Here you can see how I used the included I/O sheild but sealed the holes, as well as my network card and hard drive

The case, its not pretty but it was cheap and had a working PSU, CD-Rom, Floppy, and DVD Drive, so I'll take it

I've settled on these tests, running them now with and without the memory interleave patch

SISsoft Memory
SISsoft CPU Floating
SISsoft CPU Interger
SISsoft Whetstone
Final Reality
3D Mark 99 Max
3D Mark 2000
Expendable
GL Quake
Quake II
Quake III
Quake DOS
Doom
PC Player Benchmark

So I couldn't figure out how to do some of the time demo's but I think this is interesting either way

Also not in a bar chart

SISsoft Memory FPU 109 MB/s 220 MB/s
SISsoft CPU Floating 3267 IT/s 3287 IT/s
SISsoft CPU Interger 2669 IT/s 2682 IT/s
SISsoft Whetstone 662 MFLOPS 665 MFLOPS
3D Mark 99 Max 4099 4671
3D Mark 2000 2067 2354
Final Reality 3.57 4.15

The Memory Interleaveing patch helps, but strangly not in QuakeGL, and when its GPU Bound it doesn't matter at all, I might uninstall the patch and repeat this with a Geforce2MX and see what I find also I'm curious if swapping the single stick of ram to a 128mb + 2x64mb sticks would increase preformance as there is more SDRAM to interleave with

Well I tested with my Quadro2 MXR, I'm done running tests this took hours, what I have determined is I'm CPU limited

So what I dont understand is why GLQuake and Quake II are so low, is it just the cpu holding it back? Though if so wouldn't the scores be higher, I would have assumed my FPs would be hmuch higher with the GF2 MX vs the TNT2, as a matter of fact compared to anand tech my preformance is half of what they got

candle_86 wrote on 2022-11-08, 23:41:

So what I dont understand is why GLQuake and Quake II are so low, is it just the cpu holding it back? Though if so wouldn't the scores be higher, I would have assumed my FPs would be hmuch higher with the GF2 MX vs the TNT2, as a matter of fact compared to anand tech my preformance is half of what they got

Back in the day, Anandtech and other websites used to do their benchmarks without a sound card. That provides a slight boost to performance. Also, which Nvidia graphics drivers are you using? I wouldn't go higher than 12.41 with a TNT2 or a GeForce 2 MX card. Regarding the chipset drivers, I'm not sure which VIA 4-in-1 version is the most suitable for that motherboard, but probably something in the 4.35 range.

P.S.

That Dell SB0200 is a terrible sound card. It's a SBLive in name only, being heavily cut down. It doesn't have any hardware-based sound acceleration and may experience problems when resampling sources below 48 kHz. That might account for some of the performance issues as well.

I mean... you give up on a lot of budget stuff to be on a budget. But I guess it's also a regional thing.

You (I) can get a Socket A, Socket 754, Socket 370 or Socket 478 board with a CPU for something like 10$/€ any time.

Joseph_Joestar wrote on 2022-11-09, 05:32:

Back in the day, Anandtech and other websites used to do their benchmarks without a sound card. That provides a slight boost to […]
Show full quote

candle_86 wrote on 2022-11-08, 23:41:

So what I dont understand is why GLQuake and Quake II are so low, is it just the cpu holding it back? Though if so wouldn't the scores be higher, I would have assumed my FPs would be hmuch higher with the GF2 MX vs the TNT2, as a matter of fact compared to anand tech my preformance is half of what they got

Back in the day, Anandtech and other websites used to do their benchmarks without a sound card. That provides a slight boost to performance. Also, which Nvidia graphics drivers are you using? I wouldn't go higher than 12.41 with a TNT2 or a GeForce 2 MX card. Regarding the chipset drivers, I'm not sure which VIA 4-in-1 version is the most suitable for that motherboard, but probably something in the 4.35 range.

P.S.

That Dell SB0200 is a terrible sound card. It's a SBLive in name only, being heavily cut down. It doesn't have any hardware-based sound acceleration and may experience problems when resampling sources below 48 kHz. That might account for some of the performance issues as well.

+1 on the Dell OEM Live card. It will steal CPU cycles as needed.
This will impact performance.

