Hi,

the system I'm developing on is an AMD Athlon 64 3700+ (1MB Cache, Clawhammer, 754), it's supposed to have SSE2 (highest SSE revision which this processor supports), even the CPUID instruction confirms it.

However, when I try to employ the "movdqa xmm, xmm" instruction, the software crashes. is this some documentation issue, as the movdqa is supposed to be an SSE2 instruction (AMD64 manual says it is).

just to confirm, the XMM registers are accessible using movq xmm, xmm. but I'm not sure if they're at all useful..

by the way, I'm using the K8N Neo3 mainboard with Phoenix BIOS and apparently I have all the memory in my system available from within real mode. I can confirm the A20 line, for example, is enabled at boot time. if this is normal, then it's a surprise to me.

and as for my last comment, yes, my code is running immediately after the boot process.

Thanks,
Stefan

I have a Athlon 64 3200+ (512 KB Cache, Venice, 939) with SSE3.

Have you enabled SSE first?

See http://board.flatassembler.net/topic.php?t=6146

oh crap, you responded just as I edited my post about that...

yeah, technically, the Socket 754 Clawhammer's are older than the Socket 939 Venice.

the BIOS Setup menu ("del" key on boot) has it enabled.. is that what you're referring to?

No, I don't mean http://board.flatassembler.net/topic.php?p=59421#59421 but CR4.OSFXSR which must be set by the OS.

[edit]But if you haven't set OSFXSR but movq still works it is wierd [/edit]

thanks for the links..

yeah, the movq does work (cpuid reports sse2 is available, too) but it doesn't appear to matter anymore. I just got back from testing CR4.OSFXSR, and using movdqa doesn't crash the program anymore.

anyway.. thank you. I appreciate the help