I have now released SIV V5.86 Beta-03 which adds DIMM temperature + PWM Power Controller support for Windows 9x, below both are reported.

Below are DDR3 and DDR5 has been confirmed as working, but I have no reports for DDR4, given they work on NT I will be surprised if they don't work on Windows 9x as it's the same code.

In testing I found that Menu->System->👻 Devices reported them, but [Exorcise] did not work. I plan to look into why and wonder, are there any 👻 Device clean-up utilities for Windows 9x?

I have now released SIV V5.86 Beta-04 which adds Windows 9x support for [Ghost Devices] (Menu->System->Ghost Devices).

I have now released SIV V5.86 Beta-05 which further improves Windows 9x support.

I have now released SIV V5.86 Beta-06 which further improves Windows 9x support. The latest changes are:

I woke up this morning with a request from @red-ray forthe dump files of his software off my Supermicro S2DGU system. Well God Bless old hardware!! I haven't had this system powered up in probably atleast ~4yrs.

Of course it won't POST....it's probably mad at me for all the neglect. Suspect #1 is the enlight PSU (yes I recapped it when I built this system maybe ~10yrs ago).....well fudge, it died anyway....

Did a quick connect of another 5V heavy PSU just for this test....and yay, back up & running!

2x 10,000 RPM SCSI drives in a RAID0, it boots pretty quick for its age.

Couldn't drag this beast out and not fire up some GL Quake!! Of course the movement makes the pics look like crap.

Now for the crux of it.

TBC

Conclusion from above...

Also attached are the SIV software dump files.

Twisted Six wrote on 2026-02-24, 18:29:

Also attached are the SIV software dump files.

Thank you for checking things out and I am happy to hear things worked once you changed the PSU. In general SIV worked as I hoped, but there was one minor issue, if you look at the start of the save file you should see: BIOS R1.6\r\n 01/15/99, basically the \r\n should not be there. I have fixed this for SIV V5.86 Beta-07 which I will release later this week. With the S2DG2 the raw data is "R1.6 ", the fix was trivial.

I am half surprised the fan speeds are zero, are your fans 3-pin that report their speed?

As I suspected would be the case CPUZ reported the motherboard as a "Intel 440GX" and suspect it won't be the only utility that does.

Again, thank you for running the test. Beta-07 also reports the PIRs in colour.

red-ray wrote on 2026-02-24, 21:03:

Thank you for checking things out and I am happy to hear things worked once you changed the PSU. In general SIV worked as I hope […]
Show full quote

Twisted Six wrote on 2026-02-24, 18:29:

Also attached are the SIV software dump files.

Thank you for checking things out and I am happy to hear things worked once you changed the PSU. In general SIV worked as I hoped, but there was one minor issue, if you look at the start of the save file you should see: BIOS R1.6\r\n 01/15/99, basically the \r\n should not be there. I have fixed this for SIV V5.86 Beta-07 which I will release later this week. With the S2DG2 the raw data is "R1.6 ", the fix was trivial.

I am half surprised the fan speeds are zero, are your fans 3-pin that report their speed?

As I suspected would be the case CPUZ reported the motherboard as a "Intel 440GX" and suspect it won't be the only utility that does.

Again, thank you for running the test. Beta-07 also reports the PIRs in colour.

If you're going to issue a fix later this week, just tag me back when you do. I'll see if I can find some time over the weekend or early next week to re-run the test if you like.

Nice stroll down memory lane dragging this one out (it's actually one of my 'retro fleet' favorites)....and my retro fleet is pretty large. Maybe by then I won't be such a nubb that I can actually respond to a PM!! 😁

On PCI-Bus section you show the status of the timings and when for example a card supports 66MHz. But im missing the currently timing the card runs. Just a suggestion, would be nice to see.

Jackhead wrote on 2026-02-25, 10:39:

On PCI-Bus section you show the status of the timings and when for example a card supports 66MHz. But im missing the currently timing the card runs. Just a suggestion, would be nice to see.

AFAIK there is nothing in the common PCI header (first 0x40 bytes) that reports the current speed so a generic solution is not possible.

SIV does report the current speed for some cards. see the last line of the attached, but doing this requires special code just for that card. Which card would you like this for and is the datasheet available?

i cant see the extendet Capabilities. Is this some option i have to activate? I running a voodoo clone pci that is able to run at 66MHz. My board has pci 2.1 and i set 66MHz on bios. But there are other pci boards that dont able run @66MHz but they working. I just wonder if they still run on 33 or not.

Jackhead wrote on 2026-02-25, 14:12:

i cant see the extendet Capabilities. Is this some option i have to activate?

No, Extended Capabilities is PCIe only, it's nothing to do with Configuration (0x40) which is a register specific to the Intel 6702PXH Segment A PCIe to PCI-X Bridge

In the screen shot there are the Error Reporting @ 0x100 + Power Budgeting @ 0x300 Extended Capabilities, I suspect you need to re-read the PCIe specification.

For PCI there are Standard Capabilities offset 0x34 points to the first of them, check the PCI specification.

As I recall the 3dfx Voodoo5 6000 clones use VSA 100-220 chips and AFAIK they don't report the current PCI speed via the PCI config space, have you checked the datasheet? Have a look at strapInfo0 VMI_HD2, my reading is that it would like to run @ 66 MHz rather than it is running @ 66 MHz. Does it change if you set things to run @ 33 MHz?

