[llvm] [IA][RISCV] Support VP intrinsics in InterleavedAccessPass (PR #120490)
Min-Yih Hsu via llvm-commits
llvm-commits at lists.llvm.org
Thu Jan 9 09:56:49 PST 2025
================
@@ -248,6 +249,196 @@ static bool isReInterleaveMask(ShuffleVectorInst *SVI, unsigned &Factor,
return false;
}
+// For an (de)interleave tree like this:
+//
+// A C B D
+// |___| |___|
+// |_____|
+// |
+// A B C D
+//
+// We will get ABCD at the end while the leaf operands/results
+// are ACBD, which are also what we initially collected in
+// getVectorInterleaveFactor / getVectorDeinterleaveFactor. But TLI
+// hooks (e.g. lowerInterleavedScalableLoad) expect ABCD, so we need
+// to reorder them by interleaving these values.
+static void interleaveLeafValues(SmallVectorImpl<Value *> &Leaves) {
+ unsigned Factor = Leaves.size();
+ assert(isPowerOf2_32(Factor) && Factor <= 8 && Factor > 1);
----------------
mshockwave wrote:
> or your current approach but with support to all power_of_2 factors.
> so making it generic for all power_of_2 factors would be better.
Oh yeah absolutely, I added `Factor <= 8` here simply because RISC-V (and I believe AArch64?) support at most 8. But that said, I'm happy to relax this restriction, in which case I literally only need to remove this condition and it should support all power-of-two out of the box.
https://github.com/llvm/llvm-project/pull/120490
More information about the llvm-commits
mailing list