[Hexagon] Move selection of HVX multiply from lowering to patterns
authorKrzysztof Parzyszek <kparzysz@quicinc.com>
Thu, 1 Oct 2020 16:01:07 +0000 (11:01 -0500)
committerKrzysztof Parzyszek <kparzysz@quicinc.com>
Fri, 2 Oct 2020 21:04:34 +0000 (16:04 -0500)
commitf528816d586a42e4cf27af5b2fa9ba91213307aa
treeae65935c6d1596f34b7cc88184692aedde311a43
parent0373c768c56fc7ce6fe1efa48383f7b376e7bc2a
[Hexagon] Move selection of HVX multiply from lowering to patterns

Also, change i32*i32 to V6_vmpyieoh + V6_vmpyiewuh_acc, which works
on V60 as well.
llvm/lib/Target/Hexagon/HexagonISelLoweringHVX.cpp
llvm/lib/Target/Hexagon/HexagonPatternsHVX.td
llvm/test/CodeGen/Hexagon/autohvx/arith.ll