Jackhead wrote on 2026-02-25, 14:12:

i cant see the extendet Capabilities. Is this some option i have to activate? I running a voodoo clone pci that is able to run at 66MHz. My board has pci 2.1 and i set 66MHz on bios. But there are other pci boards that dont able run @66MHz but they working. I just wonder if they still run on 33 or not.

Basically PCI is a parallel bus.
It's 66 MHz capability will not be used when there is a single card that does not work at 66 MHz.
This is controlled by the M66EN line. Any 33 MHz device will draw this to ground.

Than i cant explain how the performance boost happen if i choose 66MHz in bios against 33 MHz. My two soundboards reported not 66MHz compatible in HWinfo.
3dmark99 half the points on 33MHz, with 66MHz its around 7200points (1024x768). Thats why i like to see what is really going on the pci bus.

Jackhead wrote on 2026-02-25, 17:48:

Than i cant explain how the performance boost happen if i choose 66MHz in bios against 33 MHz. My two soundboards reported not 66MHz compatible in HWinfo.
3dmark99 half the points on 33MHz, with 66MHz its around 7200points (1024x768). Thats why i like to see what is really going on the pci bus.

Maybe, but as I asked what happens to strapInfo0?

You see something?

Hi red-ray,

unfortunately I don't have access to my K6-3 machine for a while from sunday on, so be patient with test requests after saturday.
However, on my other place I have a 486 with 256 MB of RAM where I easily can simulate a similar situation.

My machine is a K6-3 400 MHz on an old 430TX based QDI Titanium IB+. It has installed its maximum of 256 MB RAM.
The K6-3 has an internal on-die 32 kB L1-D, 32 kB L1-I and 256 kB L2 cache.
The QDI Titanium IB+ has 512 kB external cache, which is L3 when using a K6-3, K6-2+ or K6-3+ and L2 when using a K6 or K6-2.
The 430TX chipset has limited cacheable area of 64 MB when paired with 512 kB of external cache.

That cacheable area limit means that megabytes 0-63 are covered by external cache.
But megabytes 64-255 are not covered by external cache.
This means there is a difference on where you place your memory area for doing the cache latency tests.
If you use automatic placement, it depends on the operating system where that memory area will be located.
A 9x based Windows allocates memory requests bottom-up, so it is most likely inside the cacheable area of my machine.
A NT based Windows allocates memory requests top-down, so it is most likely outside the cacheable area of my machine.
So it is clear that there are almost no differences on my screenshot in the area 384 kB and 512 kB to the higher areas and it won't be listed there.
If you fix this I would see differences in the 384 kB and 512 kB area.
Which machines may be affected? All machines with on-board cache, so all 386, 486 and Socket 5, Socket 7 and SS7 boards. Pentium Pro, Pentium II, Athlon and better machines are not affected*.
*As far as I remember there may be some CPU models with internal L2 cache that do not cover full RAM, like not all of the 4 or 64 GB.

The [Cache-o Latency] screenshot does not show the effect of my external L3-cache in the 384 and 512 kB area inside the cacheable area.
The [Machine] screenshot shows that internal L2 is not shown on this screen but the external cache is declared as L2 although it is a L3 here in fact.
It also shows some weird "K5" and "K6 400".
Memory information shows all 4 ranks of my 2 128 MB DIMMs, that are 2 ranks on each DIMM. You display it as 4 "banks". (HWinfo does the same)
It also does not show the devices in the PCI slots. Perhaps it gets confused by the PCI-to-AGP bridge on my Matrox G450 PCI. (HWinfo does the same)

-- Edit:
Debug output files here: Re: SIV support for 386/486/586 class + Alpha CPUs and 3dfx + S3 + SiS + Matrox + XGI + old ATI + NVidia GPUs - Testing
I have to edit this post again. All what I said about memory allocation differences probably is just for the kernel memory allocation.

Jackhead wrote on 2026-02-26, 09:41:

You see something?

I can see strapInfo0 ends 46 meaning AGP Disabled + 66 MHz + 64 KB + Medium.

Does this change if you set 33 MHz in the BIOS?

red-ray wrote on 2023-04-07, 22:41:

I am bored with Disruptor asking questions and not providing the information I need to sensibly answer them so have muted Disruptor.

I just don't get it, I have made it abundantly clear I don't wish Disruptor's disruptive posts to be made any threads I start so why do they keep happening?

Nothing has changed, he still asks questions and fails to attach the SIV save files that I would need were I inclined to answer them.

Disruptor wrote on 2026-02-26, 10:06:

The [Machine] screenshot shows that internal L2 is not shown on this screen but the external cache is declared as L2 although it is a L3 here in fact.

As for [Machine] not reporting the L3 cache then clearly this is a bug in the motherboard BIOS so the BIOS needs to be fixed as it's the BIOS that set's up the SMBIOS data that SIV and other utilities report.

Same value when i set 33MHz. For testing i had only the gpu on the pci bus, same value on 33 and 66.

red-ray wrote on 2026-02-26, 13:58:

https://www.vogons.org/download/file.php?id=237214 […]
Show full quote

Jackhead wrote on 2026-02-26, 09:41:

You see something?

I can see strapInfo0 ends 46 meaning AGP Disabled + 66 MHz + 64 KB + Medium.

Does this change if you set 33 MHz in the BIOS?