Vesa modes in pcpbench dont have any corruption and corruption in speedsys looks like this:

Can you post a screenshot of that particular CMOS setup screen you referred to?
Could be bad menory or also a flaky BIOS ROM chip.
If you can identify your board (via the BIOS string) you might be able to burn a new BIOS.
Found some potential boards: https://theretroweb.com/motherboards/?showIma … nufacturerId=25

jesolo wrote on 2023-06-28, 19:18:

Can you post a screenshot of that particular CMOS setup screen you referred to? Could be bad menory or also a flaky BIOS ROM chi […]
Show full quote

Can you post a screenshot of that particular CMOS setup screen you referred to?
Could be bad menory or also a flaky BIOS ROM chip.
If you can identify your board (via the BIOS string) you might be able to burn a new BIOS.
Found some potential boards: https://theretroweb.com/motherboards/?showIma … nufacturerId=25

If it was bios chip I am not sure why it would cause issues with one specific brand of video cards.
Here is the image of the bios screen and that refresh setting:

It is the bottom one in the pictures on this page https://theretroweb.com/motherboards/s/jamico … 0a-486dlc-386dx except mine has socket instead of integrated cpu and mine has crystal for 33Mhz cpu. Jumper manual on the site does not match my motherboard.

Baoran wrote on 2023-06-28, 19:30:

If it was bios chip I am not sure why it would cause issues with one specific brand of video cards. Here is the image of the bio […]
Show full quote

jesolo wrote on 2023-06-28, 19:18:

Can you post a screenshot of that particular CMOS setup screen you referred to? Could be bad menory or also a flaky BIOS ROM chi […]
Show full quote

Can you post a screenshot of that particular CMOS setup screen you referred to?
Could be bad menory or also a flaky BIOS ROM chip.
If you can identify your board (via the BIOS string) you might be able to burn a new BIOS.
Found some potential boards: https://theretroweb.com/motherboards/?showIma … nufacturerId=25

If it was bios chip I am not sure why it would cause issues with one specific brand of video cards.
Here is the image of the bios screen and that refresh setting:
P_20230628_222712[1].jpg

It is the bottom one in the pictures on this page https://theretroweb.com/motherboards/s/jamico … 0a-486dlc-386dx except mine has socket instead of integrated cpu and mine has crystal for 33Mhz cpu. Jumper manual on the site does not match my motherboard.

Could be an incompatibility with the Tseng chipset. Try enabling auto config (first option) and see what happens.

jesolo wrote on 2023-06-28, 19:40:

Baoran wrote on 2023-06-28, 19:30:

If it was bios chip I am not sure why it would cause issues with one specific brand of video cards. Here is the image of the bio […]
Show full quote

jesolo wrote on 2023-06-28, 19:18:

Can you post a screenshot of that particular CMOS setup screen you referred to? Could be bad menory or also a flaky BIOS ROM chi […]
Show full quote

Can you post a screenshot of that particular CMOS setup screen you referred to?
Could be bad menory or also a flaky BIOS ROM chip.
If you can identify your board (via the BIOS string) you might be able to burn a new BIOS.
Found some potential boards: https://theretroweb.com/motherboards/?showIma … nufacturerId=25

If it was bios chip I am not sure why it would cause issues with one specific brand of video cards.
Here is the image of the bios screen and that refresh setting:
P_20230628_222712[1].jpg

It is the bottom one in the pictures on this page https://theretroweb.com/motherboards/s/jamico … 0a-486dlc-386dx except mine has socket instead of integrated cpu and mine has crystal for 33Mhz cpu. Jumper manual on the site does not match my motherboard.

Could be an incompatibility with the Tseng chipset. Try enabling auto config (first option) and see what happens.

Auto config option did not change anything and performance is exactly the sameit turned on. Still the only option that seem to affect is the refresh hidden/normal option that I mentioned.

Can you test with only bank 0 filled? And then swap with the other 4 SIMMs and try again. My reasoning is that maybe one group of SIMMs you have there doesn't support hidden refresh (even though all the FPM chips I've seen do support it).

Baoran wrote on 2023-06-28, 18:14:

Basically the issue is that if I set it to "hidden" it will cause any tseng labs isa card I have to have corruption on screen in graphics modes but text mode in dos is fine.

