diff options
| author | Michael Zuckerman <Michael.zuckerman@intel.com> | 2017-09-25 14:50:38 +0000 |
|---|---|---|
| committer | Michael Zuckerman <Michael.zuckerman@intel.com> | 2017-09-25 14:50:38 +0000 |
| commit | 4a97df01c43170eed51d2cca0dc779d9d2ca6419 (patch) | |
| tree | de4d3f69196e6feb059f4eef0c09f0ee4c211a0e /lldb/packages/Python/lldbsuite/test/expression_command/macros/TestMacros.py | |
| parent | 31fef4d3f08a15361c0ef9c0fabba5ba083f85ae (diff) | |
| download | bcm5719-llvm-4a97df01c43170eed51d2cca0dc779d9d2ca6419.tar.gz bcm5719-llvm-4a97df01c43170eed51d2cca0dc779d9d2ca6419.zip | |
[X86][LLVM]Expanding Supports lowerInterleavedStore() in X86InterleavedAccess (VF8 stride 4):
This patch expands the support of lowerInterleavedStore to 8x8i stride 4.
LLVM creates suboptimal shuffle code-gen for AVX2.
In overall, this patch is a specific fix for the pattern (Strid=4 VF=8) and we plan to include more patterns in the future.
The patch goal is to optimize the following sequence:
At the end of the computation, we have xmm2, xmm0, xmm12 and xmm3 holding
each 8 chars:
c0, c1, , c7
m0, m1, , m7
y0, y1, , y7
k0, k1, ., k7
And these need to be transposed/interleaved and stored like so:
c0 m0 y0 k0 c1 m1 y1 k1 c2 m2 y2 k2 c3 m3 y3 k3 ....
Reviewers
DavidKreitzer
Farhana
zvi
igorb
guyblank
RKSimon
Ayal
Differential Revision: https://reviews.llvm.org/D36058
Change-Id: I3cc5c2ca5d6318901c192a4428493b99ef424c32
llvm-svn: 314109
Diffstat (limited to 'lldb/packages/Python/lldbsuite/test/expression_command/macros/TestMacros.py')
0 files changed, 0 insertions, 0 deletions
