lolo799 wrote on 2021-09-07, 10:26:

With the Test-04 version

After pondering [country] I added some debug code. Using the attached SIV32A V5.61 TAPI-04 test SIV please do:

SIV32A -DBGTAPI -save=[country]=country > SIV_DBGOUT.log | more

and post the SIV_DBGOUT.log file once SIV has stalled or completed.

red-ray wrote on 2021-09-08, 07:48:

Thank you for running the test, but you did not do Menu->File->Save->Save Selected, deselect [Country] and then press [Start] as I asked, please try again.

I did try the Test-04 before you mention the "save selected" option, haven't turned on the PWS since.

I tried Test-04 on my PWS 600au and [country] worked so I am pondering why it stalled on your system. When it stalled what % CPU time was SIV32A using?

I'll check that next time.

SIV did not work as I hoped as think SIVA32.sys was deleted/went missing from C:\SIV32A\, what happened to it? To check all the files are present look at Menu->Help->SIV Lookup which should report similar to as below, further such as [PCI Bus] reported as follows rather than the PCI Device Information because SIV was unable to find it's driver.

I extracted the content of the .zip you posted in your 2nd post, I doubt I deleted SIV32A.sys but I'll recheck that too.

lolo799 wrote on 2021-09-08, 10:31:

I'll check that next time.

OK, I just uploaded SIV 5.61 Beta-04 so ideally use that for the tests please.

Beta04 results:

As you can see it has problems identifying the SIV32A.sys file:

Continued:

Got this error a few times while trying the save- selected menu and selecting some network related things on the last column:

Taskmanager when SIV freezes:

The logs, with [country] unchecked and the one with the debug output you asked for, you'll be disappointed by the latter...

lolo799 wrote on 2021-09-12, 11:06:

As you can see it has problems identifying the SIV32A.sys file:

Thank you for running the tests and posting the results.

1. For some strange/unknown reason the SIV driver does not start, but it did start and run OK in the files you posted on posted on 2021-09-03 @ 16:28. SIVA32.sys should be 174 KB (178,432 bytes), is it? Looking in the save file at Menu->Windows->Services->Control SIV I can see that the SIVDriver service Current State is Stopped when it should be Running, does it start if you press [Start]? It not then try pressing [Delete] and see if this helps (SIV should recreate the service). Deleting C:\WINNT\System32\Drivers\SIVA32.sys may also help get the service running.
2. The reason that [country] get's stuck seems to be that RasGetCountryInfoA() never returns. AFAIK Windows 2000 reads this information from HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Telephony\Country List, can REGEDIT read that part of the registry OK?
3. I suspect your system does not have the L3 cache module, is this the case? If so the attached SIV32A 5.61 Zero-05 should allow for this, SIV did not as I thought all PWS systems had a 2MB or 4MB L3 cache.
4. The error 2184 is down to your network not being correctly configured and you need to fix this. The error is returned by NetServerEnum() which is what the net view command also uses, GIYF.

red-ray wrote on 2021-09-12, 14:24:

Thank you for running the tests and posting the results. […]
Show full quote

lolo799 wrote on 2021-09-12, 11:06:

As you can see it has problems identifying the SIV32A.sys file:

Thank you for running the tests and posting the results.

1. For some strange/unknown reason the SIV driver does not start, but it did start and run OK in the files you posted on posted on 2021-09-03 @ 16:28. SIVA32.sys should be 174 KB (178,432 bytes), is it? Looking in the save file at Menu->Windows->Services->Control SIV I can see that the SIVDriver service Current State is Stopped when it should be Running, does it start if you press [Start]? It not then try pressing [Delete] and see if this helps (SIV should recreate the service). Deleting C:\WINNT\System32\Drivers\SIVA32.sys may also help get the service running.
2. The reason that [country] get's stuck seems to be that RasGetCountryInfoA() never returns. AFAIK Windows 2000 reads this information from HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Telephony\Country List, can REGEDIT read that part of the registry OK?
3. I suspect your system does not have the L3 cache module, is this the case? If so the attached SIV32A 5.61 Zero-05 should allow for this, SIV did not as I thought all PWS systems had a 2MB or 4MB L3 cache.
4. The error 2184 is down to your network not being correctly configured and you need to fix this. The error is returned by NetServerEnum() which is what the net view command also uses, GIYF.

