1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-19 02:52:53 +02:00

[X86] Redefine KUNPCK instructions to take a narrower source register class than destination register class. Remove copies from the isel output pattern.

There's no reason for the inputs to be the destination register class. This just
forces an unnecessary copy in the output patterns.

llvm-svn: 358362
This commit is contained in:
Craig Topper 2019-04-14 20:52:42 +00:00
parent 37fe36ab5e
commit 8335a271a1

View File

@ -2988,26 +2988,24 @@ defm : avx512_binop_pat<vxnor, xnor, KXNORWrr>;
defm : avx512_binop_pat<xor, xor, KXORWrr>;
// Mask unpacking
multiclass avx512_mask_unpck<string Suffix,RegisterClass KRC, ValueType VT,
RegisterClass KRCSrc, X86FoldableSchedWrite sched,
multiclass avx512_mask_unpck<string Suffix, X86KVectorVTInfo Dst,
X86KVectorVTInfo Src, X86FoldableSchedWrite sched,
Predicate prd> {
let Predicates = [prd] in {
let hasSideEffects = 0 in
def rr : I<0x4b, MRMSrcReg, (outs KRC:$dst),
(ins KRC:$src1, KRC:$src2),
def rr : I<0x4b, MRMSrcReg, (outs Dst.KRC:$dst),
(ins Src.KRC:$src1, Src.KRC:$src2),
"kunpck"#Suffix#"\t{$src2, $src1, $dst|$dst, $src1, $src2}", []>,
VEX_4V, VEX_L, Sched<[sched]>;
def : Pat<(VT (concat_vectors KRCSrc:$src1, KRCSrc:$src2)),
(!cast<Instruction>(NAME##rr)
(COPY_TO_REGCLASS KRCSrc:$src2, KRC),
(COPY_TO_REGCLASS KRCSrc:$src1, KRC))>;
def : Pat<(Dst.KVT (concat_vectors Src.KRC:$src1, Src.KRC:$src2)),
(!cast<Instruction>(NAME##rr) Src.KRC:$src2, Src.KRC:$src1)>;
}
}
defm KUNPCKBW : avx512_mask_unpck<"bw", VK16, v16i1, VK8, WriteShuffle, HasAVX512>, PD;
defm KUNPCKWD : avx512_mask_unpck<"wd", VK32, v32i1, VK16, WriteShuffle, HasBWI>, PS;
defm KUNPCKDQ : avx512_mask_unpck<"dq", VK64, v64i1, VK32, WriteShuffle, HasBWI>, PS, VEX_W;
defm KUNPCKBW : avx512_mask_unpck<"bw", v16i1_info, v8i1_info, WriteShuffle, HasAVX512>, PD;
defm KUNPCKWD : avx512_mask_unpck<"wd", v32i1_info, v16i1_info, WriteShuffle, HasBWI>, PS;
defm KUNPCKDQ : avx512_mask_unpck<"dq", v64i1_info, v32i1_info, WriteShuffle, HasBWI>, PS, VEX_W;
// Mask bit testing
multiclass avx512_mask_testop<bits<8> opc, string OpcodeStr, RegisterClass KRC,