BigDaddyM wrote:

Is it possible that SIV calculates the chip's clocks incorrectly? I think that RAMDAC should be 230 MHz. 86x362 and 86x368 are practically the same except for the amount of supported memory and its types. I will try to find some cards today.

Most things are possible, but to sensibly comment I would need to see a screen shot.

For some GPUs SIV reports including the S3 Trio3D/2X the current RAMDAC speed which will changes as the resolution changes,

I already have everything for you. I will put everything I have in one archive, including card photos and benchmarks

debs3759 wrote:

Do you need the app testing on VIA Eden, C7 and Luke CoreFusion? I have several ITX boards with BGA chips that I plan to test over the next couple of weeks, so it would be the ideal time to test SIV and CPU-Z if either needs it.

Also is the list on the first post up to date? I am sure I can test some of the early GPUs you list, if still needed.

Thank you and sorry for the delay in replying, I only just noticed your post. Testing both the C7 and Luke would be good as I don't seem to have SIV save files from either.

In general I have tried to keep the initial post up-to-date so if you have a GPU and SIV looks to report incorrect information please let me know. Ideally use the attached SIV32L that is currently the V5.44 release and if both NT/2K/XP and W9x are bootable then use NT/2K/XP as this should in general provide better support than W9x.

Message sent

BigDaddyM wrote on 2019-12-22, 15:13:

Message sent

Thank you for all your help. I have just uploaded SIV32L 5.45 Beta-02 which should fix the system memory reporting for your VIA and a few other VIA chipsets.

Finally getting back to this! I've been doing some testing with my (upgraded 😉 ) K6-2+ system. Just tried out the newest 5.45 Beta2. It seems to do pretty well, but I noticed a couple minor things are getting misreported:

The new CPU is a K6-2+/550 ACZ (link). SIV reports it in most places as a K6-III (Sharptooth). There are a couple of lines where it shows K6-2+ in the saved report. It does get the L2 size right at 128kB (a K6-III would have 256kB.)

(Note: CPU-Z seems to share this behavior, it shows a 2+ on the main screen but as a K6-III on the benchmark screen.)

I've got it running at 5.5x105=577.5 MHz, a mild FSB overclock. SIV seems to be interpreting this as 7x82.5 instead (and also shows the RAM speed as 82.5MHz because of this.) It does report the RAM TIMINGS correctly, at 2-2-2-6-9 (agreeing with CPU-Z.)

Also, I don't see any indication of the L3 (motherboard) cache. Does SIV check this?

This system has my other Voodoo3 (a legit 3000 running at 166MHz) in it, and everything seems to be OK there.

The CPU temperature is fine on the main screen. In the Status window, 'T2 System' and 'T1 CPU' seem to be reporting OK, but there's a 'T3 AUX' that shows -48 C. I don't know what this sensor is, to be honest. AFAIK there's no actual temperature sensor in this CPU so I presume 'T1 CPU' is taken from the sensor in the socket?

Voltages, etc. are all being reported correctly as far as I can tell.

Report attached. 😉

Thank you for checking out SIV and posting the save file.

xjas wrote on 2020-01-12, 10:50:

The new CPU is a K6-2+/550 ACZ (link). SIV reports it in most places as a K6-III (Sharptooth). There are a couple of lines where it shows K6-2+ in the saved report. It does get the L2 size right at 128kB (a K6-III would have 256kB.)

AFAIK if 256KB then AMD-K6-III+ else if 128KB then AMD-K6-III else AMD-K6-2+ and given the CPU reports itself as "AMD-K6(tm)-III Processor" this is also why SIV reports "AMD-K6-III (Sharptooth)". When SIV reports "K6-2+" this is down to the data SIV is reporting from not reporting the L2 cache size.

It would be easy enough to change the code, but before doing this I need to convince myself what should be reported. Do you have an AMD document that specifies this?

xjas wrote on 2020-01-12, 10:50:

I've got it running at 5.5x105=577.5 MHz, a mild FSB overclock. SIV seems to be interpreting this as 7x82.5 instead (and also shows the RAM speed as 82.5MHz because of this.) It does report the RAM TIMINGS correctly, at 2-2-2-6-9 (agreeing with CPU-Z.)

This is down to SIV not reading MSRs on W9x and on NT/XP it should report correctly, does it? Either way 5.45 Beta-03 should do better on W9x.

xjas wrote on 2020-01-12, 10:50:

Also, I don't see any indication of the L3 (motherboard) cache. Does SIV check this?

SIV did not have the concept of L3 Motherboard Cache so I have added this for 5.45 Beta-03.

xjas wrote on 2020-01-12, 10:50:

The CPU temperature is fine on the main screen. In the Status window, 'T2 System' and 'T1 CPU' seem to be reporting OK, but there's a 'T3 AUX' that shows -48 C. I don't know what this sensor is, to be honest. AFAIK there's no actual temperature sensor in this CPU so I presume 'T1 CPU' is taken from the sensor in the socket?

As SIV does not know the motherboard model it fails to suppress T3 and on NT/XP it should not be there. Is the motherboard maker and model within the W9x registry and if so where please?

