Señor Ventura wrote on 2025-12-07, 11:29:

[...]

Yes, i'm on it, but just for information, these games doesn't performed like that in a p133+voodoo, or p166mmx alone.

The platform can make a big difference. You're running this P200 on a slug of a SiS5511 chipset. Maybe the P133 was running on a vastly faster i430TX with SDRAM. Also P166MMX has twice the L1 cache and will generally outperform a P200 even on the same platform.

We know your system is slow E2E, but until you do some benchmarks that specifically stress CPU ALU, CPU FPU, memory bandwidth etc we can't say *why* it's so slow and so can't suggest ways to speed it up.

One thing worth mentioning is sound: back in the day generating sound was CPU-intensive. Even on a decently-specced system for UT (say a P3-1000EB), turning sound off could give you a pile of FPS more, and how many FPS you lose could differ a lot depending on which sound card/onboard/integrated sound you have. If you're running that 2FPS with sound on, try disabling it and see what that does for your FPS.

Although really, I'd say don't bother with UT. The very best you can hope for is for it to run barely playably at absolute minimum settings. That's not fun, that's masochism. I was recently playing it on a dual Athlon MP 2400+ system with Geforce4Ti GPU. And even there I had to lower resolution and colour depth to get it running really smoothly. Then it was fun, but that's a whole 5 year newer hardware that what you're trying to get it to work with. Give yourself a break and play stuff from 1996 or earlier. That will be fun on that CPU.

bertrammatrix wrote on 2025-12-07, 00:59:

We were less spoiled in those days - we knew minimum requirements really meant "good luck with these" and usually just put up with some abysmal resolution to have a chance at playing (and still got slideshows when more enemies showed up)

Yeah. Minimum was more like the game launches and runs on those specs, that is it technically ”works” on such system.

People often look game performance through moden glasses although expectations were very different in the olden days. Redinf old hardware and game reciews paints completely different picture: even during socket A era 30-40FPS was generally considered fine performance.

CharlieFoxtrot wrote on 2025-12-07, 12:55:

bertrammatrix wrote on 2025-12-07, 00:59:

We were less spoiled in those days - we knew minimum requirements really meant "good luck with these" and usually just put up with some abysmal resolution to have a chance at playing (and still got slideshows when more enemies showed up)

Yeah. Minimum was more like the game launches and runs on those specs, that is it technically ”works” on such system.

People often look game performance through moden glasses although expectations were very different in the olden days. Redinf old hardware and game reciews paints completely different picture: even during socket A era 30-40FPS was generally considered fine performance.

Nah. back in the day people were acutely aware of how crap 15fps was too. They just didn't have the means to run it any faster.

There's a reason I kept running WIndows 3.1 until 1999: my P60 with 16MB RAM could have run Windows 95 or even 98FE, but it would have been slower than running stuff on 3.1, and anything that wouldn't run on 3.1 would be too slow to be enjoyable anyway. I'd rather play Doom2 at high speed than torment myself with a Quake2 slideshow.

I recall when unreal came out that I had to go to advanced settings and turn off things like reflections in order to make it playable. It then became a benchmark with each upgrade to see how many FPS I could get on Unreal.

dionb wrote on 2025-12-07, 12:20:

The platform can make a big difference. You're running this P200 on a slug of a SiS5511 chipset. Maybe the P133 was running on a […]
Show full quote

Señor Ventura wrote on 2025-12-07, 11:29:

[...]

Yes, i'm on it, but just for information, these games doesn't performed like that in a p133+voodoo, or p166mmx alone.

The platform can make a big difference. You're running this P200 on a slug of a SiS5511 chipset. Maybe the P133 was running on a vastly faster i430TX with SDRAM. Also P166MMX has twice the L1 cache and will generally outperform a P200 even on the same platform.

We know your system is slow E2E, but until you do some benchmarks that specifically stress CPU ALU, CPU FPU, memory bandwidth etc we can't say *why* it's so slow and so can't suggest ways to speed it up.

One thing worth mentioning is sound: back in the day generating sound was CPU-intensive. Even on a decently-specced system for UT (say a P3-1000EB), turning sound off could give you a pile of FPS more, and how many FPS you lose could differ a lot depending on which sound card/onboard/integrated sound you have. If you're running that 2FPS with sound on, try disabling it and see what that does for your FPS.

Although really, I'd say don't bother with UT. The very best you can hope for is for it to run barely playably at absolute minimum settings. That's not fun, that's masochism. I was recently playing it on a dual Athlon MP 2400+ system with Geforce4Ti GPU. And even there I had to lower resolution and colour depth to get it running really smoothly. Then it was fun, but that's a whole 5 year newer hardware that what you're trying to get it to work with. Give yourself a break and play stuff from 1996 or earlier. That will be fun on that CPU.

