[PATCH] D57978: [CodeGen] Generate follow-up metadata for loops with more than one transformation.
Michael Kruse via Phabricator via cfe-commits
cfe-commits at lists.llvm.org
Mon Mar 25 13:41:33 PDT 2019
Meinersbur added a comment.
In D57978#1440894 <https://reviews.llvm.org/D57978#1440894>, @dmgreen wrote:
> This could maybe do with a few extra tests. Am I correct in saying something like this:
>
> #pragma unroll_and_jam(4)
> for(int j = 0; j < n; j++) {
> #pragma unroll(4)
> for(int k = 0; k < n; k++) {
> x[j*n+k] = 10;
> }
> }
>
>
> would end up with a llvm.loop.unroll_and_jam.followup_inner with a llvm.loop.unroll_count?
correct. I added a new test `pragma-followup_inner.cpp` where this can be seen.
================
Comment at: lib/CodeGen/CGLoopInfo.cpp:500
+ // Unroll-and-jam of an inner loop and unroll-and-jam of the same loop as
+ // the outer loop does not make much sense, but we have to pick an order.
+ AfterJam.UnrollAndJamCount = Attrs.UnrollAndJamCount;
----------------
dmgreen wrote:
> I was having trouble parsing this sentance. Does it mean both the inner loop and the outer loop both have unroll-and-jam? UnrollAndJam processes loops from inner to outer, so if this is working as I think, maybe it should be put into BeforeJam.
> Does it mean both the inner loop and the outer loop both have unroll-and-jam?
Correct. I was thinking of such a case:
```
#pragma unroll_and_jam(2)
for (int i = 0; i < m; i+=1)
#pragma unroll_and_jam(3)
for (int j = 0; j < n; j+=1)
```
Assuming the inner unroll-and-jam would do something, would it happen before or after the outer transformation. As you mentioned, there is an argument to put it into `BeforeJam`. Changed.
================
Comment at: lib/CodeGen/CGLoopInfo.cpp:502
+ AfterJam.UnrollAndJamCount = Attrs.UnrollAndJamCount;
+ AfterJam.UnrollAndJamEnable = AfterJam.UnrollAndJamEnable;
+
----------------
dmgreen wrote:
> = Attrs.UnrollAndJamEnable ?
thanks
Repository:
rC Clang
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D57978/new/
https://reviews.llvm.org/D57978
More information about the cfe-commits
mailing list