Author: Eugene Nalimov
Date: 17:03:16 08/26/03
Go up one level in this thread
On August 26, 2003 at 19:29:10, Anthony Cozzie wrote: >On August 26, 2003 at 18:58:20, Eugene Nalimov wrote: > >>On August 26, 2003 at 18:32:49, Jeremiah Penery wrote: >> >>>On August 26, 2003 at 17:42:42, Gerd Isenberg wrote: >>> >>>>Hmm... that makes using these register impossible - at least with a >>>>multithreading approach. Eight bitboard (or float) register for nothing, or only >>>>a few percent faster context switch. Ok, there are still the 16 (14) general >>>>purpose bitboards and 32 XMM bitboards, nevertheless... >>>> >>>>Gerd >>>> >>>> >>>>Porting Windows® Device >>>>Drivers to AMD64 Platforms >>>> >>>>Page 7 >>>> >>>>• Convert media instruction to SSE/SSE2 Instructions >>>> >>>>Microsoft Windows for AMD64 will not context switch x87, 3Dnow!, MMX for 64-bit >>>>native threads. This code may be converted to SSE/SSE2 through the use of >>>>intrinsic functions. >>> >>>What exactly does this paragraph mean? If it means what it seems to mean, I >>>don't see how it could get away with that for regular x87 code. I also don't >>>see why that means that the MMX registers are not thread-safe. If it means >>>something else, it's certainly not very clearly stated. >> >>That paragraph means that 64-bit code on AMD64 Windows cannot use MMX or x87 >>instructions. >> >>Thanks, >>Eugene > >isn't x87 the floading point unit? So you are saying Windows for Opteron won't >be able to do any floating point math at all? > >anthony No, I am saying that you have to use non-x87 to do FP on Windows for AMD64 -- e.g. SSE2. Thanks, Eugene
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.