Ronald G Minnich <[EMAIL PROTECTED]> writes:
> Some more data on the problem.
>
> I do think that the SDRAM is still slightly wrong, I was just surprised
> that it was tickled by WBINVD instruction. But now that I have gotten
> through that by patching the in-memory WBINVD to NOP, I am seeing other
> troubles.
>
> SDRAM is really a pain.
Ron for early testing purposes throw in earlymtrr.inc and pretend
you have ECC sdram. And include to zero it. As a side effect
this sets up the mtrrs over your RAM region and lets do_ramtest.inc
work through the cache, and this should tests burst reads&writes.
Eric
