wierd_w wrote on Yesterday, 16:10:

Just enapsulating the tsr, and getting it to run as an injected optrom should be all that's needed?

There is some chance this could work and seems much simpler to be done. I have to find a spare EEPROM chip, which is suitable for this card.

If your vbios dump is missing the mysterious 6kb option rom found in Nexxen's, then there should be enough room in the padding.

There's no problem the garbage OpROM in the Nexxen's BIOS ROM dump to be removed (overwritten with FF's) or overwritten with another useful OpROM, since it is 100% redundant and useless.

My problem is that I'm out of spare EEPOMs and in the same time there are too many fake chips sold through ali-baba-orient-express site + the rising shipping costs and the new ursulian €u taxes for all the small packages from China. I have to borrow some suitable EEPROM from one of my working vintage motherboards just for this test.

analog_programmer wrote on Yesterday, 16:49:

There's no problem the garbage OpROM in the Nexxen's BIOS ROM dump to be removed (overwritten with FF's) or overwritten with another useful OpROM, since it is 100% redundant and useless.

My problem is that I'm out of spare EEPOMs and in the same time there are too many fake chips sold through ali-baba-orient-express site + the rising shipping costs and the new ursulian €u taxes for all the small packages from China. I have to borrow some suitable EEPROM from one of my working vintage motherboards just for this test.

You people are amazing 😀

I have some spare BIOS chips. W27C512-45 (64KB)
Edit: I also have 256 with UV window. EEPROM would be faster though.
Could a bigger size chip be suitable anyway?
I'm willing to test whatever.

Nexxen wrote on Yesterday, 17:12:

You people are amazing :) […]
Show full quote

You people are amazing 😀

I have some spare BIOS chips. W27C512-45 (64KB)
Could a bigger size chip be suitable anyway?
I'm willing to test whatever.

Thank you!

Sometimes double size BIOS chips also work, just the BIOS dump has to be concatenated twice. Tomorrow I will see, if it is possible to turn the VBE TSR into usable OpROM small enough to fit in the free space of 8 KB. If I came out with a VESA-capable BIOS candidate. I will share it here, so you could test it. For sure I have suitable spare EPROM chips, but not EEPROMs, and I don't have UV-eraser lamp 🙁

Nexxen wrote on Yesterday, 14:03:

I tried the 9504 and the card works. That is not preventing anything to start.

haha, real survivor 😀 maybe just the oscillator driver got fried. Crystal needs driving to oscillate/resonate, without active driver they just sit dead. replace with https://www.jameco.com/z/ACH-14-31818MHZ-EK-A … an_2344983.html

analog_programmer wrote on Yesterday, 14:28:

There are no holes or pads for capacitors near crystal resonator on my card either. I doubt those are included inside the UM9502 circuit. I often see old motherboards with no decoupling(?) capacitors connected to crystal resonators, so probably they are not so important.

there are cases where crystals are sensitive to improper capacitive load https://www.bigmessowires.com/2026/08/30/adve … cuit-debugging/ but since this was designed without any it does mean those are build into driver in the PLL chip

analog_programmer wrote on Yesterday, 14:28:

This card could be not so bad

after seeing 1MB/s Im not so sure anymore 😀

analog_programmer wrote on Yesterday, 14:28:

if it had no VESA BIOS extensions implemented in its BIOS. Unfortunately it doesn't and I'm not skilled enough to inject such things into videocard BIOSes

we got robots for that now, but I dont see a point, I never saw anything wrong with vbe TSR, games that take advantage already dont care about base ram

rasz_pl wrote on Yesterday, 18:01:

analog_programmer wrote on Yesterday, 14:28:

This card could be not so bad

after seeing 1MB/s Im not so sure anymore 😀

Back n the days I never tried to use this:

analog_programmer wrote on Yesterday, 15:03:

P.S. But there are also this lines in the READ.ME file:

RAMBIOS device driver: […]
Show full quote

RAMBIOS device driver:

- RAMBIOS SYS ...a program to speed up video read and write operation.

Put DEVICE=RAMBIOS.SYS as the first device driver command in CONFIG.SYS
file. It should be installed before any other device driver in order to be
compatible with most application programs.