I'll try pulling the sound card and see, as for prices around me for those boards, no not really I've called around local, and even the recyclers for anything Slot1/370 with an Intel Chipset its $60, and all the socket 462/754 stuff is over priced as well unless you want a micro ATX OEM board with blown caps.

candle_86 wrote on 2022-11-07, 01:16:

$Free Pentium III 500 Kamati Passive

I used a katmai p3-450 for a while in a bx motherboard around 2006 as lanparty machine.
it was not a good experience. it was able to run ut, ut2003 and cossacks without an issue.

a little bit of extra performance would have helped a lot.
maybe, if your via chipset can do a little overclock on the fsb, you can clock that cpu around 550-560 mhz and your ram around 110 or 112 if you are lucky, and then that will give notable performance boost.

candle_86 wrote on 2022-11-07, 01:16:

$10 Infineon PC133 512mb

i am surprised it even booted tbh
you probably could have get 2x256 sticks for $1 each or so

candle_86 wrote on 2022-11-07, 01:16:

$10 Creative TNT2

dont worry about glide, a tnt2 is certainly better choice than anything from 3dfx ever made, at least in a pentium3 computer, which has the muscles to drag around the sluggish drivers of the tnt2 (the situation would be massively different in an older system for the favor of 3dfx)

candle_86 wrote on 2022-11-07, 01:16:

$5 Soundblaster Live Dell SB0200

that has to come out i am afraid, unless you like 15 fps instead of 25.
go with sb0220 instead. its also $5, its dell, and its significantly faster due to a non neutered emu10k chip

well going through my parts bin I found a Mach One EV1935 PCI, how does this compare?

As for ram, No I can't get 256mb SD-Ram cheap around me, the recyclers all sell on ebay now, and want that retro computer money

alright I took yalls advice, here are the updated scores minus the Sound Card, i didn't retest with the Quadro, but here it is, I'm entirely happy with the performance now, but now I need a sound card that doesn't hurt performance as bad. Also realized the ram on my card was 5.5NS so I set clock speeds from 125/150 to 150/166 and its rock solid

candle_86 wrote on 2022-11-09, 23:46:

well going through my parts bin I found a Mach One EV1935 PCI, how does this compare?

i dont know. measure the games with or without the card, and you will see how much it slows the games down. also measure them with the sblive. and you will see which one is better.

candle_86 wrote on 2022-11-09, 23:46:

As for ram, No I can't get 256mb SD-Ram cheap around me, the recyclers all sell on ebay now, and want that retro computer money

it doesnt matters any more, as ithe 512mb module luckily works for you... thats usually not the case.

I like these terrible builds using hardware no one would touch, truth is at the time many of us had no choice but to put up with sub par hardware and majority of my favorite games would actually be fine on a P3 500, even with that Dell sound card.
I'd assume majority of generic PCI sound cards are ok as they aren't trying to do anything fancy like Live! decoding in software.

chinny22 wrote on 2022-11-11, 16:57:

I like these terrible builds using hardware no one would touch, truth is at the time many of us had no choice but to put up with sub par hardware and majority of my favorite games would actually be fine on a P3 500, even with that Dell sound card.
I'd assume majority of generic PCI sound cards are ok as they aren't trying to do anything fancy like Live! decoding in software.

not really, some early cmedia for example consumes about 50% of the cpu in some situations, making the games a totally unplayable mess.

So here we go I finally finished it, here is the Tests with No Sound vs SB0200 vs SB0220

In older games the SB0220 handles everything great but in newer games like Quake III/Expendable/Unreal its still less FPS to have sound even with hardware acceleration

What I find really interesting is that while yes the worst preformance is the SB0200 especially in Quake GL and Quake II, in newer games both cards are less overall FPS than no sound, though the SB0220 with the real EMU10k1 is still ahead, its not a night and day diffrence, overall for all games its a 20% improvement, but excluding QuakeGL, Quake II, and the DOS Benchmarks we only see a 14% improvement.

To give it some real numbers

SB0200 overall to SB0220 is a 20% improvement across all game tests, 14% exluding GL Quake and Quake II
No Sound is 5% better overall than SB0220, but excluding DOS, QuakeGL, and Quake II Titles its a 14% improvement
No Sound vs SB0200 overall is a 23% improvement, while excluding DOS, QuakeGL, and Quake II titles its a 27% improvement