Frunzl wrote on 2020-04-08, 08:05:

Phil from Philscomputerlab gets 1800 with a pretty similar setup, although he is running a K6-3+ at FSB 100 and 550 MHz

Define "pretty similar setup".
There's a lot of things that can be different, motherboard model (chipset can make a big difference). Then there's drivers, especially for videocard. Then you got other devices, each device with his driver, maybe you have a soundcard that can slow down the entire system a little.

Frunzl wrote on 2020-04-08, 08:05:

Could the higher FSB and Clock alone account for this huge difference? I was initially running at 400 MHz, 66 MHz FSB and got a score of 1000 (vs 1100 now), so only about 10% improvement by raising FSB and Clock.

If you raised the FSB by 13% and got an overall improvement of 10%, you pretty much have the answer. 100 MHz FSB means another 33% more than yours. The rest is probably due to carefully chosen drivers and a faster chipset.

ps: why all that overclocking if you even state: "Not that it matters a lot in a retro system". Want to kill the old hardware as fast as possible?

Yes, I thought about the chipset version too.
I dont remember exactly, what Phil was using, but it was more modern obviously, since it ran 100 FSB.

Also, it had AGP, which mine doesnt, but I think, the GPU is bottlenecked by sthg else anyway, since benchmark results are pretty identical all the way from 640x480 to 1024x768.

@derSammler: you are probably right, I just wanted a few ideas. And no Sir, I do not try to kill my hardware, I am just trying to cover the largest possible period of games with a single setup. I also run very decent cooling, dont overvolt the parts and use them quite moderately in general. For me, it's more about getting it to run and benchmarking.

Frunzl wrote on 2020-04-08, 09:30:

Yes, I thought about the chipset version too. I dont remember exactly, what Phil was using, but it was more modern obviously, si […]
Show full quote

Yes, I thought about the chipset version too.
I dont remember exactly, what Phil was using, but it was more modern obviously, since it ran 100 FSB.

Also, it had AGP, which mine doesnt, but I think, the GPU is bottlenecked by sthg else anyway, since benchmark results are pretty identical all the way from 640x480 to 1024x768.

@derSammler: you are probably right, I just wanted a few ideas. And no Sir, I do not try to kill my hardware, I am just trying to cover the largest possible period of games with a single setup. I also run very decent cooling, dont overvolt the parts and use them quite moderately in general. For me, it's more about getting it to run and benchmarking.

Hello,

I agree with the others, you already identified important points by yourself, too.
So:
100MHz FSB vs. 75MHz FSB (most likely the ram runs with FSB speed)
AGP vs PCI bandwidth
550MHz vs. 450MHz.

So all in all, I would say your value is pretty good.

Frunzl wrote on 2020-04-08, 08:05:

Zida 5SVA Motherboard with VIA Apollo VPX chipset (PCI only), FSB set to 75 MHz
AMD K6-3 400 CPU @ 450 MHz (6 * 75)

Hi, can you please write how your jumpers are set up for this frequency (JP2 and JP6)? I'm trying to set it up for K6-2, but it's not working: JP6 on 1-3, 4-6 and JP2 on 1-2 for this CPU does not boot PC (one long beep and a series of short ones) , JP2 on 2-3 gives FSB 60MHz. Mine is rev. 1.2. Thank you.

Revision 1.2 and 1.3 of Zida Tomato 5SVA are different, here it is mentioned:
https://www.svethardware.cz/zakladni-deska-tomato-5sva/1352
(use translator)

for k6-2 266 Mhz and higher, you need to use version 1.3 of the board.
ver: 1.2 works only up to K6-2 233 Mhz.
also BIOS needs to be 1.40 or higher.

Have you got HDD DMA mode enabled? I've found I cannot enable DMA with a TC430HX and ST380215A so there are some CPU hiccups thanks to PIO Mode 4.