1. Pressing start didn't do anything, pressing delete followed by start gave me an SIV32A internal error panel.
   Deleting the file by hand and relaunching SIV32A didn't make it start the service either.
2. No problems browsing through that part of the registry in regedit, 33 is there.
3. right, there is no L3 cache module on my machine
4. is trying to fix it easier reinstalling Windows 2000, I'll decide tomorrow evening

New SIV32A log included, country unchecked.
And for fun tried SIV32L:

Got an exception error message from Windows that resulted in a debug log, forgot to copy it, I can provide it tomorrow if you need it.

lolo799 wrote on 2021-09-12, 17:10:

[…]
Show full quote

1. Pressing start didn't do anything, pressing delete followed by start gave me an SIV32A internal error panel. Deleting the file by hand and relaunching SIV32A didn't make it start the service either.
2. No problems browsing through that part of the registry in regedit, 33 is there.
3. right, there is no L3 cache module on my machine
4. is trying to fix it easier reinstalling Windows 2000, I'll decide tomorrow evening
5. And for fun tried SIV32L got an exception error message from Windows that resulted in a debug log, forgot to copy it, I can provide it tomorrow if you need it.

Thank you for checking things out.

1. Until the driver starts playing ball SIV will not work as it should. Try exiting SIV then from a command window try sc start SIVDriver, does it start and if not what error get's reported? How about sc delete SIVDriver?
2. I wonder if this is down to the date being 1995. At a guess the CR3032 needs replacing, does it?
3. The Zero-05 SIV worked as I hoped it would and no L3 cache was reported.
4. I suspect doing a clean install would be sensible. AFAIK the RC1 + RC2 releases time out after a while (444 days I think) so when I installed back in 2009 I set the year to 2059, did the install and reverted. Thus far it has not timed out.
5. The issue is with Menu->Help->SIV Manifest and I see the same on my PWS 600au system, but all is OK on this W10 system and P-II NT4 system. I will try and fix this, but the x32 remote debugger is very slow and a PITA.

red-ray wrote on 2021-09-12, 18:51:

Thank you for checking things out. […]
Show full quote

Thank you for checking things out.

1. Until the driver starts playing ball SIV will not work as it should. Try exiting SIV then from a command window try sc start SIVDriver, does it start and if not what error get's reported? How about sc delete SIVDriver?
2. I wonder if this is down to the date being 1995. At a guess the CR3032 needs replacing, does it?
3. The Zero-05 SIV worked as I hoped it would and no L3 cache was reported.
4. I suspect doing a clean install would be sensible. AFAIK the RC1 + RC2 releases time out after a while (444 days I think) so then I installed back in 2009 I set the year to 2059, did the install and reverted. Thus far it has not timed out.
5. The issue is with Menu->Help->SIV Manifest and I see the same on my PWS 600au system, but all is OK on this W10 system and P-II NT4 system. I will try and fix this, but the x32 remote debugger is very slow and a PITA.

I'll try starting the service from the command line tomorrow.
The battery is new-ish, but I installed Windows without changing the default date in AlphaBIOS first...I did set today's date earlier and I got a prompt to let me know the system would shut down after an hour, so I reverted back to 1995, stuff to take care before reinstalling.
And trying again to install NT4....

Thanks for your program and your quick updates!

lolo799 wrote on 2021-09-12, 19:59:

I'll try starting the service from the command line tomorrow.

OK, I have just uploaded SIV 5.61 Beta-05 which should fix the SIV32L Menu->Help->SIV Manifest issue, does it? Doing [Save Local] worked OK on my PWS 600au, but took much longer than SIV32A and reported less information as several x32 system .DLLs are absent.

I feel it would be sensible to check the event logs to see if there are any events logged about SIVDriver or RAS/RasGetCountryInfo().

I wonder if the strange effects are down to your system only having 128MB of memory. I upgraded my system to have 6 x 256MB MT 18LSDT3272DG-10EB1 which work well.

Reinstalled Win2k, the network works again!

The [manifest] issue is solved.

red-ray wrote on 2021-09-13, 07:48:

I wonder if the strange effects are down to your system only having 128MB of memory. I upgraded my system to have 6 x 256MB MT 18LSDT3272DG-10EB1 which work well.

