[llvm-commits] Please Review: ZERO_EXTEND/SIGN_EXTEND/TRUNCATE optimization for AVX2
Craig Topper
craig.topper at gmail.com
Fri Apr 20 23:56:45 PDT 2012
The test case is full of windows line endings.
+ // VPERMD
+ int ShufMask[] = {0, 2, 4, 6, -1, -1, -1, -1};
That should probably be "static const int ShufMask"
+ int ShufMask[] = {0, 2, -1, -1};
Same here. And only use 1 space after the comma.
+ Op = DAG.getNode(X86ISD::PSHUFB, dl, MVT::v32i8, Op, BV);
+
+ Op = DAG.getNode(ISD::BITCAST, dl, MVT::v4i64, Op);
+
+ int ShufMask[] = {0, 2, -1, -1};
+ Op = DAG.getVectorShuffle(MVT::v4i64, dl, Op,
DAG.getUNDEF(MVT::v4i64), &ShufMask[0]);
+
+ Op = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, MVT::v2i64, Op,
DAG.getIntPtrConstant(0));
+
+ return DAG.getNode(ISD::BITCAST, dl, VT, Op);
There are some dangling spaces in there. Also need to fit in 80 columns
+ SDValue pshufbConstPool = DAG.getNode(ISD::BUILD_VECTOR, dl,
+ MVT::v32i8, &pshufbMask[0], 32);
This looks unused
if (((VT == MVT::v8i32) && (OpVT == MVT::v8i16)) ||
((VT == MVT::v4i64) && (OpVT == MVT::v4i32))) {
This isn't your code, but can you fix the alignment so that the second line
starts after the first line's opening parenthese.
+let Predicates = [HasAVX2] in {
+ let AddedComplexity = 15 in {
+ def : Pat<(v4i64 (X86vzmovly (v4i32 VR128:$src))), (VPMOVZXDQYrr
VR128:$src)>;
+ def : Pat<(v8i32 (X86vzmovly (v8i16 VR128:$src))), (VPMOVZXWDYrr
VR128:$src)>;
+ }
+}
Why the AddedComplexity? Also 80 columns.
+ AddToWorkList(Op.getNode());
} else if (Op.getValueType().bitsGT(VT)) {
Op = DAG.getNode(ISD::TRUNCATE, N->getDebugLoc(), VT, Op);
- }
+ AddToWorkList(Op.getNode());
+ }
Last bracket needs to be right 1 space.
On Thu, Apr 19, 2012 at 10:50 AM, Craig Topper <craig.topper at gmail.com>wrote:
> On Thu, Apr 19, 2012 at 10:29 AM, Chad Rosier <mcrosier at apple.com> wrote:
>
>> Mostly stylistic comments:
>>
>> + SmallVector<SDValue,32> pshufbMask;
>> + for (unsigned i=0; i<2; i++) {
>>
>> Spacing: for (unsigned i = 0; i < 2; i++)
>>
>
> That should also be "++i" to match coding standards.
>
> I'll try to review the content tonight.
>
> --
> ~Craig
>
--
~Craig
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120420/93e56539/attachment.html>
More information about the llvm-commits
mailing list