Well, UT and unreal is a fresh refference for me to compare, that is why i tried with this... Then, i go with "speedsys" and quake for benchmarking, right?.

I will comment the results tomorrow, thanks to all!!.

Señor Ventura wrote on 2025-12-06, 18:53:

Greetings. […]
Show full quote

Greetings.

Lately i begun to install some games in my ibm aptiva, this is a pentium 200 non mmx, 256KB L2 caché, S3 trio64+, voodoo 1, awe 64 gold... but, i don't know, i don't think unreal tournament at the lowest graphic settings with the lowest resolution can only reach a couple of frames per second in software renderer.

What i'm doing wrong?.

Thanks in advance!.

Why you run Software Renderer ? choose 3Dfx Glide 😉

matze79 wrote on 2025-12-07, 15:22:

Señor Ventura wrote on 2025-12-06, 18:53:

Greetings. […]
Show full quote

Greetings.

Lately i begun to install some games in my ibm aptiva, this is a pentium 200 non mmx, 256KB L2 caché, S3 trio64+, voodoo 1, awe 64 gold... but, i don't know, i don't think unreal tournament at the lowest graphic settings with the lowest resolution can only reach a couple of frames per second in software renderer.

What i'm doing wrong?.

Thanks in advance!.

Why you run Software Renderer ? choose 3Dfx Glide 😉

I'm still collecting software ^^

dionb wrote on 2025-12-07, 12:59:

CharlieFoxtrot wrote on 2025-12-07, 12:55:

bertrammatrix wrote on 2025-12-07, 00:59:

We were less spoiled in those days - we knew minimum requirements really meant "good luck with these" and usually just put up with some abysmal resolution to have a chance at playing (and still got slideshows when more enemies showed up)

Yeah. Minimum was more like the game launches and runs on those specs, that is it technically ”works” on such system.

People often look game performance through moden glasses although expectations were very different in the olden days. Redinf old hardware and game reciews paints completely different picture: even during socket A era 30-40FPS was generally considered fine performance.

Nah. back in the day people were acutely aware of how crap 15fps was too. They just didn't have the means to run it any faster.

There's a reason I kept running WIndows 3.1 until 1999: my P60 with 16MB RAM could have run Windows 95 or even 98FE, but it would have been slower than running stuff on 3.1, and anything that wouldn't run on 3.1 would be too slow to be enjoyable anyway. I'd rather play Doom2 at high speed than torment myself with a Quake2 slideshow.

I didn't say anything about 15 FPS. It is a simple fact that 30-40FPS was completely acceptable performance early 2000s and often many new AAA games didn't run faster settings cranked up and using decent resolution and AA. 60FPS magic number became common only when 60Hz LCD displays were the norm among gamers.

And I disagree how aware people were about the FPS. People weren't aware because in many cases there was realistically no hardware where you could witness the game running smoothly. For example I played Doom in 1993 with my 486SX-20 and it was horrible by modern standards and often dropped well into to single digits. I think the best system among my friends at that point was DX33, so no one had much better hardware. You simply couldn't know how much Doom sucked sub 10FPS because you practically never saw any better.

Same goes with your basic Amiga 500 that were plenty where I live. Most demanding Amiga games CPU wise were awful with your vanilla 68k Amiga, but that was how they were back in the day and nobody said that they suck balls. Again, there was no comparison to something better.

CharlieFoxtrot wrote on 2025-12-07, 15:46:

And I disagree how aware people were about the FPS. People weren't aware because in many cases there was realistically no hardware where you could witness the game running smoothly. For example I played Doom in 1993 with my 486SX-20 and it was horrible by modern standards and often dropped well into to single digits. I think the best system among my friends at that point was DX33, so no one had much better hardware. You simply couldn't know how much Doom sucked sub 10FPS because you practically never saw any better.

I don't know about anyone else but, with rare exceptions, I didn't buy the latest games in the 90s. I was at school or uni. I tended to buy older games and so whatever mid-range PC I had was able to run those games fine. I did struggle with Unreal though, as that was bought as a new game. I think I bought a graphics card upgrade specifically to play it. A Riva TNT I believe.

CharlieFoxtrot wrote on 2025-12-07, 15:46:

[...]

For example I played Doom in 1993 with my 486SX-20 and it was horrible by modern standards and often dropped well into to single digits. I think the best system among my friends at that point was DX33, so no one had much better hardware. You simply couldn't know how much Doom sucked sub 10FPS because you practically never saw any better.

I took one look at Doom on my 386-16 and went straight back to Wolfenstein3D which did run smoothly. You don't need to have seen Doom on a Pentium to know that Doom on what most of us back then had was not a great experience.

