On Pentium boards cache chips live on CPU bus, bad chip can prevent CPU from operating

rasz_pl wrote on 2024-09-08, 07:19:

On Pentium boards cache chips live on CPU bus, bad chip can prevent CPU from operating

Ahh of course. The datasheet for this chip states that it is a CMOS SRAM so maybe it is literally related to CMOS? I guess the reason why I ask is I am curious if there is any chance the MB will run with it removed entirely? I think I will give it a try either way and see if there is at least some improvement.

On second thought, isnt there L1 cache that lives on the MB and L2 cache that lives on the CPU with Pentium systems?

Think of CMOS as a technology, not the term used commonly for the memory storing BIOS settings etc.

You can safely remove the chip, the system will at least show POST codes if the chip was irritating the CPU bus.

Other way around, this is L2 pipelined burst sync cache. CMOS is just a name for technology used to build SRAM, Complementary MOS.
board will work with L2 removed, will just fail to detect L2 and initialize without it.

Ok awesome thanks for the all the clarification on this. I removed the hot chip without any issues, but still no post codes so far. However it was on bank 0 with bank 1 still on the board, not sure if this would cause problems. I am going to search around for a minute to see if anything else is causing the issue, then maybe remove bank 1 if I can't find anything else.

As was probably assumed, pulling the bank 1 chip made no difference with POST. What was interesting is there were a few bare spots on the traces. Not sure if that happened from the applied heat while removing or if it was possible there had been some work on this board previously. I did a little more looking around for things like broken traces or cold solder joints but nothing is really standing out.

Not sure where to go from here unfortunately. This board seems to be totally dead and nothing is really indicating to me where I should look.

Being fried and internally shorted is not the only reason for SRAM chip to get hot. 3.3V might be missing/broken making it power itself from data lines and latch up, or power something else thru SRAM data lines.

rasz_pl wrote on 2024-09-09, 01:20:

Being fried and internally shorted is not the only reason for SRAM chip to get hot. 3.3V might be missing/broken making it power itself from data lines and latch up, or power something else thru SRAM data lines.

This is interesting because the ISA / PCI diag card was indicating that 3.3v was missing as well. Very well could have been what was going on with the cache chip in general.

Unfortunately when I was looking over the PCI chipset that had the bent pins that I fixed to double check my work, a closer look revealed there were a bunch of lifted pads on a group of pins down the line. My attempt to straighten them out and flatten things out was not successful and made a bigger mess. I decided I could probably pull the chip with the hot air station but sort of hit the point where it could be diminishing returns for all the work it will take. I am not sure if the chip is going to work after getting the pads cleaned up so I might as well do it if I have a donor board around, or use this board as a donor for other projects.

Sad either way, it seems like a nice board if it was working.

Delphius wrote on 2024-09-10, 01:51:

This is interesting because the ISA / PCI diag card was indicating that 3.3v was missing as well. Very well could have been what was going on with the cache chip in general.

Thats normal, ISA doesnt have 3.3V rail at all, and on PCI its optional.