Why does the patch have
+#ifdef KEY
- It may be wise to break any PathScale portions into a separate and
clean commit even if it's small and may break something. Sorry I can't
provide a deeper review.
--
Benefiting fr
I know, a lot of the optimizations are added piecemeal and ended up
like that. Soon we should clean them up or let the code quality gets
worse and worse
Sun
On Tue, Apr 19, 2011 at 11:50 AM, David Coakley wrote:
> I agree with your statement in general. In the case of ipo_icall.cxx,
> there is n
I agree with your statement in general. In the case of ipo_icall.cxx,
there is no obvious class to add the functions to -- using C-style
static functions and exporting global functions from each source file
seems to be the pattern in many of the files in ipa/main/optimize.
-David Coakley / AMD Op
if possible, I'd rather not see "static" functions. Why not use
classes and member functions, we are supposed to be writing C++, at
least, the most basic form of C++?
Sun
On Tue, Apr 19, 2011 at 3:30 AM, Min Zhao wrote:
>
> I looked at the patch. It looks fine with me.
>
> Thanks,
>
> Min
>
> On
Author: pallavimathew
Date: 2011-04-18 17:53:10 -0400 (Mon, 18 Apr 2011)
New Revision: 3558
Modified:
trunk/osprey/be/cg/x8664/expand.cxx
trunk/osprey/be/lno/simd.cxx
trunk/osprey/common/com/opcode_gen_core.h
Log:
This patch
1. enables vectorization of BAND, BXOR and SHL operations as wel
I looked at the patch. It looks fine with me.
Thanks,
Min
On Fri, Apr 8, 2011 at 1:19 PM, David Coakley wrote:
> Could a gatekeeper please review the attached patch from the AMD team?
>
> It improves some specific optimizations for C++ programs. The code
> changes are mostly in the area of IP
Author: mberg
Date: 2011-04-18 14:42:38 -0400 (Mon, 18 Apr 2011)
New Revision: 3557
Modified:
trunk/osprey/be/cg/cg.cxx
trunk/osprey/be/cg/cg_flags.cxx
trunk/osprey/be/cg/cg_flags.h
trunk/osprey/be/cg/cgdriver.cxx
trunk/osprey/be/cg/ebo_special.h
trunk/osprey/be/cg/oputil.cxx
The lda-indirect can only be folded when the offset is constant. In that
case, it's safe to set the POINTS_TO to be OFST_IS_FIXED.
2011/4/18 Sun Chan
> I'm not convinced this is the right fix. I don't even know there is a
> correct fix. The code currently is conservative, granted. OTOH, try a
>
I'm not convinced this is the right fix. I don't even know there is a
correct fix. The code currently is conservative, granted. OTOH, try a
few more examples where the *(p+n) varies, where n is 1, 2, 3, .
Sun
On Mon, Apr 18, 2011 at 7:32 PM, Jian-Xin Lai wrote:
> Hi,
>
> Can a gate keeper rev
Hi,
Can a gate keeper review the patch to remove unnecessary CHIs introduced by
Lda-Indirect folding? Thank you very much.
The case is like this:
1 int in[10];
2 voif foo(int i) {
3 int *p = in;
4 *p++ = i++;
5 *p++ = i++;
6 *p++ = i++;
7 *p++ = i++;
...
After the Lda-Indirect folding, th
This patch looks fine to me.
2011/4/13 Yiran Wang
> Hi,
>
> could a gatekeeper review this patch?
>
> Such a combination is abandoned for load-execution as the ld_op is a
> load up higher part of a SSE register, while the alu_op is doing
> conversion on the lower part (implicitly).
>
> Best Rega
This patch looks fine to me.
2011/4/13 Yiran Wang
> Hi All,
>
> Could a gatekeeper please review this patch?
>
> It is about to find out constant branches. If the compare instruction
> has an operator in memory, it is not possible to evaluate it at
> compile time.
>
> Best Regards,
> Yiran
>
> I
12 matches
Mail list logo