My money is on the IGP stealing memory bandwidth.
Many of those cheap SiS boards didn't automatically disable the onboard video when installing a discreet graphics card.
I had one of the crappy PC Chips boards with the 530 chipset back in the day and it was horrendous.
5511 is a bit earlier, but the song remains the same.

RetroPCCupboard wrote on 2025-12-07, 07:12:

Ok. I take back what I said about it not working on Voodoo 1 4mb. I have just tried it on another machine that has Voodoo1 (this time paired with a Pentium MMx 233Mhz @300Mhz). It runs at 512x384 low settings at 24fps in Glide.

If I run in software mode @320x240 on this machine (with S3 Trio 3D card) I get 17FPS. So definitely the Voodoo 1 gives a better experience.

How many mb vram is the trio64 in the aptiva? 1? 2?

IBM typically went for idiot proof reliability, it would not be unusual for memory / cache timings to be set at ridiculously slow settings with no way of changing them.

Pentium 200 non mmx was the fasetest consumer cpu in moment of quake 1 release (june ’96. In moment of unreal release (july 1998) the fastest was theoretically Pentium ii Deschutes (fsb 100) 400MHz that as they say was about 3.5x faster in unreal. Those days 3 years was really a milestone in computer performance progres. And AFAIK Unreal was „Crysis” of it’s times - so a reference point for the power of your PC. Maybe not directly answerig topic question but some historical bacground may help umderstand those low fps.

Still i can't benchmark anything... i installed the unofficial service pack and i broke windows 😮... and for installing windows, first i have to copy alll the .cab in the hdd cause it tends to freeze all the process from the cd drive.

So, benchmarking is working in progress yet... ^^u

edit: all this issues from the scratchy beginning... old memories... xD

By the way... don't know if marking the cd drive from the bios how "compatible" or "full performance" has any fault...

Your motherboard\BIOS must be configured to run it as a Pentium 120 then, somehow. The performance lines up with the numbers being reported by SpeedSys and your BIOS.

The CPUID is being reported as 052C, which is a Pentium 200 non-mmx, like you said.

Did you upgrade the processor yourself? Maybe the board just doesn't support it so it is running at the wrong speed.

Jasin Natael wrote on 2025-12-08, 20:48:

My money is on the IGP stealing memory bandwidth. Many of those cheap SiS boards didn't automatically disable the onboard video […]
Show full quote

My money is on the IGP stealing memory bandwidth.
Many of those cheap SiS boards didn't automatically disable the onboard video when installing a discreet graphics card.
I had one of the crappy PC Chips boards with the 530 chipset back in the day and it was horrendous.
5511 is a bit earlier, but the song remains the same.

I tried to disable the device under windows 98 just in case it impacts on performance, but i'm not so sure if it's working. Bios doesn't seems to admit changing that.

The same with the processor, i don't know if frequency is limited by the bios, or is only readable information.

edit: i have uploaded some pictures of speed sys and quake benchmarks in its 686 benchmarks thread.

You can test the graphics shared memory issue by just removing the discreet card and using the IGP.
If it works more or less normal and not snail slow, then it is apparently not disabling it when using the other card.
The CPU issue you should be able to check the clock speed either in DOS or in Windows to see what it is running at.
CPUID, Everest, or even just dxdiag should show you that in Windows.

Jasin Natael wrote on Today, 17:17:

You can test the graphics shared memory issue by just removing the discreet card and using the IGP. If it works more or less no […]
Show full quote

You can test the graphics shared memory issue by just removing the discreet card and using the IGP.
If it works more or less normal and not snail slow, then it is apparently not disabling it when using the other card.
The CPU issue you should be able to check the clock speed either in DOS or in Windows to see what it is running at.
CPUID, Everest, or even just dxdiag should show you that in Windows.

Seems like i'm having some advances... now i only have one device (s3 pci), but it doesn't recognize the drivers and marks it with (!). I try with one for windows 95, but it just seems not to accept even the default ones.

The CPU definitely shows 120mhz in every bench program i use (119.mhz), probably my bios sets this frequency and it can't be changed from there... maybe some jumper?...

Señor Ventura wrote on Yesterday, 14:38:

Still i can't benchmark anything... i installed the unofficial service pack and i broke windows o_O... and for installing window […]
Show full quote

Still i can't benchmark anything... i installed the unofficial service pack and i broke windows 😮... and for installing windows, first i have to copy alll the .cab in the hdd cause it tends to freeze all the process from the cd drive.

So, benchmarking is working in progress yet... ^^u

edit: all this issues from the scratchy beginning... old memories... xD

By the way... don't know if marking the cd drive from the bios how "compatible" or "full performance" has any fault...

Yeah, the unofficial pack has been horrid for a few years now.

Just download a copy of the 2004 Microsoft update CD and install the updates that way.