Chadti99 wrote on 2021-11-08, 22:31:

Well dang, not seeing ARK level dos performance on the Trio64V+, not yet.

Are you referring to the a PCI Trio64V+, an STB PowerGraph Trio64V+ VLB with stock BIOS, an STB PowerGraph Trio64V+ VLB with 0ws BIOS, or Madao's replica Trio64V+ VLB with 0ws BIOS?

feipoa wrote on 2021-11-09, 10:39:

Chadti99 wrote on 2021-11-08, 22:31:

Well dang, not seeing ARK level dos performance on the Trio64V+, not yet.

Are you referring to the a PCI Trio64V+, an STB PowerGraph Trio64V+ VLB with stock BIOS, an STB PowerGraph Trio64V+ VLB with 0ws BIOS, or Madao's replica Trio64V+ VLB with 0ws BIOS?

Sorry this would be helpful.

An STB PowerGraph Trio64V+ VLB with 0ws BIOS.

It’s a nice card and glad to own it but was really hoping to get closer to the ARK.

So far the ARK1000 beats it in every dos test I’ve tried.

Maybe Madao’s board has some special magic that unleashes the Trio64V+ fury!

This is the card that was in the Stingray box. Same performance as the 2theMax ARK but much improved image quality at 800x600 and higher, not entirely sure why. Was thinking about adding the missing memory sockets but seems pointless on this card.

So if I could get Linear Frame Buffer support working on the Trio64V+ it might eek out more frames. I’ve tried an older version of SDD and no vesa modes seem to work, garbage on screen in any VESA mode. I’ve tried the S3VBE tool, and if LFB is on, the screen simply goes blank with a flash of color every now and then. With LFB off I get similar results to SDD 6.53.

It is very depent of board. I have check with Abit AN4R307 (a AN4TR2 without 3.3V Support) in most case ist distance zero, only DOOM is ARK a bit faster. (61,47 FPS versus
61,78 FPS ) It is checked with DOSbench from Phil.
Same with ASUS VL/I-486SV2GX4 (Rev. 1.7) by a guy. (He has also ARK1000VL and my Trio64V+ )

I have never written: Trio64V+ overtaken ARK1000VL. But he can hunt ARK1000, if it runs with 0 WS )

And you need patched S3VBE20 for using with Trio xx VLB Card.

Madao wrote on 2021-11-09, 14:44:

It is very depent of board. I have check with Abit AN4R307 (a AN4TR2 without 3.3V Support) in most case ist distance zero, onl […]
Show full quote

It is very depent of board. I have check with Abit AN4R307 (a AN4TR2 without 3.3V Support) in most case ist distance zero, only DOOM is ARK a bit faster. (61,47 FPS versus
61,78 FPS ) It is checked with DOSbench from Phil.
Same with ASUS VL/I-486SV2GX4 (Rev. 1.7) by a guy. (He has also ARK1000VL and my Trio64V+ )

I have never written: Trio64V+ overtaken ARK1000VL. But he can hunt ARK1000, if it runs with 0 WS )

And you need patched S3VBE20 for using with Trio xx VLB Card.

No worries Madou. It's a great chipset and board. I think getting LFB working will close the gap for sure.

LFB makes sense for SVGA modes and has no effect in VGA modes. In SVGA modes a S3 Vision 864 with LFB is even faster than the ARK1000. In all tests without LFB the ARK rules the competition.

Madao wrote on 2021-11-09, 14:44:

And you need patched S3VBE20 for using with Trio xx VLB Card.

Is this available to download? Thanks.

Unleash the S3 fury! LFB FTW!

Unfortunately seeing some screen corruption in direct draw games like StarCraft and Red Alert. Always something 🤣. This is without any VESA software of course.

Very nice score.
Unfortunately the VESA tweaks with these early cards is a whack a mole of make one thing work bust another.

BTW, how much ram you are testing with ?
I am noticing that with l2 cache in wb mode 64mb ram is unstable in some games and apps, so need to half it, while in wt mode all is good.

Thanks pshipkov. So far everything I’ve tried with the VESA drivers is working fine. So the directx issue is particularly annoying.

I have to double check my numbers but I think the ARK still beats the Trio64V+ in most of the dosbench tests. In actual games like Quake, Doom, and Duke 3D they appear to be tied.

Would be really nice to see the Trio32 and Trio64 hitting the same speeds with a bios mod.

I’m using a single 32MB stick, I’ve tried two 32MB sticks but experience stability issues like you mentioned, and this was in L1 WB mode.

Spent a moment tonight on this limited ram issue.
It is DRAM WRITE WS related. If set it to 1 then 64mb is solid, otherwise if set to 0 only 32mb is possible.
This motherboard does not speak EDO obviously which could be a path to 50ns chips, but it may be possible to find 4x16 or 2x32 FPM modules that do it. My brief attempt was unsuccessful.
Can you provide a picture of your ram module ?

Sorry to butt in here, but could you remind me which motherboard you are referring to and a repeatable recipe which leads to the failure?

feipoa wrote on 2021-11-11, 09:25:

Sorry to butt in here, but could you remind me which motherboard you are referring to and a repeatable recipe which leads to the failure?

Are you asking about how to reproduce the instability with 64MB of ram? If so we’re talking about the Asus SV2GX4. If I have two 32MB sticks installed Quake will crash shortly after it starts running the demo. This is with 1024k L1 in WB.

pshipkov wrote on 2021-11-11, 09:12:

Spent a moment tonight on this limited ram issue. It is DRAM WRITE WS related. If set it to 1 then 64mb is solid, otherwise if s […]
Show full quote

Spent a moment tonight on this limited ram issue.
It is DRAM WRITE WS related. If set it to 1 then 64mb is solid, otherwise if set to 0 only 32mb is possible.
This motherboard does not speak EDO obviously which could be a path to 50ns chips, but it may be possible to find 4x16 or 2x32 FPM modules that do it. My brief attempt was unsuccessful.
Can you provide a picture of your ram module ?

Here’s what I’m using atm. It has chips on both sides. Panasonic 60ns chips.

Are you already using the BIOS I posted which corrected two register settings directly related to L2: WB issues? I had to run L2 in WT mode until this fix, assisted by users GigAHerZ and PC-Engineer.

See this post Re: Evolution of a Socket3 System to a POD @100MHz and my reply to myself has the BIOS linked.

feipoa wrote on 2021-11-11, 14:00:

Are you already using the BIOS I posted which corrected two register settings directly related to L2: WB issues? I had to run L2 in WT mode until this fix, assisted by users GigAHerZ and PC-Engineer.

See this post Re: Evolution of a Socket3 System to a POD @100MHz and my reply to myself has the BIOS linked.

Yes, using your L2 fix bios.

Do all VLB graphics cards cause issues in Quake with L2:1024:WB on your SV2GX4?

I will pull out my Asus SV2GX4 system this evening and see if I can replicate your crashes. My system has an S3 968 installed. Are you also able to replicate the crash with this card installed??

feipoa wrote on 2021-11-11, 14:30:

Do all VLB graphics cards cause issues in Quake with L2ဤWB on your SV2GX4?

I will pull out my Asus SV2GX4 system this evening and see if I can replicate your crashes. My system has an S3 968 installed. Are you also able to replicate the crash with this card installed??

The crashes only seem to happen with two 32MB modules installed, a single 32MB module is stable.

I do need to try more cards, I should have a 968 to test with by the end of the week.