Cool, thanks for the fast update! I'll try out the new version and post back.

I don't have time to check everything right now, but on the cache sizes:

red-ray wrote on 2020-01-12, 12:08:

xjas wrote on 2020-01-12, 10:50:

The new CPU is a K6-2+/550 ACZ (link). SIV reports it in most places as a K6-III (Sharptooth). There are a couple of lines where it shows K6-2+ in the saved report. It does get the L2 size right at 128kB (a K6-III would have 256kB.)

AFAIK if 256KB then AMD-K6-III+ else if 128KB then AMD-K6-III else AMD-K6-2+ and given the CPU reports itself as "AMD-K6(tm)-III Processor" this is also why SIV reports "AMD-K6-III (Sharptooth)". When SIV reports "K6-2+" this is down to the data SIV is reporting from not reporting the L2 cache size.

It would be easy enough to change the code, but before doing this I need to convince myself what should be reported. Do you have an AMD document that specifies this?

Pretty sure the K6-III/III+ models universally have 256kB L2, and only the 2+/2E+ have 128kB. All the datasheets can be found here; I didn't look at every single one but at least the 'standard' (original) K6-III is definitely a 256kB chip.

The motherboard is an ASUS P5A-B; CPU-Z can identify it so that ID must exist somewhere... Should I just search the registry for a text string and see where it's located, or is there some ID code you'd need instead?

xjas wrote on 2020-01-12, 21:42:

The motherboard is an ASUS P5A-B; CPU-Z can identify it so that ID must exist somewhere... Should I just search the registry for a text string and see where it's located, or is there some ID code you'd need instead?

Thank you for the update. AFAIK CPUZ reads the ASUS P5A-B from the SMBIOS and SIV will do the same on NT/XP, but as the initial post says the W9x support is limited. If SIV can read this from the registry I will get it to do this, but I have no plans to get SIV to read the SMBIOS W9x.

On Windows 7 this information is in HKLM\Hardware\Description\System\BIOS, but I don't know if W9x loads the information at all.

xjas wrote on 2020-01-12, 21:42:

red-ray wrote on 2020-01-12, 12:08:

The CPU reports itself as "AMD-K6(tm)-III Processor" this is also why SIV reports "AMD-K6-III (Sharptooth)".

I didn't look at every single one but at least the 'standard' (original) K6-III is definitely a 256kB chip.

The AMD-K6(tm)-III Processor is down to what the CPU reports and I am unsure what to do about this. I wonder if it's down to a BIOS bug and if there is an updated BIOS that fixes this, what do you think please?

Update: didn't find any way to read the board type from the registry. (I don't have the "Bios" entry at all...) There's one reference to "ASUS" in HKLM\Enum\Root\*PNP0C08\0000 but I don't think that can be relied upon.

Regarding the K6-2+/III issue, I'm not sure what can be done about this. I have the latest BIOS which specifically supports it, and it correctly identifies on POST as a K6-2+, but in the registry (in HKLM\Hardware\Description\System\CentralProcessor\0\ProcessorNameString) it's listed as a K6-III. Possibly Windows was never updated to support the + chips? To be honest, I think you know far more about the ins and outs of how Windows IDs hardware than I do.

BTW is 5.45 Beta-3 out, or did I misinterpret your post? The link on page one still leads to Beta2.

Incidentally, I'm actually looking for one of those swapable IDE bays for this machine so I can experiment with other OSes on it (including Win2K, which would give you some better results.)

xjas wrote on 2020-01-14, 08:06:

Update: didn't find any way to read the board type from the registry.

Incidentally, I'm actually looking for one of those swapable IDE bays for this machine so I can experiment with other OSes on it (including Win2K, which would give you some better results.)

OK, I suspected the board type would not be in the registry and was waiting you to confirm this before I uploaded Beta-03; It's now there.

On my systems I usually set then up as multi-boot with a single disk typically containing:

• C: (FAT16) 2GB Windows 9x/Me
• D: (FAT16) 2GB Installation Kits
• E: (FAT16) 2GB Windows NT 4.0
• F: (FAT16) 2GB Windows 2K
• G: (NTFS) 8GB Windows XP
• H: (NTFS) 16GB Windows 2003

red-ray wrote on 2020-01-13, 13:05:

The CPU reports itself as "AMD-K6(tm)-III Processor" this is also why SIV reports "AMD-K6-III (Sharptooth)".
I didn't look at every single one but at least the 'standard' (original) K6-III is definitely a 256kB chip.

The AMD-K6(tm)-III Processor is down to what the CPU reports and I am unsure what to do about this. I wonder if it's down to a BIOS bug and if there is an updated BIOS that fixes this, what do you think please?

For CPUID 05Dx
if x = 0-3, it's a K6-III+
if x = 4-7, it's a K6-2+

For CPUID 059x, I have no data on how to tell them apart, and I haven't worked on my CPUID code in years. Still have all the data I had back then though.

