zuldan wrote on 2025-04-04, 09:44:

No performance issues with a Banshee (PCI) on MVP3. What’s strange is my Voodoo 3 (AGP) was getting a score of 2400 on this same motherboard? I had a refresh install of Win98 SE with the chipset drivers installed (including the AGP driver).

This further proves that it's NOT the same issue as the one described in this thread which affects certain Ali Aladdin V chipset revisions, because those have the same (low) performance with either an AGP or a PCI Voodoo card.
My guess is that the issues with MVP3 can probably be mitigated by playing with the chipset driver versions & co.

I agree. The Aladdin V issues seem to be a very specific hardware issue.

I'd suggest opening a new thread to discuss your issue with the MVP3 motherboard.

I never did test the Rev 1.1 gigabyte board. Will need to try again.

Test results.

- DFI K6BV3+/66 Rev B2 - MVP3 + 596B
- 128MB RAM
- K6-2 500MHz (standard, no plus or anything else)
- 8 GB WD Caviar 28400
- VIA 4in1 4.35

Cards tested:
- Voodoo 3 2000 PCI

- Voodoo 3 3000 AGP

PcBytes wrote on 2025-04-04, 19:42:

Test results. […]
Show full quote

Test results.

- DFI K6BV3+/66 Rev B2 - MVP3 + 596B
- 128MB RAM
- K6-2 500MHz (standard, no plus or anything else)
- 8 GB WD Caviar 28400
- VIA 4in1 4.35

Cards tested:
- Voodoo 3 2000 PCI

- Voodoo 3 3000 AGP

Yeah, those scores are abysmal. But they are NOT related to the 596B southbridge, something else is going on here.
Then again, you are using a regular K6-2 CPU, I haven't tested one of these in a very long time, so not sure what the expected performance should be in 3DMark99. Do you have a K6-2/3+ to test as well?

I have a K6-2+ at the same speed I could test, currently sitting in a GA-5AX rev 4.1.

K6-II+ 500, same cards tested.

V3 2000 PCI:

V3 3000, AGP:

And another set of tests. This time, FIC PA-2013 (with older 586 southbridge) and K6-II+ , otherwise same cards.

V3 2k PCI

V3 3k AGP

I'l test a 693/596B Katmai board just to rule out VIA being VIA.

This is where it gets fishy. Below results are from a 500MHz Katmai, VIA693+596B based MSI 6156VA. I'm pretty sure these are NOT the scores a Pentium 3 should be getting.

Same settings were applied to all three mainboards tested:
- no other cards inserted
- USB, serial, LPT disabled
- RAM set to CL2

V3 2k PCI

V3 3k AGP

I would say this is in line with what I would expect from VIA, compared to a 440BX board.

Here's a Pentium 3 Katmai 600 MHz benchmark video I created a few years ago (using a 440BX board) and a GeForce 2 Ti @ Ultra clockspeeds: https://youtu.be/2UkFnM_rYqE?si=BLbWf3SCaCtighJt

~5600 points in 3DMark99 with a much faster video card, much faster chipset and a faster CPU... so, I think your 3DMark99 score is normal for an Apollo Pro 133 board with a Katmai 500 and a Voodoo 3.

I'm still having some doubts. There's something wrong that the P3 500 (which is 100MHz less than yours) scores on par with the K6-II+ @ FIC PA-2013. I would've expected both cards to at least score 4100 for AGP and 3500-3700 for PCI, given I've tried to keep the same settings for all of them:

- disabled USB, Serial, Parallel ports
- disabled onboard ES1373 for the 6156VA
- used VIA 4.35 drivers, as most recommend

PcBytes wrote on 2025-04-05, 15:24:

I'm still having some doubts. There's something wrong that the P3 500 (which is 100MHz less than yours) scores on par with the K6-II+ @ FIC PA-2013. I would've expected both cards to at least score 4100 for AGP and 3500-3700 for PCI.

3DMark99 is very well optimized for AMD's 3DNow! instructions, so those high scores you will see in that benchmark will very rarely translate into real world performance.

Just look at these two:
https://youtu.be/2UkFnM_rYqE?t=214
https://youtu.be/jrV6vPGR3Yc?t=410

In the first video, it's the Pentium 3 Katmai 600 on a 440BX board with the GeForce 2 Ti, scoring 5619 3DMarks / 9139 CPU marks
In the second video it's the K6-3+ @ 633 MHz, on an Ali Aladdin V board (Asus P5A), with the same GeForce 2 Ti card. And, surprisingly, it managed to score 5420 3D marks / 9276 CPU marks.

Is the K6-3+ comparable to the Katmai 600 (especially when the latter is running on a 440BX board)? Of course not, not even close, in actual games the Katmai 600/440BX combo will be anywhere between 30% to 50% faster, with few exceptions.

Chaintech 5AGM2
K6-2+ 550
128MB
Voodoo 3 3000 AGP 16MB

Win98 SE

The via 4-in-1 drivers are not installed because Windows won’t complete boot when they’re installed.

These results are massively better than my GA-5AA rev 1.1.

Asus P5A-B Rev. 1.05 with resistor mod
AMD K6-2+ 550MHz
128MB PC100
Voodoo 3 3000 AGP 16MB

Windows 98SE
DX7 and
AGP GART 2.13