This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: x86/x86-64: Support Intel MNI intrinsics
- From: Andrew Pinski <pinskia at physics dot uc dot edu>
- To: hjl at lucon dot org (H. J. Lu)
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 24 Feb 2006 16:12:31 -0500 (EST)
- Subject: Re: PATCH: x86/x86-64: Support Intel MNI intrinsics
> These 2 patches implement Intel MNI intrinsics and fix the 64bit SSE3
> intrinsics. The corresponding assembler change is at
First you should not be sending two unrelated patches together.
Second from what I heard mwait and monitor are not really SSE3
intrinsics at all.
Third there is little explaination on what these new MNI intrinsics
do?
Fourth why are you doing this:
-(include "sse.md")
(include "mmx.md")
+(include "sse.md")
(include "sync.md")
There is no explaination for that.
Likewise for:
- PTA_MMX = 8,
- PTA_PREFETCH_SSE = 16,
- PTA_3DNOW = 32,
- PTA_3DNOW_A = 64,
- PTA_64BIT = 128
+ PTA_MNI = 8,
+ PTA_MMX = 16,
+ PTA_PREFETCH_SSE = 32,
+ PTA_3DNOW = 64,
+ PTA_3DNOW_A = 128,
+ PTA_64BIT = 256
Why are you reordering them?
Why not just put PTA_MNI as 128 instead?
- { MASK_SSE3, CODE_FOR_sse3_movshdup, 0, IX86_BUILTIN_MOVSHDUP, 0, 0 },
- { MASK_SSE3, CODE_FOR_sse3_movsldup, 0, IX86_BUILTIN_MOVSLDUP, 0, 0 },
+ { MASK_SSE3, CODE_FOR_sse3_movshdup, "__builtin_ia32_movshdup", IX86_BUILTIN_MOVSHDUP, 0, 0 },
+ { MASK_SSE3, CODE_FOR_sse3_movsldup, "__builtin_ia32_movsldup", IX86_BUILTIN_MOVSLDUP, 0, 0 },
That seems very unrelated to your patch anyways.
-- Pinski