VOGONS


First post, by morgul12

User metadata
Rank Newbie
Rank
Newbie

Hi all,

I have a QDI 486 motherboard with SIS 471 chipset and 256KB L2 cache that has been awesome to use.. I have used all sorts of CPUs in this motherboard, including Intel 486 DX thru DX4 CPUs, AMD 486 and 5x86 CPUs, all sorts of Cyrix CPUs, and a few of those AMD 5x86 QFP overdrive-style chips such as Evergreen 586 and Kingston TurboChip. Intel 486 Overdrive DX4-100 also worked fine. Most of the time, I'm using an AMD 5x86-133, overclocked to 160 MHz. Everything is stable and works well.

However... yesterday I obtained a Pentium Overdrive 83MHz CPU... and it's been giving me trouble.

The long and the short of it is that the system will POST and I can boot into DOS if I skip config.sys/autoexec.bat (i.e. press F5). If I do a full boot, the system will hang after loading a couple of device drivers. So.. I quickly boot into DOS and try running some benchmark programs from the collection that Phil's Computer Lab assembled. Running SysInfo immediately hangs the system before giving any info. Running Doom causes a crash while loading, usually with a few lines of the letter "D". Running one of the others (Chris' 3D Benchmark, I think) results in a Page Fault error.

Disabling L2 cache immediately resolves all of these failures. All of the benchmarks run smoothly; no crashing, no hanging, and running them multiple times always results in success.

I've read about faulty cache causing such problems.. but my system has been perfectly stable with all other CPUs that I own, including those that are presumably running faster (overclocked 5x86's.. admittedly, Quake ran faster on the Pentium Overdrive even with L2 cache disabled, but the overclocked 5x86's beat the Pentium Overdrive in everything else).

I did try changing all memory timings in the BIOS to the slowest settings and then enabling L2 cache; same problem.

So... does anyone have thoughts on this? Is there, perhaps, something that might be tried besides disabling the L2 cache? Has anyone had a bad cache chip that has resulted in these symptoms (everything works perfectly for most CPUs but fails with one particular type)?

Thanks for any insight you have on this.

Reply 1 of 7, by athlon_p0wer

User metadata
Rank Newbie
Rank
Newbie

I don't know if this is what's going on in your case, but I had 256KB of L2 cache in my 486 for the longest time. Then the upper 128K stopped working. Permanently.

I have never been able to get this motherboard to take any cache higher than 128k since. I couldn't even install DOOM because it would crash halfway through extracting DOOM.WAD, and HIMEM.SYS would say I had bad RAM when I loaded into DOS (I tested the RAM, it was fine). The worst part is when I figured out it was the upper part of the cache, I swapped all the chips in the lower 128k out with the chips that were in the upper 128k, and there were no problems whatsoever. The motherboard just didn't like having 256KB of L2 cache.

From what I've read, it's a ~5% performance loss because once you try to go above 128KB, it starts giving diminishing returns, at least with 486s. I don't know how true this is, but I like to believe it because it makes me feel better about being stuck with only 128k.

Reply 2 of 7, by Eep386

User metadata
Rank Member
Rank
Member

@athlon_power Maybe a pin broke loose on an ASIC on your board? Stuff is getting to that age where pins come loose and TTLs begin to fail in nice scattershot patterns.

Life isn't long enough to re-enable every hidden option in every BIOS on every board... 🙁

Reply 3 of 7, by bakemono

User metadata
Rank Oldbie
Rank
Oldbie

I would try running a memory test program, with and without the L2 enabled. If it does fail when enabled, maybe try it with another CPU as well just to be sure. There could be a fault with an SRAM chip (or socket or motherboard trace?) that tends not to reveal itself except with the Overdrive CPU. The Overdrive has separate instruction and data L1 caches so memory access patterns would be different compared to other socket 3 CPUs. Maybe that scenario is too far-fetched, but a memory test would help to clarify what is happening.

again another retro game on itch: https://90soft90.itch.io/shmup-salad

Reply 4 of 7, by retardware

User metadata
Rank Oldbie
Rank
Oldbie
bakemono wrote on 2021-10-28, 22:50:

I would try running a memory test program, with and without the L2 enabled.

Discussion about suitable memtest software and download links here: memtest86 that works for 486

Reply 5 of 7, by TheMobRules

User metadata
Rank Oldbie
Rank
Oldbie

From what year is your motherboard? Can you post a picture of it?

The POD is very finnicky unless you're running it on later 486 boards with proper support. Reason for this is Intel changed some of the POD specs at the last minute before release, so many boards which were supposed to be compatible with that CPU can experience all kinds of instability. A good rule of thumb is if your board is from early 1995 or before then it is likely to have issues with the POD.

Reply 6 of 7, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Try putting L1 cache in write through mode.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 7 of 7, by morgul12

User metadata
Rank Newbie
Rank
Newbie

Alright.. my issue has been resolved. I went through the jumper settings for cache, planning to lower it to 128KB, when I found one cache jumper (out of seven) that was set improperly . I haven't touched the cache jumpers since I bought the motherboard; looks like that one was set incorrectly because pin 1 was pointing the opposite direction than all the other jumpers around it (it's a 3-pin jumper.. most of the 3-pin jumpers on this motherboard have pin 1 on the left, from the direction that I view the motherboard, but this jumper had pin 1 on the right and was set incorrectly).

So.. I fixed that jumper, and everything worked.. POD is working like a champ, and the 5x86 that I'd been using is also working fine.