Let's add another point to the list of weird stuff about this system...

[5] I tried to test some HDDs with MHDD - well, I couldn't; no matter which IDE controller I use (on-board or ISA card), MHDD would always identify the drive wrongly (like 0MB in size or 6MB cache, always non-human-readable model/firmware) and would either refuse running because the drive "doesn't support LBA", or sometimes completely disable the IDE port "for your safety".

Other than that, the BIOS detects the drives just fine - I'll try installing DOS/Win3x on one of them, to see what happens.

About the RAM, it "might" be that the system doesn't like many double sided SIMMs?

And if you have a spare >= Pentium 133 laying around, you might as well use one of those. Pentiums >= 133Mhz were made using a smaller process and thus produce less heat (even though I realize this should pose hardly any problems, if any).
The VRE was used on many Pentium 166's (3.52v instead of the usual 3.3v).

And VLSI chipset is odd, iirc that was usually for servers and maybe workstations. Not for consumer computers

Well, DOS and WfW installed just fine... I think I'll keep the system as it is (P75) for now, and use it for DOS gaming, while trying to complete it with some missing small parts (cache module, FDD/HDD mounting kits, front logo bezel (it has one from a VL 5/90 Series 3 now)). After I get bored I'll probably donate it, so currently there are no special pieces in its configuration (2*4+2*8MB of RAM, 2GB NEC HDD with a suspected bad block/sector that's very difficult to find, RTL8019AS NIC and ES1868F sound).

Next thing on the TODO list is to decide whether to install the CMD640 IDE drivers or not (available both for DOS and for Win3.11) - I know how to recover Windows from a badly installed network or video driver, but not IDE 😀. I also don't know if it's worth bothering (performance-wise)... Pretty much the same thoughts apply for the ICU (ISA Configuration Utility, that also seems to provide some sort of PnP driver).

Tetrium wrote:

About the RAM, it "might" be that the system doesn't like many double sided SIMMs? […]
Show full quote

About the RAM, it "might" be that the system doesn't like many double sided SIMMs?

And if you have a spare >= Pentium 133 laying around, you might as well use one of those. Pentiums >= 133Mhz were made using a smaller process and thus produce less heat (even though I realize this should pose hardly any problems, if any).
The VRE was used on many Pentium 166's (3.52v instead of the usual 3.3v).

And VLSI chipset is odd, iirc that was usually for servers and maybe workstations. Not for consumer computers

If the 166's used VRE, what was the core voltage on them?

I do recall now that the 133 and higher used a smaller process, and actually the 120 was different in some way too. I think that only part of the die was shrank too, which is unusual.

SavantStrike wrote:

If the 166's used VRE, what was the core voltage on them?

3.52v. The Pentium classic is single rail

It's basically a factory overclock

If the 166's used VRE, what was the core voltage on them?

I do recall now that the 133 and higher used a smaller process, and actually the 120 was different in some way too. I think that only part of the die was shrank too, which is unusual.

There were actually three voltage specifications for the various P54 chips... STD, VR, and VRE. Technically, STD denotes a range of ~3.15-3.6V (basically 3.3V nominal with pretty loose tolerances), VR is for 3.3-3.45 (3.38V nominal with fairly tight tolerances), and VRE is for 3.45-3.6V. However, many boards just simplified things down to two settings: 3.3V, usually noted as STD/VR, or 3.45V, for VRE.

I think all the P54 chips were listed generically as 3.3V chips, even the ones that used 3.45V. The actual voltage probably depended on the stepping, but I'm not sure. I know it was just as confusing back then as it is now, and you pretty much had to look at the particular chip you had and set it according to how it was marked.

You're right that 120mhz and higher chips used a smaller process, 0.35 micron... I think it was 0.6 micron for the <120mhz ones.

I'm not sure how accurate the info on this site is, but it seems that there should be no difference in voltage settings between P75 and P90:

http://www.cpu-collection.de/?tn=0&l0=co&l1=I … 2=Pentium%20P54

Anyway, I also have a P90 at home - one of these days I'll try it to see what happens...

Now, coming back to the issue list about the Vectra, I tested the CMD IDE drivers yesterday - there seems to be a difference in performance (I now have the "use 32bit disk access" option available in Windows), but the system seems to have become unstable (it happened several times that the computer froze just after exiting Windows, with only a blinking text-mode cursor on a blank screen). So I left the Windows drivers out, and just kept the DOS one in CONFIG.SYS...

I also tested MP3 playing - VBR (over 200kbps average) on P75 sounds much better than 128kbps on DX2/66, but still it's not continuous. I wonder how much the absence of L2 cache matters for MP3 playing...

Be careful with that CMD IDE chip, it was infamous for data corruption problems.

I'm not sure about the 75mhz chips, but I know 90mhz and up can be either 3.3 or 3.45V chips. Though for the most part it was 100mhz and up that used VRE.

I did some reading about CMD640's data corruption issue, and apparently it only occured when trying to access both IDE channels at once; since I only have one IDE connector onboard, I assume I only have one IDE channel, so I should be clear of the infamous bug 😀.

My P90 is this one - http://www.cpu-world.com/sspec/SX/SX968.html

Update 😀: I just tested four cache modules of different timings, one of them HP-branded (not the exact model for this Vectra, but pretty close as far as I could find) - the system would not POST with any of them.

And not only it would not POST, it wouldn't even show any signs of power-up (like lit front-panel LEDs or spinning PSU fan) - is that normal behaviour in case of wrong or malfunctioning cache module? The system boots just fine with or without the cache terminator...