VOGONS


First post, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

FINALLY finished benchmarking that last CPU 😵

Windows benchmarking takes forever, that's for sure.

I tested a few CPUs ranging from the Pentium 100 all the way to one of the fastest options, a K6-III+/550.

The card scales very nicely and it is clear that the fast AMD K6 chips are a great CPU to go with a Voodoo 2.

A short video with more details and some funky music from the game Incoming:

http://www.youtube.com/watch?v=ru2juYAXiug&feature=youtu.be

And here just the chart:

ztgAOcq.png

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

Reply 1 of 23, by NitroX infinity

User metadata
Rank Member
Rank
Member

Nice benchmarks 😀

If you're open to suggestions;
Add K5 & K6 benchmarks.
Compare all cpu's at the same speed to see which one has the best performance per megahertz. (all with a 66MHz FSB, to keep it equal)
Maybe add A Pentium 2 233MHz and 450MHz to the mix?

Reply 2 of 23, by leileilol

User metadata
Rank l33t++
Rank
l33t++

Cyrix 6x86?

apsosig.png
long live PCem

Reply 3 of 23, by F2bnp

User metadata
Rank l33t
Rank
l33t

Ubercool! The jump from the MMX233 to the K6-2 300 is quite big.
Very impressive mau, what else do you have in store ? 😀

Reply 4 of 23, by sunaiac

User metadata
Rank Oldbie
Rank
Oldbie

Voodoo 2 is much faster than Voodoo 1 in GLQuake even on a P100 !
Would never have guessed.

R9 3900X/X470 Taichi/32GB 3600CL15/5700XT AE/Marantz PM7005
i7 980X/R9 290X/X-Fi titanium | FX-57/X1950XTX/Audigy 2ZS
Athlon 1000T Slot A/GeForce 3/AWE64G | K5 PR 200/ET6000/AWE32
Ppro 200 1M/Voodoo 3 2000/AWE 32 | iDX4 100/S3 864 VLB/SB16

Reply 5 of 23, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Yea the jump to K6-2 is quite dramatic. The 100 MHz FSB is likely part of the reason. One you have on-die cache things really start to fly.

The good old Pentium seems so out of date.

In terms of other exotic chips, donate them to me and I will add them to the chart 😀

What's next? Well I really want to look at some Socket 487 Pentium 4 stuff.

Voodoo 2 / SLI scaling on slot 1 and S370 would be interesting but I think I need a break from 3Dfx 😊

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

Reply 7 of 23, by kixs

User metadata
Rank l33t
Rank
l33t

Quantum 3D's Obsidian2 X-24 had a software SLI disable (obvious).

Can you test the Pentium MMX on 100MHz FSB (it should run at 300MHz with not too much problems) to see how it compares with K6-2@300MHz.

Back in the days I went from P-MMX 250MHz (2.5x100MHz) to K6-2 500MHz on some MVP3 board and I really didn't see that much improvement in games with Voodoo3.

Visit my AmiBay items for sale (updated: 2025-10-29). I also take requests 😉
https://www.amibay.com/members/kixs.977/#sales-threads

Reply 8 of 23, by GL1zdA

User metadata
Rank Oldbie
Rank
Oldbie

Thanks for this benchmark, very good work. I'm always interested in benchmarks showing how GPUs scale, helps a lot when choosing what parts to use for retro rigs.

getquake.gif | InfoWorld/PC Magazine Indices

Reply 9 of 23, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

The latest 3DFX Voodoo 2 driver has a tick box to disable SLI. But it's missing the disable V-Sync boxes. However you can get these back by updating a DLL 😀

So I can now disable / enable SLI on the fly which will make benchmarking faster.

I will start tomorrow but everything is ready to go 😀

Slot 1 scaling comes first. I don't have that many CPUs, but a PII 233, a PII 350, a PIII 500, 600, 700 and 800 I believe.

And after that I will move to Socket 370 and test a bunch of 133 MHz FSB CPUs all the way up to 1.4 GHz.

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

Reply 10 of 23, by gandhig

User metadata
Rank Member
Rank
Member

You must have quite a test lab and loads of patience to change the processor, reinstall the heatsink every time with fresh thermal paste for these benchmark marathons or do you do it somewhat differently(and easy)?

Dosbox SVN r4019 + savestates Build (Alpha)
1st thread & the only one related to the forum(?)...warning about modern-retro combo
Dead, but, Personal Favourite
Replacement for Candy Crush...Train the Brain

Reply 11 of 23, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

As if I apply thermal paste 🤣

These CPUs run quite cool and the cooler I use is not the tiniest.

On the other hand I only need to do this once. And then... become immortal 🤣

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

Reply 12 of 23, by vetz

User metadata
Rank l33t
Rank
l33t
gandhig wrote:

You must have quite a test lab and loads of patience to change the processor, reinstall the heatsink every time with fresh thermal paste for these benchmark marathons or do you do it somewhat differently(and easy)?

Thermal paste is not needed for a Pentium. Just use a heatsink with a fan and you're fine (even without fan works as well). It's quite open and testing doesn't last long. These CPU's produce very little TDP.

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes

Reply 13 of 23, by Nahkri

User metadata
Rank Member
Rank
Member

I see that difference between a k6-2+ and a k6-3+ are small,the extra l2 cache doesn't affect gaming to much.

Reply 14 of 23, by gandhig

User metadata
Rank Member
Rank
Member

No pentium1 experience for me.

Immortality(if there is such thing) or not in the narrow minded world nowadays, your contribution is definitely way up there for us souls who have no means to gather data from such old hardware. It almost makes me consider buying a PCI voodoo card with my hard earned money. Looking forward to Socket 370 benching(can't escape applying thermal paste 🤣 ).

Dosbox SVN r4019 + savestates Build (Alpha)
1st thread & the only one related to the forum(?)...warning about modern-retro combo
Dead, but, Personal Favourite
Replacement for Candy Crush...Train the Brain

Reply 16 of 23, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Ran into a weird issue and worth letting others know about it!

I'm working on Slot 1 Voodoo 2 scaling and started with a P III 600. I recorded all results, but when doing other CPUs and coming back to a faster model (750 MHz), the scores of Unreal with SLI where too slow.

I tried all software tricks, re-installing OS, different drivers, but nothing worked.

Then I remembered that I changed the RAM at some point. From a single 128MB stick to two 64 MB sticks.

So I put the single 128MB stick back and the results scaled correctly again. We are talking of a difference of 100 vs 70 fps.

Sad news is I will start all over again, the good news is that I found out what the issue was 😀

But what a weird one...

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

Reply 17 of 23, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Found drivers (They are Creative drivers) which have the "disable VSync" tick boxes for Glide and Direct 3D and also a "disable SLI" tick box!

Works with all the games and makes benchmarking very easy 😀

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

Reply 18 of 23, by leileilol

User metadata
Rank l33t++
Rank
l33t++

Will you go all the way down to PII 233? 😀

apsosig.png
long live PCem

Reply 19 of 23, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++
leileilol wrote:

Will you go all the way down to PII 233? 😀

That is the slowest chip I have, yes 😀

The fastest is a Coppermine 750 and a Celeron Coppermine-128K 900.

Beyond that I will switch to a S370 board.

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