William R. Walsh wrote:
Hi!

Right here it says that all the 815 are limited to 512MB:

Oh, I don't doubt it one bit. I just wonder if that's really true as opposed
to being "officially" true. It wouldn't be the first time I've seen
undocumented support for something that isn't supposed or allowed to work
per the official speculation.

What you are seeing is a defect in the BIOS allowing it to count more
memory than is supported. Not exactly for successful BIOS diddling.

Where the diddling comes in is with the POST error message. At this time it
says "you must uninstall some memory and try again". It simply won't let you
continue. What if it was willing to go right on by? I suspect it might crash
in some creative way later on in the boot process, but who knows?

William


Yeah, you could patch out the jump instruction with some NOP
instructions to see what happens. Not too dangerous, and less than
rocket science. Still, I think it is telling that I have never come
across a system with 815 chipset and more than 512MB. And I cannot
understand why. After all, the predecessor 440BX chipset supported up
to 1GB, although most 440BX boards were documented with a max of 768MB.

This is not the first time Intel trimmed back the maximum memory on its
desktop chipsets, mostly to make its server-workstation chipsets (e.g.
early Xeon vs P3) look better. The first 430 chipset board allowed for
up to 512MB with a low-end Pentium, but the later ones were maxed out at
128MB, until the very end with the 430HX or 430VX (I forget which)
handled up to 256MB. This made the Pentium Pro look much more attractive.

Intel has always been pretty honest about its specs, which are sometimes
driven in strange directions by marketing considerations... Ben Myers