- RAMBIOS EXE ...a program for AUTOEXEC.BAT or under dos prompt.

rasz_pl wrote on Yesterday, 18:01:

analog_programmer wrote on Yesterday, 14:28:

if it had no VESA BIOS extensions implemented in its BIOS. Unfortunately it doesn't and I'm not skilled enough to inject such things into videocard BIOSes

we got robots for that now, but I dont see a point, I never saw anything wrong with vbe TSR, games that take advantage already dont care about base ram

I'm not sure if I used the VBE TSR either.

I don't use "ai" chatbots for "coding", debugging or patching. The free ones are useless and the paid ones (somehow useful) became too expensive to be used for such stupid things like ancient VGS BIOS patching.

You're just trying to stop me from wasting my time with the slow card's BIOS, but you haven't commented my other even more stupid idea for 1 MB VRAM upgrade 😁

The card being slow may well be due to it being ISA bus with waitstates.

The OEM was clearly scraping the barrel by using defect ram, so the implementation of waitstates and such, makes sense.

No idea what the speedup driver 'does'.

Well, before I jump straight into a BIOS modifying attempts, it will be better to set my UM85C408AF card with these drivers and TSRs to check out what exactly gives each of them.

With a Pentium 75 I got for Doom benchmark:

fast option: 6647 realtics
slow was like 2200 IIRC

For 80, 70 and 60 ns what MHz selection would be better? These tests were with 32MHz Mclk.

wierd_w wrote on Yesterday, 18:42:

The card being slow may well be due to it being ISA bus with waitstates.

The OEM was clearly scraping the barrel by using defect ram, so the implementation of waitstates and such, makes sense.

No idea what the speedup driver 'does'.

http://vgamuseum.info/index.php/benchmarks

There is worse than this one. I'm somewhat hopeful that our Programmer could up it a bit.
My mistake in 2022 turned out to be not a bad thing after all.
After decades this card could get what it deserves. Crazy 😀

Nexxen wrote on Yesterday, 22:56:

fast option: 6647 realtics
slow was like 2200 IIRC

that is opposite? 2200 is very fast 35fps, would put it faster than ET4000AX and WD90C31 https://thandor.net/benchmarks/vga/vga?a=69&c=0&o=0&s=Submit maybe overclocked ISA?
and 22000 would be 3 fps 😀

Nexxen wrote on Yesterday, 22:56:

For 80, 70 and 60 ns what MHz selection would be better? These tests were with 32MHz Mclk.

faastest UM85C408AF go all the way to 70MHz 😀

Nexxen wrote on Yesterday, 22:56:

With a Pentium 75 I got for Doom benchmark: […]
Show full quote

With a Pentium 75 I got for Doom benchmark:

fast option: 6647 realtics
slow was like 2200 IIRC

For 80, 70 and 60 ns what MHz selection would be better? These tests were with 32MHz Mclk.

6647 realtics ~ 11 FPS
2200 realtics ~ 33 FPS

By "fast option" do you mean, that you used the RAMBIOS.SYS/RAMBIOS.EXE driver? And why "fast" is three times slower than "slow"? Did you swapped the results?

So, it seems like we have a hidden ISA VGA "gem" here and its original cr*ppy BIOS is the main problem.

And here is our fast BIOS candidate I quickly made for less than 5 minutes:

After all, I have to purchase some "new" chinese EEPROMs for the experiments with this card. F*ck ursulians!

P.S. Next experiment should be to add VBE 1.2 or 2.0 into this fast BIOS. I think, a simple VESA.COM conversion into OpROM will not work, 'cause it stays resident in the RAM, not in the OpROM or shadowed memory parts and this makes such direct conversion pointless. The VBE table has to be "glued" into the BIOS.

rasz_pl wrote on Today, 03:53:

faastest UM85C408AF go all the way to 70MHz 😀

This explains "hardwired" JP3.

I saw "DRAM: 1M" string in the BIOS, so my idea for VRAM upgrade by combining two 4-bit data lines into one 8-bit line connected to one 512x8 FPM DRAM chip is still valid.

I think I made some mistake with benchmarks...
Let's redo them. I use Phil's pack.

