Is there a reason for setting JP7 (near RAM) to 2-3 (3.6 V) instead of 1-2 (3.45 V)? Your CPU is overvolted...

That's 1-2.

Oh, you're right.
I've got same board revision and jumper settings and I do not have any problem.
AMD 5x86 133 ADZ running 40 MHz x4 and 2-1-1-1 timing.

I would check:
0. Scratches on the board
1. Jumpers. Not only correct settings but if they really connect pins
2. Soldering of SIS 496 - slightly shove its pins with a toothpick or a needle

And maybe swap F244 buffers (I think they're U20 U21) to new ones

Have you tried other srams - do you have a "master set" where you know for sure that it's working?

I assume that you've tested SRAM ICs in another motherboard and confirmed that they are OK. If it's so, then you should follow Roman555's advice and try to gently push pins on chipset ICs with a needle, for possibly cold-soldered pins, test if jumpers are actually shorting headers and look for PCB cuts/missing SMDs.

Question to Roman555:

Have you ever encountered faulty buffer ICs on a motherboard before, that were not I/O bus related?

Board is in good shape, no parts are missing or damaged. I've checked the chipset pins with a needle and they look okay too, everything is solid and without short circuits. Tried to change some jumpers with no luck, will try to change all later. But it seems that buffer theory is the most plausible.

SSTV2 wrote:

Question to Roman555:
Have you ever encountered faulty buffer ICs on a motherboard before, that were not I/O bus related?

Never
P.S. but IMHO they are connected to cache chips, that could be inserted backward, and also connected to VLB.
Of course some measurements are necessary to make decision.

Yeah, buffers dying on themselves would be an unusual case, but possible. I've never encountered such problem myself personally, just curious if others did 😀

But I wonder: can it be wrong BIOS? BIOS has to init L2 cache.
So there are more questions.
The Serpent Rider,
Did you see this mainboard had worked before?
Is the BIOS correct?
About two another sets of SRAM. Are they working for sure? Did you change all the 9 chips or just 8?
Were L1 cache activated?

Motherboard was received "as is". It works fine without L2 cache.
BIOS time stamp and description: 06/25/96-SIS-496-497B-PVI-4SP3C-00
All SRAM works fine in another board.
Yes.'

P.S.
Replaced all jumpers - no luck.

Ok. No new ideas.
Please pay attention that this revision of the chipset supports only FPM RAM !

Actually it works with EDO.
Probably will put it on the back burner. Because I don't have required parts and not so sure if this will help.

I would change the F244 buffers. Have one mainboard that would hang during POST after installing real cache in place of the plastic blocks, because it also needed the buffers. All ok after soldering them.
Maybe you have a faulty one.

The Serpent Rider wrote:

Actually it works with EDO.

No. Only SiS496PR (revision B5) supports EDO.
Yours is SiS496NV (revision B4). It supports only FPM.
So if you use EDO with your board the system doesn't work properly (random crashes)
If you are not sure about type of your DRAM just upload here photos of SIMMs or models of chips

quicknick wrote:

I would change the F244 buffers.

Only if the board doesn't work with proper DRAM type ... There's some confusing with this chipset.

I did extensive tests only with FPM. so that's not the case.