CkRtech wrote on 2026-02-28, 20:14:

I am up to doing this as I am still messing with this video card if you have the interest and the time. […]
Show full quote

Mumak wrote on 2025-11-09, 18:25:

Well, I guess I will have to get such a card and run tests...
The only alternate way would be to create a special HWiNFO version which would require a key press prior to each VGA chip test (HWiNFO displays at the bottom of screen which chip is tested) so you could capture the moment when it goes wrong..

I am up to doing this as I am still messing with this video card if you have the interest and the time.

The last line in the debug file was:
Checking: AT&T DEB

I know that it could be we were several steps ahead of the last item logged at the time the error occurred, but stopping before each test with a list of which test is about to be active would hopefully help us narrow things down.

OK, I have created a special build that requires a key-press after each status row displayed by HWiNFO. This should help to pin-point the exact place where it goes wrong.
Please try this build and let me know what was the last message displayed before the display went off.

Ok. Launched. Pressed spacebar through some init. Selected video info.

Saw “checking matrox…” and pressed space.
Saw “checking S-MOS,” pressed space, and then… screen corrupted.

CkRtech wrote on 2026-02-28, 22:50:

Ok. Launched. Pressed spacebar through some init. Selected video info.

Saw “checking matrox…” and pressed space.
Saw “checking S-MOS,” pressed space, and then… screen corrupted.

Thank you!
Please try the attached version and let me know if it fixes the problem.

Alas, it appears to still have the same problem.

OK, I have put again the key-pressing to check further. Please try this build and let me know.

OK. Now we are talking.

"Checking: Weitek" gave a momentary flash of just a few scanlines near the bottom quarter of the screen. Everything continued to work fine. Standard fare for when hardware is detected by software in my opinion.

And of course the Power 9000 chip on this Diamond Viper VLB is made by Weitek, so perhaps it managed to communicate with it.

I passed through a few other items until "Checking: Oak." I pressed spacebar after that and BOOM. Corruption.

Oak is, of course, the base VGA chip on the card used for handling text and low res modes. The P9000 is accessed for VESA modes.

CkRtech wrote on 2026-03-01, 17:59:

OK. Now we are talking. […]
Show full quote

OK. Now we are talking.

"Checking: Weitek" gave a momentary flash of just a few scanlines near the bottom quarter of the screen. Everything continued to work fine. Standard fare for when hardware is detected by software in my opinion.

And of course the Power 9000 chip on this Diamond Viper VLB is made by Weitek, so perhaps it managed to communicate with it.

I passed through a few other items until "Checking: Oak." I pressed spacebar after that and BOOM. Corruption.

Oak is, of course, the base VGA chip on the card used for handling text and low res modes. The P9000 is accessed for VESA modes.

Thanks. Please try the attached build and let me know.

Bingo.

No TSR:

With Diamond VESA driver loaded:

CkRtech wrote on 2026-03-01, 19:01:

Bingo. […]
Show full quote

Bingo.

No TSR:

The attachment IMG_6083.jpeg is no longer available

With Diamond VESA driver loaded:

The attachment IMG_6084.jpeg is no longer available

The attachment IMG_6085.jpeg is no longer available

Perfect 😀 Thanks for you help!
I will release a new version with this and a few other fixes.

Mumak wrote on 2026-03-01, 19:05:

Perfect 😀 Thanks for you help!
I will release a new version with this and a few other fixes.

Happy to help!

Version 6.2.7 released. Details and download in OP.

Hello,
I have PB450 motherboard with built-in video card and no PCI. I know there were both PCI and ISA (VLB?) versions.
My MB has ISA riser card option and looks like this:

It has VLBus chip built in, but it seems it might be used for IDE only, since it's too far from the video chip. So I'm not so sure whether the video is ISA or VLB. VRAM benchmark shows speeds very close ISA.
However, HWINFO screens show the opposite:

I wonder how HWINFO detects VLB?

radozd wrote on 2026-04-04, 09:22:

Hello, I have PB450 motherboard with built-in video card and no PCI. I know there were both PCI and ISA (VLB?) versions. My MB h […]
Show full quote

Hello,
I have PB450 motherboard with built-in video card and no PCI. I know there were both PCI and ISA (VLB?) versions.
My MB has ISA riser card option and looks like this:

The attachment pb450-small.jpeg is no longer available

It has VLBus chip built in, but it seems it might be used for IDE only, since it's too far from the video chip. So I'm not so sure whether the video is ISA or VLB. VRAM benchmark shows speeds very close ISA.
However, HWINFO screens show the opposite:

The attachment IMG_7592.jpeg is no longer available

The attachment IMG_7593.jpeg is no longer available

I wonder how HWINFO detects VLB?

