.

Savage 2000 brings some nice numbers in Q3. From what I've read, S3 mainly optimized for Q3 and UT. The games do run pretty well on it. The same can't be said for most other games though 🤣

Reply 30 of 90, by idspispopd

User metadata
Rank Oldbie
Rank
Oldbie
elianda wrote:

You have to be aware that early K6-2s have no MTRR support and K6-3+ have already extended 3DNow! support, so the K6-3+ is already a rather special CPU in the K6 line.
Be sure to have the K6 systems correctly setup, regarding MTRR, cacheable area, the right driver versions, game patches.

All CXT core K6-2 had MTRRs which includes all models > 350 MHz.
The extended 3DNow! instruction are not utilized by common software, they won't make any difference in benchmarks.

swaaye wrote:

I'd like to see a Covington Celeron (or a simulated one via disabled L2). These CPUs existed to compete with K6 chips and I recall them being recommended over K6 for gaming, for good reason.

This test uses a K6 300 MHz, not a K6-2 so no 3dNow!.

feipoa wrote:

AnandTech’s tests illustrated a 4% increase in performance when using 3DNow! accelerated Quake 2 OpenGL drivers as opposed to using the standard OpenGL drivers under Quake 2 with the K6-3 test system’s Riva TNT video card.
Ouch, only 4%

Tom's Hardware claims a 87% advantage of a K6-2 over a K6 (300MHz, 100MHz FSB, Voodoo2, Quake II Massive1 demo) putting it between a Pentium II 300 and a Pentium II 333.
IMO Quake II using a 3dfx card, AMD's 3DNow! patch and the included 3DNow! MiniGL is probably the best showcase of 3DNow!.

Reply 31 of 90, by Skyscraper

User metadata
Rank l33t
Rank
l33t

I think its a bit unfair to compare a K6-2/3 running on a Via or Ali chipset to a P2/P3 running on one of the greatest chipsets of all time.

I have a slot-1 PC-Chips m727 with Ali Aladdin Pro II chipset. (One of the motherboards i found at my parents place.)
I just bought a second Voodoo Banshee that I hope will arri... lets just say the time frame is not clear 😁. Perhaps it will get stuck on a boat for a month.
ODwilly donated a K6-3 which I hope will get here in a week or two.
When stuff has arrived I will contribute some benchmarks with K6-2/3 on a Pc-Chips m577 vs P2/P3 on a Pc-Chips m727.
Appels to appels.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 32 of 90, by swaaye

User metadata
Rank l33t++
Rank
l33t++

I don't think it's unfair at all. When you bought a Intel P2/P3/Celeron you would usually get an Intel chipset. 440BX was not really a luxury item, but there were even cheaper models like 440LX, EX and ZX. They were all competent.

AMD's K6 was on a shaky platform without doubt. But this is just part of what you were buying into with AMD back then for your low price "value". In fact this is another reason AMD had to be so cheap.

idspispopd wrote:

Tom's Hardware claims a 87% advantage of a K6-2 over a K6 (300MHz, 100MHz FSB, Voodoo2, Quake II Massive1 demo) putting it between a Pentium II 300 and a Pentium II 333.
IMO Quake II using a 3dfx card, AMD's 3DNow! patch and the included 3DNow! MiniGL is probably the best showcase of 3DNow!.

Yeah it is probably the most significant 3DNow effort. I'm not sure if AMD had to write everything or if maybe 3dfx did the OpenGL driver. But it is notable that id wasn't interested in working on 3DNow. It's a lot of work (money) to make this stuff happen and they didn't see it as worthwhile. I'm sure it sold some K6-2s though which would be why AMD got involved.

Reply 33 of 90, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Yea it's what it is 😀

Those are the prime chipsets available in that time...

Personally I think it's an easy choice and a BX440 machine is the way to go if you're building a machine with such performance in mind. Super Socket 7 is a great platform, but for me much more interesting because it can be slowed down to 386 levels. If I need speed (and stability, compatibility and all of that) I would really just go with a nice BX440 board 😀

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 34 of 90, by d1stortion

User metadata
Rank Oldbie
Rank
Oldbie
swaaye wrote:

But it is notable that id wasn't interested in working on 3DNow. It's a lot of work (money) to make this stuff happen and they didn't see it as worthwhile.

They sure said something different though!