I hope to get round to the testing I planned for Christmas week soon. Had to turn everything off to conserve power for a few weeks due to overspending 😀

debs3759 wrote on 2020-01-15, 01:18:

For CPUID 05Dx
if x = 0-3, it's a K6-III+
if x = 4-7, it's a K6-2+

Thank you, I was not aware the K6 designation could be derived from the stepping. When I checked the SIV save files I have it seems to work, but noted the CPU reports itself as a K6-III rather than a K6-III+.

I know it's not old, but I have just released SIV 5.45 Beta-05 that adds support for Intel Ice Lake-U and it would be good to get reports from some. Thus far I have only tested it on an Intel i7 Quad 1065G7 (Ice Lake-U)

For SIV 5.45 Beta-10 I added GeForce 6000 series direct clock + fan PWM % and NVCPL temperature reporting for ForceWare drivers. The ForceWare drivers don't support NVAPI and are typically found on Pentium M laptops.

I would be particularly interested to see what get's reported for a GeForce 6000 with two fans.

For SIV 5.45 Beta-12 I added XGI Volari GPU clock speed reporting. If you have an XGI Volari GPU please post a screen shot similar to as below so I can check how well the new code does.

I also fixed a few small glitches in the DMP + RDC CPU support. You need to SIV 5.45 Beta-12 or later which is attached to SIV support for 386/486/586 class + Alpha CPUs and 3dfx + S3 + SiS + Matrox + XGI + old ATI + NVidia GPUs - Testing Help and things should work on all of Windows NT/2K/XP/9x and later.

I did not know SIV - System Information Viewer before I found this forum, it seems to be good, tons of information available. Must try it out on a normal hardware to see what it can do.

But first 2 screen shots from my antique Micronics ASIC Eisa 486 computer with (newly setup) Windows 98SE. I just tried hwinfo before, so now SIV, will also try cpu-z 😀
The mainboard only has EISA slots, so only ISA and EISA cards are possible in this computer, I don't know if that is interesting for you, it is likely relevant only for 0.0001% of the users.

I ran the siv32l.exe v5.44, works fine, no crashes.

The gpu detection is not too exciting, what's nice is that all video modes possible of the ATI Graphics Ultra EISA are reported.

For the CPU the external bus is reported with 28MHz, but it should be 33MHz, so also the cpu speed is wrong only 4 times 28MHz. The external cache of 256kB on the mainboard is not detected.
Also no harddisks are detected, most likely because of the EISA bus.

CoffeeOne wrote on 2020-02-10, 19:37:

The gpu detection is not too exciting, what's nice is that all video modes possible of the ATI Graphics Ultra EISA are reported.
For the CPU the external bus is reported with 28MHz, but it should be 33MHz, so also the cpu speed is wrong only 4 times 28MHz. The external cache of 256kB on the mainboard is not detected.
Also no harddisks are detected, most likely because of the EISA bus.

Thank you for trying out SIV and given I don't think SIV has ever been run on an EISA system then it did OK, at least it did not crash.

In general SIV uses the PCI Device ID for GPU reporting which will not be present, that said it does report my Kelvin 64 ISA GPU so I suspect I will be able to improve what it reports for your GPU. To enable me to do this do Menu->File->Save Local and post the generated save file.

The 28MHz is down to the CPU speed being measured as 112 MHz, so I suspect I need to adjust this. Please will you post Menu->Help->Conversions->CPU MHz so I can check.

SIV does not report Disk SMART on Windows 98, but will if you can run Windows NT 4.0 SP6a. As I recall HWiNFO is the same.

red-ray wrote on 2020-02-10, 19:55:

Thank you for trying out SIV and given I don't think SIV has ever been run on an EISA system then it did OK, at least it did not […]
Show full quote

CoffeeOne wrote on 2020-02-10, 19:37:

The gpu detection is not too exciting, what's nice is that all video modes possible of the ATI Graphics Ultra EISA are reported.
For the CPU the external bus is reported with 28MHz, but it should be 33MHz, so also the cpu speed is wrong only 4 times 28MHz. The external cache of 256kB on the mainboard is not detected.
Also no harddisks are detected, most likely because of the EISA bus.

Thank you for trying out SIV and given I don't think SIV has ever been run on an EISA system then it did OK, at least it did not crash.

In general SIV uses the PCI Device ID for GPU reporting which will not be present, that said I does report my Kelvin 64 ISA GPU so I suspect I will be able to improve what it reports for your GPU. To enable me to do this do Menu->File->Save Local and post the generated save file.

The 28MHz is down to the CPU speed being measured as 112 MHz, so I suspect I need to adjust this. Please will you post Menu->Help->Conversions->CPU MHz so I can check.

SIV does not report Disk SMART on Windows 98, but will if you can run Windows NT 4.0 SP6a. As I recall HWiNFO is the same.

Hi,

Thx for having interest. I am getting tired, I think I will continue tomorrow. I will for sure not switch to Windows NT, because I will mostly use the DOS part of Windows 98SE. The windows part I use to transfer files and screenshots 😁

I will provide those log files, but - after reading your private message I have a question - shall I continue with the latest beta version or with the latest version 5.44?