[SVE256] Handle SSE insertions for ADDSUBPD/ADDSUBPS

See #3799
This commit is contained in:
LC committed 2026-06-17 01:54:55 -04:00
1 parent 454c0b31cb
commit be1025901a
3 files changed
+57 -1

No files matched your search

@@ -3464,7 +3464,7 @@ void OpDispatchBuilder::ADDSUBPOp(OpcodeArgs, IR::OpSize ElementSize) {
Ref Src = LoadSourceFPR(Op, Op->Src[0], Op->Flags);
Ref Result = ADDSUBPOpImpl(OpSizeFromSrc(Op), ElementSize, Dest, Src);
StoreResultFPR(Op, Result);
StoreResult_WithAVXInsert(VectorOpType::SSE, RegClass::FPR, Op, Result);
}
void OpDispatchBuilder::VADDSUBPOp(OpcodeArgs, IR::OpSize ElementSize) {
@@ -0,0 +1,28 @@
%ifdef CONFIG
{
"HostFeatures": ["AVX"],
"RegData": {
"XMM0": ["0xfdecd28fab3fa4a5", "0x7d7ccd8836d09fc2", "0xccdbcfc31f3ff0f3", "0x108390defebac4be"],
"XMM1": ["0xc4be43cc1ad6970b", "0x4559bd7eb46a1278", "0", "0"],
"XMM2": ["0xc4be43cc1ad6970b", "0x4549bd7eb46a1278", "0xf793ef673dac6e4c", "0xbb7b5b3d85d34271"],
"XMM3": ["0x000043cc1ad6970b", "0x4549bd7eb46a1278", "0xf793ef673dac6e4c", "0xbb7b5b3d85d34271"]
}
}
%endif
vmovapd ymm0, [rel .data]
vmovapd ymm1, [rel .data + (1 * 32)]
vmovapd ymm2, [rel .data + (2 * 32)]
vmovapd ymm3, [rel .data + (3 * 32)]
addsubpd xmm0, xmm1
vaddsubpd xmm1, xmm2, xmm3
hlt
align 32
.data:
dq 0xfdecd28fab3fa4a5, 0x7d7ccd8836d09fc2, 0xccdbcfc31f3ff0f3, 0x108390defebac4be
dq 0x43cc1ad6970b4549, 0xbd7eb46a1278f793, 0xef673dac6e4cbb7b, 0x5b3d85d342718be9
dq 0xc4be43cc1ad6970b, 0x4549bd7eb46a1278, 0xf793ef673dac6e4c, 0xbb7b5b3d85d34271
dq 0x000043cc1ad6970b, 0x4549bd7eb46a1278, 0xf793ef673dac6e4c, 0xbb7b5b3d85d34271
@@ -0,0 +1,28 @@
%ifdef CONFIG
{
"HostFeatures": ["AVX"],
"RegData": {
"XMM0": ["0xfdecd28fab3fa4a5", "0x7d7ccd8836d09fc2", "0xccdbcfc31f3ff0f3", "0x108390defebac4be"],
"XMM1": ["0xc4be43cc00000000", "0x45c9bd7e00000000", "0", "0"],
"XMM2": ["0xc4be43cc1ad6970b", "0x4549bd7eb46a1278", "0xf793ef673dac6e4c", "0xbb7b5b3d85d34271"],
"XMM3": ["0x000043cc1ad6970b", "0x4549bd7eb46a1278", "0xf793ef673dac6e4c", "0xbb7b5b3d85d34271"]
}
}
%endif
vmovapd ymm0, [rel .data]
vmovapd ymm1, [rel .data + (1 * 32)]
vmovapd ymm2, [rel .data + (2 * 32)]
vmovapd ymm3, [rel .data + (3 * 32)]
addsubps xmm0, xmm1
vaddsubps xmm1, xmm2, xmm3
hlt
align 32
.data:
dq 0xfdecd28fab3fa4a5, 0x7d7ccd8836d09fc2, 0xccdbcfc31f3ff0f3, 0x108390defebac4be
dq 0x43cc1ad6970b4549, 0xbd7eb46a1278f793, 0xef673dac6e4cbb7b, 0x5b3d85d342718be9
dq 0xc4be43cc1ad6970b, 0x4549bd7eb46a1278, 0xf793ef673dac6e4c, 0xbb7b5b3d85d34271
dq 0x000043cc1ad6970b, 0x4549bd7eb46a1278, 0xf793ef673dac6e4c, 0xbb7b5b3d85d34271