4x70ns @75 MHz Pentium
1) 33.3
2)32.6
3) bench score 40.5 = 24.3FPS
4) VESA not found (640x480)
5) 15.4
6) xx
a) Doom minimum details = 1286 realtics
b) max details (full screen) 6646

@166MHz
1) 43.4
2)42.7
3) 50.2 = 30.1fps
5)22.4
a) 1091
b) 5878

Doom minimum is a tiny window, max is all screen with no hud.

With RAMBIOS.SYS loaded in CONFIG.SYS ("DEVICE=RAMBIOS.SYS" as first line) I got doubled results from CheckIt! 3.0 for "BIOS VIDEO CPS" compared to the test with no RAMBIOS.SYS loaded. Landmark's SST 2.0 gives me some insane boost of 4+ times, but SST 6.0 does not make any difference. Very weird results.

RAMBIOS.EXE executed from AUTOEXEC.BAT gives almost no difference.

Nexxen, did you tried the BIOS I prepared?

By the way, your SST6 result on 486DX is inferior to mine on 286 12 MHz (~1550 to ~1570 chr/ms depending on if RAMBIOS.SYS is loaded or not), which is also strange.

analog_programmer wrote on Today, 10:00:

With RAMBIOS.SYS loaded in CONFIG.SYS ("DEVICE=RAMBIOS.SYS" as first line) I got doubled results from CheckIt! 3.0 for "BIOS VID […]
Show full quote

With RAMBIOS.SYS loaded in CONFIG.SYS ("DEVICE=RAMBIOS.SYS" as first line) I got doubled results from CheckIt! 3.0 for "BIOS VIDEO CPS" compared to the test with no RAMBIOS.SYS loaded. Landmark's SST 2.0 gives me some insane boost of 4+ times, but SST 6.0 does not make any difference. Very weird results.

RAMBIOS.EXE executed from AUTOEXEC.BAT gives almost no difference.

Nexxen, did you tried the BIOS I prepared?

By the way, your SST6 result on 486DX is inferior to mine on 286 12 MHz (~1550 to ~1570 chr/ms depending on if RAMBIOS.SYS is loaded or not), which is also strange.

Not yet, I'll test this late afternoon.
I'll add the line in Config.sys and redo the tests.
I'm using a Pentium 166 @75 + Phil's pack, I don't know how optimized it is.
The board is a https://theretroweb.com/motherboards/s/qdi-p5 … -explorer-ii-sp

Nexxen wrote on 34 minutes ago:

Not yet, I'll test this late afternoon.

I'm waiting to see, if there will be some real speed improvements compared to the stock BIOS. Thanks for your detailed test reports!

Nexxen wrote on 34 minutes ago:

I'll add the line in Config.sys and redo the tests.
I'm using a Pentium 166 @75 + Phil's pack, I don't know how optimized it is.
The board is a https://theretroweb.com/motherboards/s/qdi-p5 … -explorer-ii-sp

Perhaps some motherboard's BIOSes difference in wait state setting for the ISA slots could explain the differences in SST6 results, despite the different platforms (Pentium - 486 - 286).

Now I will test VESA.COM to see what it will give or take away.

analog_programmer wrote on 20 minutes ago:

I'm waiting to see, if there will be some real speed improvements compared to the stock BIOS. Thanks for your detailed test repo […]
Show full quote

Nexxen wrote on 34 minutes ago:

Not yet, I'll test this late afternoon.

I'm waiting to see, if there will be some real speed improvements compared to the stock BIOS. Thanks for your detailed test reports!

Nexxen wrote on 34 minutes ago:

I'll add the line in Config.sys and redo the tests.
I'm using a Pentium 166 @75 + Phil's pack, I don't know how optimized it is.
The board is a https://theretroweb.com/motherboards/s/qdi-p5 … -explorer-ii-sp

Perhaps some motherboard's BIOSes difference in wait state setting for the ISA slots could explain the differences in SST6 results, despite the different platforms (Pentium - 486 - 286).

Now I will test VESA.COM to see what it will give or take away.

Shoot, I didn't think of modifying the ISA settings in BIOS......
Shame on me! 🤣