[x64] introduce vex prefix version of float64 arithmetic binop

[x64] introduce vex prefix version of float64 arithmetic binop

BUG=
R=bmeurer@chromium.org

Committed: https://chromium.googlesource.com/v8/v8/+/50c4d8826b838596cda892d76ec84eb7a0eab9c1

[Weiliang, 2014-11-27 14:26:49]

weiliang.lin@intel.com changed reviewers:
+ bmeurer@chromium.org, danno@chromium.org, dcarney@chromium.org,
titzer@chromium.org

[Weiliang, 2014-11-27 14:26:50]

PTAL

[Benedikt Meurer, 2014-11-27 14:31:07]

Thanks for the patch. Please address the instruction selector/code generator
concern, and also add unit tests for the instruction selector.

https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
File src/compiler/x64/instruction-codes-x64.h (right):

https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
src/compiler/x64/instruction-codes-x64.h:65: V(AVXFloat64Add)                 \
We should be smarter here, not having SSE and AVX versions, but only one
version. The instruction selector will do the SameAsFirst constraint if AVX is
not supported, and the CodeGenerator will first check whether output and input0
match and if so generate the SSE version, otherwise the AVX version.

https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
File src/compiler/x64/instruction-selector-x64.cc (right):

https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
src/compiler/x64/instruction-selector-x64.cc:739: if
(CpuFeatures::IsSupported(AVX)) {
Use the IsSupported() from InstructionSelector for unittestability.

https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
src/compiler/x64/instruction-selector-x64.cc:751: if
(CpuFeatures::IsSupported(AVX)) {
Use the IsSupported() from InstructionSelector for unittestability.

https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
src/compiler/x64/instruction-selector-x64.cc:763: if
(CpuFeatures::IsSupported(AVX)) {
Use the IsSupported() from InstructionSelector for unittestability.

https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
src/compiler/x64/instruction-selector-x64.cc:775: if
(CpuFeatures::IsSupported(AVX)) {
Use the IsSupported() from InstructionSelector for unittestability.

[Benedikt Meurer, 2014-11-28 06:20:59]

https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
File src/compiler/x64/instruction-codes-x64.h (right):

https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
src/compiler/x64/instruction-codes-x64.h:65: V(AVXFloat64Add)                 \
Forget about this comment, keep both versions. I just noticed that the AVX
encoding doesn't matter here.

[Weiliang, 2014-11-30 07:45:26]

On 2014/11/28 06:20:59, Benedikt Meurer wrote:
>
https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
> File src/compiler/x64/instruction-codes-x64.h (right):
> 
>
https://codereview.chromium.org/764863002/diff/1/src/compiler/x64/instruction...
> src/compiler/x64/instruction-codes-x64.h:65: V(AVXFloat64Add)                
\
> Forget about this comment, keep both versions. I just noticed that the AVX
> encoding doesn't matter here.

Thanks. Please take a look at patch set 2.

Thanks
-Weiliang

[Benedikt Meurer, 2014-12-01 10:33:09]

The CQ bit was checked by bmeurer@chromium.org

[Benedikt Meurer, 2014-12-01 10:33:09]

lgtm

[commit-bot: I haz the power, 2014-12-01 10:33:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/764863002/20001

[Benedikt Meurer, 2014-12-01 10:33:26]

Thanks. Can you port the changes to ia32?

[Benedikt Meurer, 2014-12-01 10:36:49]

The CQ bit was unchecked by bmeurer@chromium.org

[Benedikt Meurer, 2014-12-01 10:45:25]

Committed patchset #2 (id:20001) manually as
50c4d8826b838596cda892d76ec84eb7a0eab9c1 (presubmit successful).