[SVE][LoopVectorize] Verify support for vectorizing loops with invariant loads
authorKerry McLaughlin <kerry.mclaughlin@arm.com>
Thu, 25 Mar 2021 13:37:02 +0000 (13:37 +0000)
committerKerry McLaughlin <kerry.mclaughlin@arm.com>
Thu, 25 Mar 2021 14:10:21 +0000 (14:10 +0000)
commit1f4649969062fd2b43dcc09c79eb977c83749caa
tree6842ebc8eb6e552b0442578b61a118157b6828b1
parent96a4167b4c7e7e43d11b98f30bed84e4a626281a
[SVE][LoopVectorize] Verify support for vectorizing loops with invariant loads

D95598 added a cost model for broadcast shuffle, which should enable loops
such as the following to vectorize, where the load of b[42] is invariant
and can be done using a scalar load + splat:

  for (int i=0; i<n; ++i)
    a[i] = b[i] + b[42];

This patch adds tests to verify that we can vectorize such loops.

Reviewed By: joechrisellis

Differential Revision: https://reviews.llvm.org/D98506
llvm/test/Transforms/LoopVectorize/AArch64/sve-cond-inv-loads.ll
llvm/test/Transforms/LoopVectorize/AArch64/sve-inv-loads.ll [new file with mode: 0644]