theelf wrote on 2025-12-30, 17:13:

Then you are faster than ANY benchmark out there

https://thandor.net/benchmark/33

My numbers are in line from most benchmark online of other people, and of course a classic pentium chipset. If you use new chipsets, PC100 ram etc faster bus you can get better performance, here i talk about 66x3 and standar 72pin ram

The current Pentium 200 system I am using is a period correct 1996 build. Motherboard is a Biostar MB-8500TUC-A with a 430HX (Triton II) chipset with 66 MHz FSB. Only non-period correct things are fans and CF card for storage.

For what it's worth, I just re-ran the Quake benchmarks (both demo1 and demo3) just to confirm numbers.

Quake demo1 hit 44.9 FPS and demo3 hit 42.4 FPS. Screenshots posted below. Also posted Speedsys results.

As I said above, I wonder if the performance difference might be L2 cache size related. 5-6 FPS difference is what I've observed in Pentium Pro benchmarks (256 vs 512), so it seems reasonable similar differences might arise for a regular Pentium.

Video card could also account for some of the difference. In my experience using a Trio 64+ vs Matrox Mystique with the Pentium Pro 256k, the Matrox was about 1.5 FPS faster (47.3 FPS vs 45.9).

Ozzuneoj wrote on 2025-12-30, 17:32:

Am I missing something or does that page not show what video card is being used for all of the CPU tests? That basically invalid […]
Show full quote

theelf wrote on 2025-12-30, 17:13:

Then you are faster than ANY benchmark out there […]
Show full quote

Shponglefan wrote on 2025-12-30, 17:12:

I just edited my post to note that it's running full screen.

Then you are faster than ANY benchmark out there

https://thandor.net/benchmark/33

My numbers are in line from most benchmark online of other people, and of course a classic pentium chipset. If you use new chipsets, PC100 ram etc faster bus you can get better performance, here i talk about 66x3 and standar 72pin ram

Am I missing something or does that page not show what video card is being used for all of the CPU tests? That basically invalidates using that entire page to compare to anything outside of that page.

Likewise, the video card benchmarks do not say which CPU is being used. A very odd omission after doing all of that work. 🤷

Using a faster CPU look at how huge the differences are between graphics chips:
https://www.vgamuseum.info/images/vlask/bench/quake320.png

Very likely the reason that Shponglefan's system is faster is because the Matrox Mystique 4MB is extremely fast in this test compared to other cards from the time. It is faster than any S3 card on the chart.

yup, is a problem always with benchmarks, never give full spects. Matrox are fast, but not too much, and super incompatible with DOS, out of question for any good dos build

In fact S3 cards are in the slow area, i get some extra FPS with the ARK 2000 for example, maybe 4 extra fps in quake, some more in DOOM (1005 realtiks in S3, 890 in ARK!!!!!)

But S3 cards are super compatible, incredible good ones, yet i need to find something not work in a S3

theelf wrote on 2025-12-30, 17:40:

Matrox are fast, but not too much, and super incompatible with DOS, out of question for any good dos build

It really depends on what games one is using it with.

We had a Matrox Millennium card in our Pentium system growing up and it was a good performer for the types of games we played back then (mainly first person shooters like Duke 3D). We never noticed any compatibility issues at the time.

Shponglefan wrote on 2025-12-30, 17:39:

As I said above, I wonder if the performance difference might be L2 cache size related. 5-6 FPS difference is what I've observed in Pentium Pro benchmarks (256 vs 512), so it seems reasonable similar differences might arise for a regular Pentium.

33 to 43 FPS is +~30%, I think we'd know by know if increasing the cache yielded this performance increase 😉

konc wrote on 2025-12-30, 17:52:

Shponglefan wrote on 2025-12-30, 17:39:

As I said above, I wonder if the performance difference might be L2 cache size related. 5-6 FPS difference is what I've observed in Pentium Pro benchmarks (256 vs 512), so it seems reasonable similar differences might arise for a regular Pentium.

33 to 43 FPS is +~30%, I think we'd know by know if increasing the cache yielded this performance increase 😉

To be clear, I don't think cache difference accounts for the full performance difference. But I suspect it may be a significant portion of it.