It had no problems at first, just when the network stopped working.
And about the memory, I'm going to get rid of the miata, not going to upgrade it to the max.

lolo799 wrote on 2021-09-14, 20:05:

Reinstalled Win2k, the network works again!

Thank you, all the strange effects have now gone away 😀 I guess we will never know their root cause and I feel they must have been triggered by the previous installation going wonky.

I spotted that SIV32A failed to report the Matrox MGA-1164SG Mystique RAMDAC Clock and think I have fixed this in SIV32A V5.61 Beta-06, have I?

The display was set to 800 x 600 @ 60Hz, but looking at Menu->Machine->Graphics it looks like it can manage 1280 x 1024 @ 60Hz, can it?

When you generate the new save files may I have both SIV_PWS433.dmi (binary dump of what [Machine] reports) + SIV_PWS433.txt as SIV works better in test mode when I have both.

dormcat wrote on 2021-09-15, 17:09:

My Gainward CARDEXpert SG4 Savage4 Pro PCI has a similar problem: while the sticker says "SG4 Pro" (86c397), software like Everest or CPU-Z identify it as either non-Pro (86c395) or "Savage4V," a model that I can't find on most websites. Memory clock is even slower at just 100 MHz.

AFAIK a 86c395 is a S3 Savage4 GT and a 86c397 a S3 Savage4 Pro, but I have no idea if Everest decodes the chip type correctly.

I suspect SIV will do better, please try it and post the Menu->Machine->GPU Info information, better still do Menu->File->Save Local and post the save file.

Below is what SIV reported for my S3 Savage4 Pro-M (86C396), notice how the screen shows a dump of the CRnn registers that need to be used to know the chip type.

red-ray wrote on 2021-09-14, 23:08:

Thank you, all the strange effects have now gone away :) I guess we will never know their root cause and I feel they must have […]
Show full quote

Thank you, all the strange effects have now gone away 😀 I guess we will never know their root cause and I feel they must have been triggered by the previous installation going wonky.

I spotted that SIV32A failed to report the Matrox MGA-1164SG Mystique RAMDAC Clock and think I have fixed this in SIV32A V5.61 Beta-06, have I?

The display was set to 800 x 600 @ 60Hz, but looking at Menu->Machine->Graphics it looks like it can manage 1280 x 1024 @ 60Hz, can it?

When you generate the new save files may I have both SIV_PWS433.dmi (binary dump of what [Machine] reports) + SIV_PWS433.txt as SIV works better in test mode when I have both.

1280*1024 works fine, yes, although the first time I ran SIV32A 5.61Beta06 file>save local after taking that screenshot it crashed at [cpuid], which was the last thing written in the report log, its size was about 75k.
I attached the drwatson log from the event.

SIV32A also reports the card as MGA-1164SG Mystique for some reason...

lolo799 wrote on 2021-09-19, 09:22:

1280*1024 works fine, yes, although the first time I ran SIV32A 5.61Beta06 file>save local after taking that screenshot it crashed at [cpuid]

SIV32A also reports the card as MGA-1164SG Mystique for some reason...

Thank you for the report, it crashed doing a divide by zero processing the [cpu-levels] section. I will try and make it happen on my PWS 600au and get back to you. I suspect the trigger was that SIV had failed to figure out the x6.5 CPU multiplier.

The text in PCIDEVS.txt was not ideal, the attached should get the name correct, does it please? Both the MGA-1064SG + MGA-1164SG have the same PCI ID of 102B:051A and you need to check the revision to know which is which.

red-ray wrote on 2021-09-19, 11:01:

The text in PCIDEVS.txt was not ideal, the attached should get the name correct, does it please? Both the MGA-1064SG + MGA-1164SG have the same PCI ID of 102B:051A and you need to check the revision to know which is which.

The GPU core clock shows the MGA-1064SG, didn't have time to test with the new file.
I also have a 1164 Mystique that came with my BeBox, i'll put it in the Alpha to check with both old and new PCIDEVS.

lolo799 wrote on 2021-09-19, 11:35:

I also have a 1164 Mystique that came with my BeBox, i'll put it in the Alpha to check with both old and new PCIDEVS.

OK, I would be inclined to install both in Alpha.

