[PATCH] D19984: [LV] Preserve order of dependences in interleaved accesses analysis

[Matthew Simpson via llvm-commits]

mssimpso marked 2 inline comments as done.

================
Comment at: lib/Transforms/Vectorize/LoopVectorize.cpp:5196-5200
@@ -5106,1 +5195,7 @@
 
+      // If A and B cannot be reordered, we can't add additional instructions
+      // to A's group since this may cause illegal code motion. If B is already
+      // in a group release it.
+      if (!canReorderStridedAccesses(&*I, &*II)) {
+        if (isInterleaved(B)) {
+          InterleaveGroup *StoreGroup = getInterleaveGroup(B);
----------------
> It would be also great to add a testcase to the earlier problem that only candidate accesses were dependency-checked.

I've tried constructing a testcase that would generate incorrect code for this, but I think the current limitations of LAA might actually make this unrealizable at the moment. After the memory checks, two access can be dependent only at constant distances. Thus, if one access is strided, a dependent access at a constant distance would have to be strided as well.

> This sounds right to me. Let me think it over before posting another update.

After thinking about this more carefully, I think we will miss cases with a simplification like the one you suggested. For example, if we have a set of interleaved loads followed by a set of interleaved stores that access the same location (or vice versa), giving up in the outer loop would prevent the two groups from being created.


http://reviews.llvm.org/D19984