It does account for an 11% - 12% difference in my Pentium Pro tests.

theelf wrote on 2025-12-28, 02:12:

Maybe my english is not good... no way you will get more performance […]
Show full quote

Maybe my english is not good... no way you will get more performance

Man, i have a triton board, exept for triton ii, istop performance for xlassic pentium, and i get 33fps

You get 34 !!

Do some test, 3dbench 1.0c 148fps, doom shareware 1.9 fullscreen no life bar, -timedemo demo3 2134/1005 realtiks = 74fps

You want more fps? forget classic pentium, buy a pentium 2

I'm not interested in any other pc than a classic pentium, the thing is, i want the best version possible that my chasis can have.

Jasin Natael wrote on 2025-12-29, 15:22:

Pretty much as theelf said. There isn't anything wrong with your current performance. It's on par with other similarly configure […]
Show full quote

Pretty much as theelf said. There isn't anything wrong with your current performance. It's on par with other similarly configured systems. Your CPU isn't underperforming.
If you want better performance you need to upgrade.

There is no need to limit yourself to a Pentium. You need a better CPU, or a better system.
K6-2, Cyrix MII or a later system.

Having in mind that mi SIS chipset is write throug, and my aptiva 2144 has some mainboard with write back chipset, swapping to one of these is a gain, so not to do it is itself limiting myself to an underperformance.

I don't want play 1999 games in my pentium, i'm only benchmarking, i just want the best version of my aptiva, cause... Why not?.

Señor Ventura wrote on 2025-12-30, 18:01:

Having in mind that mi SIS chipset is write throug, and my aptiva 2144 has some mainboard with write back chipset, swapping to one of these is a gain, so not to do it is itself limiting myself to an underperformance.

I suspect only having 256kB of L2 cache is a big part of the bottleneck in your current system.

Shponglefan wrote on 2025-12-30, 18:10:

Señor Ventura wrote on 2025-12-30, 18:01:

Having in mind that mi SIS chipset is write throug, and my aptiva 2144 has some mainboard with write back chipset, swapping to one of these is a gain, so not to do it is itself limiting myself to an underperformance.

I suspect only having 256kB of L2 cache is a big part of the bottleneck in your current system.

Supossedly with 64MB of ram, 256KB of L2 cache is pretty enough, although at the limit. I always understand that with 512KB i barely could gain max 2 or 3fps.

So, right now i need an intel mainboard for my aptiva 2144, cache L2, and the rest is external.

P.D: And updating my bios too, but i'm stucked trying to use the program to flash.

Señor Ventura wrote on 2025-12-30, 18:16:

Supossedly with 64MB of ram, 256KB of L2 cache is pretty enough, although at the limit. I always understand that with 512KB i barely could gain max 2 or 3fps.

As I posted above, with a Pentium Pro 200 MHz, the difference in Quake with an extra 256kB was 5-6 FPS (11-12% increase).

Because the Pentium Pro processors have L2 cache embedded directly, it's easier to do a comparison with everything else being the same.

With a regular Pentium system it would be harder to test, since you'd need the same motherboard, chipset, video card, etc. Changes to other variables would also influence the results.

Shponglefan wrote on 2025-12-30, 17:53:

konc wrote on 2025-12-30, 17:52:

Shponglefan wrote on 2025-12-30, 17:39:

As I said above, I wonder if the performance difference might be L2 cache size related. 5-6 FPS difference is what I've observed in Pentium Pro benchmarks (256 vs 512), so it seems reasonable similar differences might arise for a regular Pentium.

33 to 43 FPS is +~30%, I think we'd know by know if increasing the cache yielded this performance increase 😉

To be clear, I don't think cache difference accounts for the full performance difference. But I suspect it may be a significant portion of it.

It does account for an 11% - 12% difference in my Pentium Pro tests.

How many realtiks or fps you get in doom? can you check?

Fullscreen, nolife bar, no sound, shareware 1.9, doom timedemo demo3

I get 890 with ark and 1005 with S3 Virge

148 fps in 3dbench 1c

About speed i have another PC super socket 7 pentium 233mmx overclock to 266 83bus, riva 128agp, 1mb cache, 64mb pc100

I did not check quake because is not a game have interest but in doom, 3dbench and emulators like callus fps i did not see a big difference in proportional 266-200

sadly in in a train now going to see family i cant do tests

theelf wrote on 2025-12-30, 18:42:

How many realtiks or fps you get in doom? can you check?

964 realtics (77.5 FPS).

