muchas gracias for the offer, I've sent you a PM.

DonutKing wrote:

Thanks but I already ordered some form utsource.net. They're only 15ns but that's still good for 66MHz bus speed - and I'm not planning to overclock so that's fine by me. They're a little more expensive at $2.50 each but shipping was only $8 for basic registered post.

The part number is IS61C102415N and you'll need 9 of them, including the TAG RAM which is near the large SIS chip.

I finally got around to testing this cache out - the system reports 1024 installed correctly but Himem.sys reports errors and the system is unusable. I tried a few other 256kb sets I have and they work fine, so it appears to be these chips.

Have you tried them DonutKing?

Also tried that updated BIOS you sent me and it works like a charm. I've whacked a am5x86 in there and it's very feisty indeed (for a 486).

Yep, my SV2G board hums away happily with 1MB of cache.
They also pass cachechk and speedsys tests.

Sounds like you may have a dodgy chip 🙁 Did you upgrade the TAG RAM chip as well?

Good to hear the BIOS worked 😀

I've had motherboards that just wouldn't work with 1024 KB cache, although the cache was good and the board's spec. supported 1024 KB. For your '1024' cache pieces, try setting it up as 512 KB cache (4 chips), set the TAG and jumpers appropriately, and run HIMEM. If it works fine with both sets of 4 '1024' cache pieces, then your issue is not with your cache modules. If you still get HIMEM errors w/512 KB cache, try jugling around the cache pieces around to isolate the bad pieces. Also, you'll want to use concervative cache and memory timings in the BIOS for this test. As I noted earlier, perhaps in another post, was that of all the '1024' cache pieces, about 10% were faulty. Good luck.

feipoa wrote:

For your '1024' cache pieces, try setting it up as 512 KB cache (4 chips), set the TAG and jumpers appropriately, and run HIMEM. If it works fine with both sets of 4 '1024' cache pieces, then your issue is not with your cache modules. If you still get HIMEM errors w/512 KB cache, try jugling around the cache pieces around to isolate the bad pieces.

Thanks for the ideas, gave it a shot but no luck, although I've realised since that the manual I have for this board (ASUS VL/I-486SV2GX4) is for an older revision than mine, I have 2.1, so I suspect I was setting one of the jumpers incorrectly which would probably explain it.

In the end I managed to get it working with a set of 512k chips I pulled off a socket 5 board I'm throwing out, but even with the correct jumper settings (silk screened onto the board it turns out) it wasn't easy. For a while it was reporting 128k and only after much fiddling did they work correctly.

DonutKing wrote:

Yep, my SV2G board hums away happily with 1MB of cache.
They also pass cachechk and speedsys tests.

I've had 1 MB cache 486 boards which would run fine in cachechk and speedsys but failed HIMEM and MemTest. Can you verify with himem and 2 hours of MemTest?

@badmofo
Seems strange that you couldn't find 4 out of 8 1024 chips that are good. A 50% failure seems unlikely. I wonder if your chipset or some component on the PCB is marginally faulty?

I've had 1 MB cache 486 boards which would run fine in cachechk and speedsys but failed HIMEM and MemTest. Can you verify with himem and 2 hours of MemTest?

That system loads himem.sys and I've never seen it fail the memory test. I won't be able to run memtest for a while as its all packed away, but I was playing games for several hours and never had a problem.

feipoa wrote:

@badmofo
Seems strange that you couldn't find 4 out of 8 1024 chips that are good. A 50% failure seems unlikely. I wonder if your chipset or some component on the PCB is marginally faulty?

I think it was because I was using jumper settings from an older version of the manual, but the fun factor was dwindling after swapping chips for 2 hours so I gave up once I had a different set of 512 chips working. 512 is fine for my needs, as was 256 I suspect...

Thanks for your interest though, I'll give the 1024 chips another go eventually I'm sure.

How about testing the '1024' chips in a different motherboard setup for 512KB cache (4 chips)? You should be able to find the faulty pieces.

You're right and I will try again at some stage, I think I'll save them up until I'm luck enough to come across a nice PCI socket 3 board.