For this graphics adapter (CL-GD5428) the bus is detected directly from the chip straps (System Bus Select CF[5], CF[7], CF[14]).

Mumak wrote on 2026-04-04, 11:40:

For this graphics adapter (CL-GD5428) the bus is detected directly from the chip straps (System Bus Select CF[5], CF[7], CF[14]).

Thanks. So I should trust HWINFO. Good!

radozd wrote on 2026-04-04, 15:06:

Thanks. So I should trust HWINFO. Good!

You can also check where the LFB is located - IBM PS/1 Integrated Graphics, how to tell if ISA or VLB?

Grzyb wrote on 2026-04-09, 03:39:

You can also check where the LFB is located - IBM PS/1 Integrated Graphics, how to tell if ISA or VLB?

How could I miss this thread?! Thanks a lot!

Under 16MB while using UNIVBE. Not sure if this rule is applicable here.

radozd wrote on 2026-04-09, 07:48:

Under 16MB while using UNIVBE. Not sure if this rule is applicable here.

Interesting...
LFB above 16 MB clearly states non-ISA, but below 16 MB can be anything.
Can you check where Windows 9x drivers place the LFB ?

Grzyb wrote on 2026-04-09, 10:40:

Interesting... LFB above 16 MB clearly states non-ISA, but below 16 MB can be anything. Can you check where Windows 9x drivers p […]
Show full quote

radozd wrote on 2026-04-09, 07:48:

Under 16MB while using UNIVBE. Not sure if this rule is applicable here.

Interesting...
LFB above 16 MB clearly states non-ISA, but below 16 MB can be anything.
Can you check where Windows 9x drivers place the LFB ?

MS-DOS 6.22 only, sorry

I was looking for ways to load ASPI8U2.SYS high, not specifically how to run HWiNFO and ended up here by accident. But I did run HWiNFO. Here's what I found.

The board is a Abit AB-SM5 (HWiNFO identifies this as "IT5V")

@feipoa Your photos of mem /c/p made me look into boot optimizations. Here's what I found.

Both ASPI8U2.SYS ASPICD.SYS are possible to load high.
Consider using SHSUCDX.COM instead of MSCDEX.EXE. It is smaller, open source, and can be loaded high with the /C option. It's BlueSCSI compatible where MSCDEX.EXE is not, which is why I use it.
https://github.com/adoxa/shsucd/blob/master/shsucdx.nsm

I also changed "EMM386.EXE RAM" to "EMM386.EXE NOEMS" and got a few more UMBs (which I believe are more useful). This has given me a new best "Largest executable program size" of 599K. Close to the 615K that was mentioned. No problem running HWiNFO though. Maybe the 615K was a "report included" specific limitation?

Maybe my configuration can spark some ideas. Some notes about it,
MSDOS bigger than feipoa's. I assume because of 7.1 from Windows 98 SE.
COMMAND mentioned twice (10K + 7K) so I am wondering why there is only one COMMAND in feipoa's configuration. Is there some further optimization hiding there?

Text version (I manually typed this in on a laptop).

1Modules using memory below 1 MB:
2  Name          Total           Conventional       Upper Memory
3  -------  ----------------   ----------------   ----------------
4  MSDOS      33,776   (33K)     33,776   (33K)          0    (0K)
5  HIMEM       1,168    (1K)      1,168    (1K)          0    (0K)
6  EMM386      4,320    (4K)      4,320    (4K)          0    (0K)
7  ASPI8U2    27,312   (27K)          0    (6K)     27,312   (27K)
8  ASPICD     12,784   (12K)          0    (6K)     12,784   (12K)
9  COMMAND    10,144   (10K)          0    (6K)     10,144   (10K)
10  CTMOUSE     3,328    (3K)          0    (6K)      3,328    (3K)
11  COMMAND     6,976    (7K)          0    (6K)      6,976    (7K)
12  SHSUCDX     5,872    (6K)          0    (6K)      5,872    (6K)
13  Free      771,056  (753K)    613,680  (599K)    157,376  (154K)
14
15Memory Sumary:
16
17  Type of Memroy       Total         Used          Free
18  ----------------  -----------   -----------   -----------
19  Conventional          653,312        39,632       613,680
20  Upper                 223,792        66,416       157,376
21  Reserved                    0             0             0
22  Extended (XMS)*    65,836,496       316,880    65,519,616
23  ----------------  -----------   -----------   -----------
24  Total memory       66,713,600       422,928    66,290,672
25
26  Total under 1 MB      877,104       106,048       771,056
27
28  Largest executable program size         613,664   (599K)
29  Largest free upper memory block         133,376   (130K)
30  MS-DOS is resident in the high memory area.