VOGONS

Shuttle HOT 307 motherboard 386DX-33 cache problem

Topic actions

First post, by Jamieson

Posted on 
User metadata
Rank Newbie
Rank
Newbie

I've got a Shuttle HOT 307 motherboard with an Intel 386DX-33 and Intel 387-33 installed. Memory is 4 x 1M SIMMs.

It appears to have 64kB cache installed:
bank 0 = 4 x 8Kx8 SRAM -25
bank 1 = 4 x 8Kx8 SRAM -25
tag = 3 x 16Kx4 SRAM -20
Jumpers JP5-JP10 configured for 64K cache

With cache enabled in BIOS:
* POST reports 32K cache, not 64K.
* CACHECHK reports that it doesn't find any cache.
* MEMTEST86 crashes instantly

With cache disabled in BIOS:
* MEMTEST86 runs just fine, no memory errors

Should I be looking for a bad SRAM chip? Solder joints, pins, sockets in the cache area of the motherboard look good. All cache SRAM chips removed and reinstalled, no change.

Reply 1 of 3, by Jamieson

Posted on 
User metadata
Rank Newbie
Rank
Newbie

I'll add that CACHECHK reports BETTER performance with cache disabled in BIOS. That's a weird one!

Reply 2 of 3, by SSTV2

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

Set the cache size to 32KB, this way you can narrow down the problem to Bank0, as it is the primary one. If the cache is still not detected, move all SRAM chips from Bank1 to Bank0 and test again, if that still doesn't work - look for physical damage or check if the jumpers make good contact.

Reply 3 of 3, by Jamieson

Posted on 
User metadata
Rank Newbie
Rank
Newbie

Found the problem!

On this motherboard there is a trace connected to JP6, center pin. JP6 is one of the cache size select jumpers. The trace runs on the top layer really close to a mounting hole and it looks like a screw was really cranked down and cut through the trace. I soldered in a bodge wire (does it have to be green?) and it's good to go.

Now 64KB cache is reported at POST and CACHECHK confirms it. Getting typical 836DX-33 benchmarks now. MEMTEST86+ 2.11 runs without issue.

Go to top of page Go to top of page
Back to General Old Hardware
Close window
Close window