I'm not a huge fan of Doom for benchmarking fast Pentiums or later, since it doesn't scale well on faster systems.

On my Pentium 4 3.4 GHz with a GeForce 4200 Ti, I get 835 realtics (89.4 FPS). That's only 15% faster, but the performance difference between a Pentium and Pentium 4 is much more than that.

If I had a Pentium system set up I would run some tests. I have a Classic 100 or 120 Packard Bell but it's mothballed. I do have a 166mmx chip I COULD throw in my ss7 rig.
But all the other hardware is much newer and I don't see the point.
The only other options I have are laptops and are both MMX as well with weird mobile VGA chips so again....apples and oranges.
I still don't think the OPs results are very far out of the ordinary.
FWIW I would rather have the extra CPU cache of the MMX chips and leave the classic in the bin where it belongs.

Ozzuneoj wrote on 2025-12-30, 17:32:

theelf wrote on 2025-12-30, 17:13:

Then you are faster than ANY benchmark out there

https://thandor.net/benchmark/33

My numbers are in line from most benchmark online of other people, and of course a classic pentium chipset. If you use new chipsets, PC100 ram etc faster bus you can get better performance, here i talk about 66x3 and standar 72pin ram

Am I missing something or does that page not show what video card is being used for all of the CPU tests? That basically invalidates using that entire page to compare to anything outside of that page.

Likewise, the video card benchmarks do not say which CPU is being used. A very odd omission after doing all of that work. 🤷

The VGA benchmarks are done with a SiS6326 which is slightly slower than an ‘average’ S3. So the 43 FPS is on a Pentium 20p is fine as someone already mentioned. I scored 41.4 with the SiS card. The CPU I used in the VGA benchmarks is mentioned in brackets as suffix in the name of the graphics card (P100) 😀.

I have been busy on more system information but these pages aren’t finished yet.

Thandor wrote on 2025-12-31, 07:36:

Ozzuneoj wrote on 2025-12-30, 17:32:

theelf wrote on 2025-12-30, 17:13:

Then you are faster than ANY benchmark out there

https://thandor.net/benchmark/33

My numbers are in line from most benchmark online of other people, and of course a classic pentium chipset. If you use new chipsets, PC100 ram etc faster bus you can get better performance, here i talk about 66x3 and standar 72pin ram

Am I missing something or does that page not show what video card is being used for all of the CPU tests? That basically invalidates using that entire page to compare to anything outside of that page.

Likewise, the video card benchmarks do not say which CPU is being used. A very odd omission after doing all of that work. 🤷

The VGA benchmarks are done with a SiS6326 which is slightly slower than an ‘average’ S3. So the 43 FPS is on a Pentium 20p is fine as someone already mentioned. I scored 41.4 with the SiS card. The CPU I used in the VGA benchmarks is mentioned in brackets as suffix in the name of the graphics card (P100) 😀.

I have been busy on more system information but these pages aren’t finished yet.

Oh boy, I don't know how I missed the P100 on the video card benchmark chart. I'm sorry about that.

And thank you for clarifying which VGA card you used for the CPU testing. I thought this was an old page, I didn't realize it was still a work in progress, so again, my apologies. I find myself at your site frequently when researching things and I appreciate all the time and effort you've put into it. 🙂

Shponglefan wrote on 2025-12-30, 18:56:

964 realtics (77.5 FPS). […]
Show full quote

theelf wrote on 2025-12-30, 18:42:

How many realtiks or fps you get in doom? can you check?

964 realtics (77.5 FPS).

I'm not a huge fan of Doom for benchmarking fast Pentiums or later, since it doesn't scale well on faster systems.

On my Pentium 4 3.4 GHz with a GeForce 4200 Ti, I get 835 realtics (89.4 FPS). That's only 15% faster, but the performance difference between a Pentium and Pentium 4 is much more than that.

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.philscomputerlab.com/dos-benchmark-pack.html

theelf wrote on 2026-01-01, 20:45:

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.phi […]
Show full quote

Shponglefan wrote on 2025-12-30, 18:56:

964 realtics (77.5 FPS). […]
Show full quote

theelf wrote on 2025-12-30, 18:42:

How many realtiks or fps you get in doom? can you check?

964 realtics (77.5 FPS).

I'm not a huge fan of Doom for benchmarking fast Pentiums or later, since it doesn't scale well on faster systems.

