Robbbert wrote on 2024-12-02, 03:26:

I have attached the Debug info.

There's nothing in your link of how I can get the HWINFO Report for you.

The attachment HWiNFO32.7z is no longer available

Thanks. You can create the Report file from the main window by pressing the Report icon.
Does "UEFI" or "UEFI Boot" show green? Could you please also attach a screenshot?

Here's the files you requested.

You can see the UEFI things in the BMP file.

EDIT: i was just reading the report and noticed that the AWE64 DMA numbers both show 0; they should be 1 and 5.

Also, the memory module CAS Latency is missing; I think it probably should be 3.

Robbbert wrote on 2024-12-02, 09:39:

Here's the files you requested. […]
Show full quote

Here's the files you requested.

You can see the UEFI things in the BMP file.

The attachment ip101.7z is no longer available

EDIT: i was just reading the report and noticed that the AWE64 DMA numbers both show 0; they should be 1 and 5.

Also, the memory module CAS Latency is missing; I think it probably should be 3.

Thanks. Looks like the problem is because HWiNFO can't access the BIOS here.
I will try to fix this in the next build.
The AWE64 DMA numbers are reported by Windows.
CAS Latency reported is 2.

Here's another one. This is a WIN98FE machine running (rather slowly) on a Pentium 2.

It has the same UEFI issue, and the memory view says the machine is fitted with 2x 128MB RAM sticks, however the System applet claims it has 384MB.

(see attached)

Robbbert wrote on 2024-12-08, 14:04:

Here's another one. This is a WIN98FE machine running (rather slowly) on a Pentium 2. […]
Show full quote

The attachment IP102.7z is no longer available

Here's another one. This is a WIN98FE machine running (rather slowly) on a Pentium 2.

It has the same UEFI issue, and the memory view says the machine is fitted with 2x 128MB RAM sticks, however the System applet claims it has 384MB.

(see attached)

Try the latest Beta version 8.17 if it fixes the UEFI reporting issue.
The 3rd memory module wasn't found because it's SPD doesn't contain a correct checksum. I'll try to workaround this in the next build.

I don't normally use betas, but I tried for this test and yes it does now say Legacy on both machines.

Robbbert wrote on 2024-12-09, 00:14:

I don't normally use betas, but I tried for this test and yes it does now say Legacy on both machines.

Thanks. I also found a problem in the latest build which caused inability to read the BIOS under Win9x. That will be fixed in the next (Beta) build too.

Problem with BIOS readout on Win9x should be resolved in v8.17-5615 Beta

Just tried that beta on IP102, the bios date now shows 05/11/98 instead of blank. Not sure of the date format, is it M/D/Y, or D/M/Y, or is it just a string out of the bios rom? The version and motherboard fields are still blank though.

Also, it now detects the extra ram stick, which shows as unknown. Given that there's a problem with that stick, and the fact that the machine takes forever to boot first time (it's ok after a reboot), I'm going to try replacing that RAM stick. I'll also see if the stick has any identifying info on it.

The unknown RAM has no brand, it has a AST Warranty sticker, another sticker with PC-100 128MB on it, and one more that says LIFETIME WARRANTY. It has 16x V54C365804VCT8PC chips.

Where does HWINFO get the descriptions from? I ask because the RAM that shows as KGM100x64C2/128 is actually KTD-OPGX1N/128 (with 8x HYB39S128800CT-7)

So I changed slot 0 to 256MB, and used the Samsung module in slot 2. If I try to put 512MB in, the bios can only see 384MB, windows gives a protection error, and freezes in safe mode. So lo0ks like 384MB is all it can handle. Even then, the main bios screen can't see anything in slot 0, even though the initial memory count can see it, windows sees it, as does HWINFO. I guess there's a bug in this bios.

BIOS date format should be MM/DD/YY.
HWiNFO detects the RAM module name from SPD integrated in the DIMM that must be properly programmed by the vendor. But it's quite common that legacy vendors didn't want to pay JEDEC to become a member and get an unique ID, so in such case the vendor can't be reliably determined. The rules for SPD back then weren't so strict.