Reply 35 of 90, by Skyscraper

User metadata
Rank l33t
Rank
l33t
swaaye wrote:

I don't think it's unfair at all. When you bought a Intel P2/P3/Celeron you would usually get an Intel chipset. 440BX was not really a luxury item, but there were even cheaper models like 440LX, EX and ZX. They were all competent.

AMD's K6 was on a shaky platform without doubt. But this is just part of what you were buying into with AMD back then for your low price "value". In fact this is another reason AMD had to be so cheap.

I diddnt say that it is unreasonable 😁, just a little bit unfair.
It would be a huge surprise if the fat kid won the high jump in the school championship, especially wearing led boots.

Last edited by Skyscraper on 2013-10-11, 20:20. Edited 2 times in total.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 36 of 90, by elfuego

User metadata
Rank Oldbie
Rank
Oldbie

I dont want to be a nagging diva here, but I must warn you not to expect wonders. Especially since an intel equivalent to K6 2/3 is actually a P233 MMX and not a P2, let alone P3! Comparing a P3 450Mhz to a K6/3 @ 600Mhz (overclocked like hell) would be like comparing a Ferrari to a Dacia in a drag race. Especially in FPU performance (*the* thing that games make most use of). But, set up the systems and see for yourself 😉

Reply 37 of 90, by F2bnp

User metadata
Rank l33t
Rank
l33t

A lot of things to cover, a lot of people to answer to 😁.

@d1stortion

I like emulators as much as the next guy, I just don't see the point in running them in old hardware, unless you want to see how well it runs them out of curiosity. Like you say, playing N64 and PS1 games on 99' hardware is pretty amazing.
But on my modern PC I can use software rendering and play all those PS1 games at a higher resolution. It runs them better, that's the key word here. You could stress the fact that a lot of games can run better on modern PCs and I would recognise that. In the end, we're both right 😜.

@feipoa

I think my board has 512K indeed, no need to go beyond 256MB though, even there it is mostly useless!

@mau1

All those reviews are interesting and I've spent countless hours reading them, but in the end there is not one big comparison between K6-3+ and Pentium 2/3 ( and Celeron) CPUs. I want to test as many games as possible and see where the true weaknesses of the K6 are and how that affects performance!
To me, once I found a stable setup, the K6-3+ has been a great CPU. Fast enough for most games I wanted to play and slow and compatible enough with older games. It's been a few years since I've used a 440BX machine, but damn I loved my Pentium 3 500!!!

@everyone 😁

Quake 2's 3Dnow patch is phenomenal. Much like swaaye says, it is probably the most impressive usage of 3Dnow! Unfortunately, most games will not show anywhere near that kind of performance leap.
Was there any 1999-2000 game that made use of SSE in a similar matter? This could be interesting to take a look at.
Alas, it seems I was mistaken and I do not possess a Pentium II 450, but rather a 400. This might actually suit my needs better! Is it possible to run a 100MHz FSB Pentium II at 66MHz FSB? Does that depend on the motherboard?

I'm not expecting miracles, I'd just like to see how much the Pentiums really matter when it comes to Hardware Accelerated 3D ,and of course, I just want to satisfy my curiosity and spend my time doing something that I absolutely love to do 😀.

Reply 38 of 90, by d1stortion

User metadata
Rank Oldbie
Rank
Oldbie

It should be possible running that CPU at 66 MHz FSB, but what would be the point in doing that?

A few words concerning UT: actually compared to Unreal I think this game is not as optimized for contemporary hardware. It's mainly the models that look a lot better but I find the lighting effects in Unreal to be more atmospheric, which is perhaps just a mappers design choice but still. Also, the game has an annoying design flaw where if you fire many rockets at once the framerate will drastically go down even on an "awesome system" (readme: PIII 500, 128MB, Voodoo3). I'm sure this is even worse on Voodoo2 (SLI). But of course competitive players would have played in low resolutions anyway so the point may be moot.

Last edited by d1stortion on 2013-10-11, 20:47. Edited 1 time in total.

Reply 39 of 90, by F2bnp

User metadata
Rank l33t
Rank
l33t

Also, quick answer please, would using a 40GB WD drive from 2004 for the PII/III be unfair?
Since both systems will be running at ATA33, I'm not sure if there would be any actual performance differences. Both are 7200 drives, but the 10GB has 2MB Buffer where as the 40GB has 8MB.

@d1stortion

The point would be to have access to slower speeds, for example the PII 400 should run at 266MHz and the PIII 500 at 333MHz. Not saying I'm going to use them, but it is an option.