On my Pentium 4 3.4 GHz with a GeForce 4200 Ti, I get 835 realtics (89.4 FPS). That's only 15% faster, but the performance difference between a Pentium and Pentium 4 is much more than that.

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.philscomputerlab.com/dos-benchmark-pack.html

The attachment IMG_20260101_213758.jpg is no longer available

The attachment IMG_20260101_213808.jpg is no longer available

Not sure what you mean by "not fullscreen". The benchmarks are set to run at a specific resolution.

If it's not stretching to the borders of your monitor for a given resolution, you need to adjust the horizontal and vertical scaling on the monitor.

If you look at the screenshots I previously posted, the 320x200 resolution benchmark demos are (mostly) stretching to the borders of the monitor I used it with.

Señor Ventura wrote on 2025-12-30, 18:01:

I'm not interested in any other pc than a classic pentium, the thing is, i want the best version possible that my chasis can hav […]
Show full quote

theelf wrote on 2025-12-28, 02:12:

Maybe my english is not good... no way you will get more performance […]
Show full quote

Maybe my english is not good... no way you will get more performance

Man, i have a triton board, exept for triton ii, istop performance for xlassic pentium, and i get 33fps

You get 34 !!

Do some test, 3dbench 1.0c 148fps, doom shareware 1.9 fullscreen no life bar, -timedemo demo3 2134/1005 realtiks = 74fps

You want more fps? forget classic pentium, buy a pentium 2

I'm not interested in any other pc than a classic pentium, the thing is, i want the best version possible that my chasis can have.

Jasin Natael wrote on 2025-12-29, 15:22:

Pretty much as theelf said. There isn't anything wrong with your current performance. It's on par with other similarly configure […]
Show full quote

Pretty much as theelf said. There isn't anything wrong with your current performance. It's on par with other similarly configured systems. Your CPU isn't underperforming.
If you want better performance you need to upgrade.

There is no need to limit yourself to a Pentium. You need a better CPU, or a better system.
K6-2, Cyrix MII or a later system.

Having in mind that mi SIS chipset is write throug, and my aptiva 2144 has some mainboard with write back chipset, swapping to one of these is a gain, so not to do it is itself limiting myself to an underperformance.

I don't want play 1999 games in my pentium, i'm only benchmarking, i just want the best version of my aptiva, cause... Why not?.

Recheck your Quake benchmarks, if you get 34fps on fullscreen you have a good pentium benchmark, if you get 34 at small screen (phil bench use fullscreen minus two status bars) then is slow

Shponglefan wrote on 2026-01-01, 21:39:

Not sure what you mean by "not fullscreen". The benchmarks are set to run at a specific resolution. […]
Show full quote

theelf wrote on 2026-01-01, 20:45:

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.phi […]
Show full quote

Shponglefan wrote on 2025-12-30, 18:56:

964 realtics (77.5 FPS).

I'm not a huge fan of Doom for benchmarking fast Pentiums or later, since it doesn't scale well on faster systems.

On my Pentium 4 3.4 GHz with a GeForce 4200 Ti, I get 835 realtics (89.4 FPS). That's only 15% faster, but the performance difference between a Pentium and Pentium 4 is much more than that.

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.philscomputerlab.com/dos-benchmark-pack.html

The attachment IMG_20260101_213758.jpg is no longer available

The attachment IMG_20260101_213808.jpg is no longer available

Not sure what you mean by "not fullscreen". The benchmarks are set to run at a specific resolution.

If it's not stretching to the borders of your monitor for a given resolution, you need to adjust the horizontal and vertical scaling on the monitor.

If you look at the screenshots I previously posted, the 320x200 resolution benchmark demos are (mostly) stretching to the borders of the monitor I used it with.

this is fullscreen, no life or amo bar, minus two bar, is running at i think, 320x152

320x152 is 320x200 is 76%, if you get 34fps at fullscreen you will get more or less 42-45fps at 320x152

theelf wrote on 2026-01-01, 20:45:

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.phi […]
Show full quote

Shponglefan wrote on 2025-12-30, 18:56:

964 realtics (77.5 FPS). […]
Show full quote

theelf wrote on 2025-12-30, 18:42:

How many realtiks or fps you get in doom? can you check?

964 realtics (77.5 FPS).

