What version of Win95?

Caluser2000 wrote on 2021-09-04, 04:37:

What version of Win95?

I’ve tried OSR2 and 95d (the unofficial one)

I wanted to ask as well — the SD to IDE adapter isn’t auto detected at all. I made up the settings in the picture and it seemed to work okay — but it didn’t seem right to just make them up.

I tried using gnu parted on my desktop on the as card but it was obviously using LBA because there were a stupid number of cylinders… and the sectors weren’t 512kb.

Something funky at least. How do I figure out what it’s supposed to be? Hook it up to another pc with a better bios? The next level up of pc has lba and supports 137/128gb drives — so I’m not sure that would be helpful.

aspiringnobody wrote on 2021-09-04, 04:16:

Also: does anyone know what these extra settings after CHS do? When it auto detects the “Lnd” is one less than the cylinders. Pre seems to always be -1.

Lnd is landing zone and Pre is write precompensation cylinder. Both of those are for pre-IDE drives and are irrelevant.
XL likely stands for translation.

How big is the drive that it came with?

And the CF card was totally blank when you put it in this system and you used this system (and the same BIOS you are using now) to partition and format it? If anything there was inconsistent it could cause a problem especially since "XL" rather than "LBA" leads me to believe it might be using a pre-LBA ad hoc translation scheme (aka "Large" in Award BIOS).

What if you let scandisk run a surface scan? Maybe you can figure out if it has a read error and how far into the disk?

jakethompson1 wrote on 2021-09-04, 04:50:

How big is the drive that it came with?

HP Branded/Quantum/ProDrive -- 730MB/1416C/16H/63S/-1 pcomp/lnd 1415

jakethompson1 wrote on 2021-09-04, 04:50:

And the CF card was totally blank when you put it in this system and you used this system (and the same BIOS you are using now) to partition and format it? If anything there was inconsistent it could cause a problem especially since "XL" rather than "LBA" leads me to believe it might be using a pre-LBA ad hoc translation scheme (aka "Large" in Award BIOS).

I've tried several CF cards. All of them blank except one, which was formatted using Linux using the "no format (empty)" option -- e.g. neither MBR nor GPT. I also used dd to blank the first 100 blocks.

jakethompson1 wrote on 2021-09-04, 04:50:

What if you let scandisk run a surface scan? Maybe you can figure out if it has a read error and how far into the disk?

scandisk runs pre-windows install and the surface scan completes without issue. Is there a way I can do a more detailed scan than the one windows does on its own? After I start seeing corruption it is happening in the "files" section of scandisk btw.

I'm leaning towards the board being bad in some way. This chipset is one of the last of the 486 era. It supports write-back cache, for sure. That HP wouldn't have enabled it in their last 486 machine doesn't check out (which explicitly supports DX4 CPUs), so the fact that I can't get the system to post with any writeback enabled CPU installed is suspicious.

Sounds like a bad CD Disk if the install stops at the same location. Its having trouble reading file.

Intel486dx33 wrote on 2021-09-04, 13:14:

Sounds like a bad CD Disk if the install stops at the same location. Its having trouble reading file.

CD works find when I’m using the SD adapter instead of the CF adapter

What is the brand and model of this CF and is it really 4GB?

Cheers,

pentiumspeed wrote on 2021-09-04, 18:44:

What is the brand and model of this CF and is it really 4GB?

I've tried a transcend 4gb CF card and several verbatim 4gb CF cards. I've also tried a 16gb verbatim card just to see if 4gb was some kind of magic number. The board autodetects the 16gb card as 8gb -- so that's probably the limit to whatever "large" method it is using.

I'm reasonably sure that the CF cards aren't the problem in and of themselves -- it is likely an issue with the board's bios. The SD card adapter makes it further and actually installs windows -- but eventually will show data corruption. There is either a bug in the bios with drives larger than were common at the time (and nobody noticed because drives larger than 4gb were prohibitively expensive in 1995) or more likely I've got some kind of conflict going on. The OPTi IDE controller is attached to the VESA Local Bus. The SD card is more like 3.8gb so I suppose it's possible that there is some kind of hard limit to this board between 3.8gb and 4.0gb -- but like I said either way I eventually end up with drive corruption.

I don't want to corrupt the working install on the actual HDD so I have avoided hooking it back up other than to get the CD-ROM driver I needed off of it. The board may just be broken and no matter what I do it will eventually corrupt any drive attached to it.

I pulled the ISA Rom card out of my Compaq Portable today I'm going to try the XT-Universal-Bios tonight even though it says it's not compatible with my controller.

aspiringnobody wrote on 2021-09-04, 19:13:

I pulled the ISA Rom card out of my Compaq Portable today I'm going to try the XT-Universal-Bios tonight even though it says it's not compatible with my controller.

I don't think it's so much that it's "incompatible" as that XT-IDE doesn't know how to recognize and program speeds into VLB/PCI IDE controllers except for a select few. At worst, it will use 16-bit I/O even when 32-bit would be possible.

What's your BIOS string? You have latest BIOS for this board?

jakethompson1 wrote on 2021-09-04, 19:28:

aspiringnobody wrote on 2021-09-04, 19:13:

I pulled the ISA Rom card out of my Compaq Portable today I'm going to try the XT-Universal-Bios tonight even though it says it's not compatible with my controller.

I don't think it's so much that it's "incompatible" as that XT-IDE doesn't know how to recognize and program speeds into VLB/PCI IDE controllers except for a select few. At worst, it will use 16-bit I/O even when 32-bit would be possible.

What's your BIOS string? You have latest BIOS for this board?

Yeah, I'm on GD.05.22 -- I haven't tried re-flashing it in case something is wrong with the board I don't want to brick it. But it's supposedly on the newest bios.

Edit: This board had a super capacitor that it was using to power the CMOS backup ram. I removed it thinking it was a button cell battery -- and replaced it with a CR2032 and a diode to prevent charging. I suppose it is possible that the PC needed that extra capacitance for some reason but I doubt it. There is no real way to know without a schematic for the board. In any case the cap was dead and completely unmarked so I wouldn't know what to replace it with even if I wanted to. It's probably a 5.5v 1F cap but they all look the same except for the physical size so I wouldn't know unless I ordered one.

So, the problem is that this board appears to be incompatible with AMD CPUs. I'm not sure how or why that could be possible -- but with a DX2 66 overclocked to 80mhz it picks up the IDE CD-rom with the normal oak driver and just works. The second I put either of my AMD CPUs in it chokes, even if I slow the bus and multiplier all the way down to 25mhz. Just complete garbage. It will boot and "work" -- but things don't work right in the end. Won't detect the CD-ROM -- won't auto-detect the SD to IDE adapter. Data corruption. All apparently because it doesn't work with AMD CPUs... which since they are register compatible means HP must have programed the bios to check for AMD and not work. What a joke.

I guess I'm back to looking for a new 486 desktop. AGAIN. finding one with L2 cache that actually works is proving to be quite the exercise in frustration.