Get your least pretty DX4, engage it in the socket without fully inserting, pull the lever over until you feel resistance, and scraunch it up and down a few times with a bit of pressure on the lever... the idea being to break through any oxidation that's causing problems.

Thanks, tried that but it made no difference. The only thing that made a slight difference (made it post more often than not and be a little bit more stable) was to lower the clock multiplier so it ran at 75mhz instead of 100mhz.... 🙁 this is the same behaviour with boxh DX4-100 CPU's so it must be something on the board causing the issue (but yet not affecting the DX2-66 which is rock solid stable). Maybe Capacitors or something?

Could be your cache or memory timings are just marginal enough that the DX-2 can't hammer them 100% and the brief respites allow enough recovery, whereas the DX4 at 100 has enough CPU power to keep up the pressure and they flake.

Cache could be the reason indeed. I've got the same board and have 15ns cache chips installed. The board runs with about every CPU I've thrown in (Intel, amd, cyrix) including DX2, dx4@100, am5x86@150 and cx5x86@100.

Too short this out you may try running it without l2 cache enabled, so the cache is ruled out. If it runs, you can add some wait states to l2.

By the way, could it be you're DX2 is wb l2 and you're dx4 is wt only, with the board set to wb? Which CPUs did you use exactly?

(And for jumpers you may refer to ultimateretro, in cast i hadn't found it yet: https://www.ultimateretro.net/nl/motherboards/640)

I have got 15ns Cache installed for the "External Cache" which I assume is L2 (the BIOS refers to this as External Cache). The chips are W24257AK-15.

The CPU I have got (in both cases) is a P24C Intel 80486 DX4-100 "SK051" which I believe only does write through (which is set in the BIOS). I have set the jumpers for that CPU and the Voltage Jumpers and I measured the voltage at the CPU socket before even attempting to use either DX4 Chip.

If I disable External Cache, the performance of the machine is significantly lower than with it enabled which I assume is to be expected?

Basically there should be 2 DX4 jumper settings - one for older write through ones, one for write back ones.
Your SK051 seems to not have write back L1 cache. https://www.cpu-world.com/sspec/SK/SK051.html

I just found the same Spec page and realised it only supports Write through.

I can only see 1 set of jumper settings for P24C / DX4. Nothing about Write Back or Write Through and I have made sure the BIOS is set to Write Through mode.

Just to add to this, when I do get the thing to eventually fire up and POST & Boot with the DX4-100, if I run 3D Bench from Phil's Computer Lab's dosbench pack, the S logo it shows usually has some corruption (some black dots in the S that shouldn't be there). However, when I run this with the DX2-66 it works perfectly everytime and the S looks as it should do.

That corruption almost looks like something is running a bit too fast for things to keep up. Have you tried relaxing the cache/memory timings or added waitstates to 8 or 16-bit I/O? If you're using a VLB video card, is there a waitstate jumper on it?

I can safely assure you electrolytic caps aren't the issue. Probably.

Also somewhat possible is the BIOS not really liking the DX4 at 100MHz. I've had a motherboard with an -ahem- mentally challenged BIOS that would interpret Am5x86-133s as DX2-100s and run them with the L1 cache disabled and other "nice" glitchy stuff. When I forced the multiplier to 3X (hence turning them into DX4-100s) it ran fine.

I can't see any options to change the Cache or Memory Timings? There are no jumpers on the board either for this that I can see.

I have a VLB Video Card which is a Cirrus Logic PT-524/8G VGA ISA Graphics / Video Card (Chipset CL-GD5428-80QC-A), I would need some assistance with wait states. There is a jumper on the board related to VLB Wait State which I can set to 1 but I am pretty sure that made no difference.

Edit: Actually hold the boat, UH '19 has a BIOS I can rummage through. Let's take a look...

Are you able to dump the BIOS? I'd like to take a look-see at it to find out if the manufacturer disabled the CMOS setup options for the wait states.
Best way would be to dump it using a TL866 or the like, but a DOS utility (not NSSI) may work provided you turn off all BIOS shadowing first (if the thing even allows for that).

Blech, the one on UH'19 gives a checksum error when loaded into AWDHACK. So it would seem I still need a dump after all...

You will likely find it online.

Award Modular BIOS v4.50G

11/17/94-UMC-498GP-2C4X6A31-00

I tried the one on UH '19, but it gives a checksum error under AWDHACK. However, it runs in PCem. (Probably an earlier version without checksum check, or something.)

Looking at the contents, there are L2 Wait State options but they're dummied out. There's also an Auto Configuration option, also dummied out.

So could that be the issue? I need additional controls that I don't have?

It could very well be the problem. Unfortunately I am not able to load the BIOS from UH '19 into any version of MODBIN or AWDHACK I can get, it always gives a checksum error. If I had to guess, the manufacturer, knowing that this was a version of Award BIOS that didn't have a checksum check, crudely patched those options out without bothering to update the in-file checksum. (Or maybe they simply disabled the file's own checksum check so that their ugly hacks just worked.)

One possible workaround: try a BIOS for a different UM8498F based motherboard (but keep your current BIOS on its own chip as a spare in case it doesn't work). One possible candidate would be the PC Chips M912 V1.7. The file I'm linking is from Ultimate Retro '19 and is an AMIBIOS, 12/01/95.
https://www.ultimateretro.net/motherboard/bio … 9a868814393.zip

DaveDoc1984 wrote on 2022-04-20, 15:25:

Hi Everyone, I have recently been putting together a retro build and initially went with a DX2-66 Intel CPU but have since bough […]
Show full quote

Hi Everyone, I have recently been putting together a retro build and initially went with a DX2-66 Intel CPU but have since bought a DX4-100 Intel CPU.

The issue I have got, is whilst it works fine when setup with the DX2-66 and is stable, when I set the jumpers for the DX4-100 (Referred to as P24C on the Motherboard Silkscreen), I will be lucky if the machine posts at all...

Often after many attempts at powering off an on again it will eventually post and show the DX4-100 on the BIOS.. however upon boot-up of the OS (Windows 95) it will crash shortly afterwards (even in MS DOS mode). Sometimes I can get it to boot to the DOS Prompt but if I as much as run a game or even a Benchmark from Phil's Computerlab's pack of Benchmarks, it will freeze very shortly afterwards.

Any ideas as to why I can't see to run a DX4-100 off the board? I have tried 2 different DX4-100 chips (1 that is confirmed to work on another board) and no dice. I made sure with my multimeter that the voltage was set correctly at 3.45 volts before installing the DX4-100 but to no avail.

Any help you can give me to figure out why a DX4-100 won't run would be greatly appreciated.

JP20, JP21, JP22 are opened or in what position?

Are there "waitstates" or "VLBUS speed" jumpers on videocard?

JP21 and JP22 both closed. I have also tried with JP22 open (JP21 should be closed as the CPU is greater than 33mhz).

JP20 Open as its 3X 33mhz for the DX4-100.

I will need a Jumper Manual for the Video Card to know what the Jumpers do as yes there are Jumpers on the Card.