I'm not a huge fan of Doom for benchmarking fast Pentiums or later, since it doesn't scale well on faster systems.

On my Pentium 4 3.4 GHz with a GeForce 4200 Ti, I get 835 realtics (89.4 FPS). That's only 15% faster, but the performance difference between a Pentium and Pentium 4 is much more than that.

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.philscomputerlab.com/dos-benchmark-pack.html

The attachment IMG_20260101_213758.jpg is no longer available

The attachment IMG_20260101_213808.jpg is no longer available

Ahá, so, yo got 42fps, and i barely reach 34.

I need:
-512KB L2 cache.
-Write back mainboard aptiva 2144 compatible.
-Update bios (i don't know how, i have the image bios, but not the flasher program).

And that's it.

Señor Ventura wrote on 2026-01-02, 02:26:

Ahá, so, yo got 42fps, and i barely reach 34. […]
Show full quote

theelf wrote on 2026-01-01, 20:45:

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.phi […]
Show full quote

Shponglefan wrote on 2025-12-30, 18:56:

964 realtics (77.5 FPS).

I'm not a huge fan of Doom for benchmarking fast Pentiums or later, since it doesn't scale well on faster systems.

On my Pentium 4 3.4 GHz with a GeForce 4200 Ti, I get 835 realtics (89.4 FPS). That's only 15% faster, but the performance difference between a Pentium and Pentium 4 is much more than that.

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.philscomputerlab.com/dos-benchmark-pack.html

The attachment IMG_20260101_213758.jpg is no longer available

The attachment IMG_20260101_213808.jpg is no longer available

Ahá, so, yo got 42fps, and i barely reach 34.

I need:
-512KB L2 cache.
-Write back mainboard aptiva 2144 compatible.
-Update bios (i don't know how, i have the image bios, but not the flasher program).

And that's it.

You made the test again to be sure the screen size?

About 512kb is pretty clear made a small difference here since my fps are similar to the ones with 512kb, better you put in second priority, i tested in my system WB and WT and the difference is about 8%, i get 38fps in WT, this is a small but nice diference, yes maybe a bios update can be useful

Of course if you have WT and a bootleck in graphic card can be possible to have 34fps in fullscreen minus two bars, what yor video card? i read in first post S3 trio in other pages integrated SIS. I tested my system with a sis 6202 PCI and i get 33fps, tested again a trident 9440 card and i barely reach 36fps, tested with a SIS 6326 and i reach 39fps not bad

https://aptivasupport.com/uas/bios/

Found this, maybe check a update before, but take care, is not risk free

theelf wrote on 2026-01-02, 10:15:

You made the test again to be sure the screen size? […]
Show full quote

Señor Ventura wrote on 2026-01-02, 02:26:

Ahá, so, yo got 42fps, and i barely reach 34. […]
Show full quote

theelf wrote on 2026-01-01, 20:45:

Sorry late reply, i just came home from family, i downloaded latest phil benchmark and Quake is NOT fullscreen https://www.philscomputerlab.com/dos-benchmark-pack.html

The attachment IMG_20260101_213758.jpg is no longer available

The attachment IMG_20260101_213808.jpg is no longer available

Ahá, so, yo got 42fps, and i barely reach 34.

I need:
-512KB L2 cache.
-Write back mainboard aptiva 2144 compatible.
-Update bios (i don't know how, i have the image bios, but not the flasher program).

And that's it.

You made the test again to be sure the screen size?

About 512kb is pretty clear made a small difference here since my fps are similar to the ones with 512kb, better you put in second priority, i tested in my system WB and WT and the difference is about 8%, i get 38fps in WT, this is a small but nice diference, yes maybe a bios update can be useful

Of course if you have WT and a bootleck in graphic card can be possible to have 34fps in fullscreen minus two bars, what yor video card? i read in first post S3 trio in other pages integrated SIS. I tested my system with a sis 6202 PCI and i get 33fps, tested again a trident 9440 card and i barely reach 36fps, tested with a SIS 6326 and i reach 39fps not bad

https://aptivasupport.com/uas/bios/

Found this, maybe check a update before, but take care, is not risk free

Yes, same test as you, but 34fps.

Upgrading hardware appart (motherboard M51, M63, M71, although is confusing to find one), i would begin for updating the bios... i have the bios image, the problem is how to flash it.