I have found the issue with the divide by zero crash and should have fixed it in the attached SIV32A V5.61 Zero-07 test SIV.

red-ray wrote on 2021-09-19, 08:25:

AFAIK a 86c395 is a S3 Savage4 GT and a 86c397 a S3 Savage4 Pro, but I have no idea if Everest decodes the chip type correctly. […]
Show full quote

dormcat wrote on 2021-09-15, 17:09:

My Gainward CARDEXpert SG4 Savage4 Pro PCI has a similar problem: while the sticker says "SG4 Pro" (86c397), software like Everest or CPU-Z identify it as either non-Pro (86c395) or "Savage4V," a model that I can't find on most websites. Memory clock is even slower at just 100 MHz.

AFAIK a 86c395 is a S3 Savage4 GT and a 86c397 a S3 Savage4 Pro, but I have no idea if Everest decodes the chip type correctly.

I suspect SIV will do better, please try it and post the Menu->Machine->GPU Info information, better still do Menu->File->Save Local and post the save file.

Below is what SIV reported for my S3 Savage4 Pro-M (86C396), notice how the screen shows a dump of the CRnn registers that need to be used to know the chip type.

SIV reported my video card as S3 Savage4 Pro-M (86C396); on the other hand, Everest reported it as Savage4 GT (86C395).

Another discrepancy is RAMDAC clock: SIV reported 162 MHz but Everest reported 300 MHz. The memory clock is the same slow 100 MHz, though.

Different versions of CPU-Z (the last "native Win9x" 1.57, the "return of Win9x support" 1.78, and the Vintage Edition 1.03), however, gave little useful information, as "Savage4V" doesn't seem to be a valid model name.

dormcat wrote on 2021-09-20, 02:57:

[…]
Show full quote

1. SIV reported my video card as S3 Savage4 Pro-M (86C396); on the other hand, Everest reported it as Savage4 GT (86C395).
2. Another discrepancy is RAMDAC clock: SIV reported 162 MHz but Everest reported 300 MHz. The memory clock is the same slow 100 MHz, though.
3. Different versions of CPU-Z (the last "native Win9x" 1.57, the "return of Win9x support" 1.78, and the Vintage Edition 1.03), however, gave little useful information, as "Savage4V" doesn't seem to be a valid model name.

Thank you for checking out SIV.

1. Given 86C395 is a S3 Savage4 GT and the sticker on the card says SG4 Pro I expect Everest is incorrect. We really need to see what it says on the chip, but as there is a heatsink in the way we can't, what does HWiNFO32 report and what do you think it should be?
2. SIV is reporting the current RAMDAC speed whereas Everest reports the maximum. If you change the resolution the speed reported by SIV should change, does it?
3. You need to report CPUZ issues to the CPUZ thread. AFAIK "native Win9x 1.57" is deprecated and been replaced by "Vintage Edition 1.03".
4. Below you you should see that both SIV + HWiNFO32 correctly report my Savage4 as 86C396 which is why I feel AIDA64 (Everest) + the CPUZs need fixing. I have found that in general the CPUZ reporting of GPUs is limited.

red-ray wrote on 2021-09-20, 06:30:

Thank you for checking out SIV. […]
Show full quote

Thank you for checking out SIV.

1. Given 86C395 is a S3 Savage4 GT and the sticker on the card says SG4 Pro I expect Everest is incorrect. We really need to see what it says on the chip, but as there is a heatsink in the way we can't, what does HWiNFO32 report and what do you think it should be?
2. SIV is reporting the current RAMDAC speed whereas Everest reports the maximum. If you change the resolution the speed reported by SIV should change, does it?
3. You need to report CPUZ issues to the CPUZ thread. AFAIK "native Win9x 1.57" is deprecated and been replaced by "Vintage Edition 1.03".
4. Below you you should see that both SIV + HWiNFO32 correctly report my Savage4 as 86C396 which is why I feel AIDA64 (Everest) + the CPUZs need fixing. I have found that in general the CPUZ reporting of GPUs is limited.

Thanks for the tip. However, HWiNFO reported conflicting model on the same "video adapter" tab:

IMHO it is no longer important whether it be 86C395 or 86C396 at this time as I only play DOS games on that build; the only way to verify it would be removing the heatsink, but it is glued on the chip really hard and the only heat gun I have is my hair dryer.

FYI: SIV32L crashes on my recent acquisition and I haven't been able to find a solution.