That is a known issue with ET4000AX cards, you simply cannot use hidden refresh with them end of story

TheMobRules wrote on 2023-06-29, 00:12:

Can you test with only bank 0 filled? And then swap with the other 4 SIMMs and try again. My reasoning is that maybe one group of SIMMs you have there doesn't support hidden refresh (even though all the FPM chips I've seen do support it).

There is nothing wrong with the memory on his motherboard, that's just how those video cards behave

maxtherabbit wrote on 2023-06-29, 01:58:

That is a known issue with ET4000AX cards, you simply cannot use hidden refresh with them end of story

maxtherabbit wrote on 2023-06-29, 01:59:

There is nothing wrong with the memory on his motherboard, that's just how those video cards behave

Curious, is this something specific to the ISA versions of these cards or certain games? The reason I'm asking is because I'm currently testing a local bus version of the ET4000AX and all the games and graphics modes I've tried run just fine with Hidden Refresh enabled (Doom, Quake, ROTT, and even fired Keen 1 after seeing this post).

I plan to test that board with an ISA version of the Tseng as soon as I find it somewhere in my house. My idea was to compare the custom local bus performance vs. an ISA card with the same chip, I'll keep this in mind.

Honestly I don't know, a local bus 4000AX is a mystery to me, but I would assume they would not be affected since hidden refresh specifically decouples ISA attached DRAM from system DRAM refresh

Does Hidden Refresh disable ISA REFRESH# generation?

maxtherabbit wrote on 2023-06-29, 01:59:

TheMobRules wrote on 2023-06-29, 00:12:

Can you test with only bank 0 filled? And then swap with the other 4 SIMMs and try again. My reasoning is that maybe one group of SIMMs you have there doesn't support hidden refresh (even though all the FPM chips I've seen do support it).

There is nothing wrong with the memory on his motherboard, that's just how those video cards behave

I guess the best option for me would be to find an isa card that would have similar performance to tseng labs cards. I only have 2 trident (9000A and 8900D) isa cards and one ATI (28000-5) isa video card and all the rest are cards with tseng labs chip. I did try to search in vogons and using google for information if other people have had similar issues but nothing came up so the problem is probably not too well known.

Baoran wrote on 2023-06-30, 11:36:

I guess the best option for me would be to find an isa card that would have similar performance to tseng labs cards.

I'm curious, why do you need to eek out the last 1% of performance from ISA bus so badly? I've read some opinions that TSENG cards are the fastest, sure, but the price is compatibility. Or in other words, there is a reason why these cards are the fastest and it's not that other manufacturers were idiots. Though obviously everybody tried to make their product cheap to manufacture.

Trident 8900D with 1MiB of VRAM should be very close to max ISA performance. Well behaved chipset that either already has VESA extensions in the BIOS or will work with small TSR (or UniVBE) in pretty much every game. If you need a card with linear framebuffer access then ISA cards will not be fast enough anyway. The only possible downside to 8900D is the DAC noise can be sometimes seen on modern LCD screens.
Another fine card is WDC WD90C31 - I have never tested it head to head with the Trident but it does seem to have less DAC noise (and the dual BIOS should make it faster in 286 and 386 systems that do not have ROM shadowing).

Conversely, why care so much about using hidden refresh? It's not that significant of an improvement. I'd just keep the Tseng

maxtherabbit wrote on 2023-06-30, 14:10:

Conversely, why care so much about using hidden refresh? It's not that significant of an improvement. I'd just keep the Tseng

I have this 386 build Early 90s 386 build as close to period correct as possible. and I was considering switching the motherboard that has the hidden refresh option to that system because that motherboard has much more effective slow down with turbo button. Basically the motherboard that currently is in the system simply just halves the clockspeed of the cpu when you press the turbo button but this motherboard basically slows it down to around 20% of the normal speed, but it is not as fast as the motherboard currently in the system without setting the hidden refresh option when running at normal speed. I guess I was trying to have best of both worlds by trying to solve this problem. The slowdown is between 5-10% without the hidden refresh option depending on which benchmark I use.

Both ATI and Trident 8900D cards I have are 512kb cards. Somehow I have only ended up with 1MB cards that are made by tseng labs. The reason for wanting to use them is probably the same. Not wanting to slow down system I have already built.