Lines Matching refs:insertps
144 ; X32-NEXT: insertps {{.*#+}} xmm0 = zero,xmm1[0],zero,xmm0[3]
149 ; X64-NEXT: insertps {{.*#+}} xmm0 = zero,xmm1[0],zero,xmm0[3]
151 …%tmp1 = call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %t1, <4 x float> %t2, i32 21) nounwi…
155 declare <4 x float> @llvm.x86.sse41.insertps(<4 x float>, <4 x float>, i32) nounwind readnone
157 ; When optimizing for speed, prefer blendps over insertps even if it means we have to
174 ; When optimizing for size, generate an insertps if there's a load fold opportunity.
176 ; generate an insertps for X32 but not for X64!
180 ; X32-NEXT: insertps {{.*#+}} xmm0 = mem[0],xmm0[1,2,3]
192 ; is always just a blendps because blendps is never more expensive than insertps.
267 ; This used to compile to insertps $0 + insertps $16. insertps $0 is always
276 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[2,3]
285 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[2,3]
303 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
308 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
319 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1],xmm1[1],xmm0[3]
324 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1],xmm1[1],xmm0[3]
332 ; instead of insertps
373 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[2,3]
378 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[2,3]
408 ;;;;;; Shuffles optimizable with a single insertps or blend instruction
453 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,1],zero
458 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,1],zero
472 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,3],zero
477 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,3],zero
492 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[3],zero,zero,xmm0[3]
497 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[3],zero,zero,xmm0[3]
510 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,zero,xmm1[0]
515 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,zero,xmm1[0]
528 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,zero,xmm0[0]
533 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,zero,xmm0[0]
546 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
547 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[2]
552 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
553 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[2]
793 ; X32-NEXT: insertps $48, (%{{...}}), {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
798 ; X64-NEXT: insertps $48, (%{{...}}), {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
801 %2 = tail call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %a, <4 x float> %1, i32 48)
805 ;; Use a non-zero CountS for insertps
811 ; X32-NEXT: insertps $32, 4(%{{...}}), {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
816 ; X64-NEXT: insertps $32, 4(%{{...}}), {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
819 %2 = tail call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %a, <4 x float> %1, i32 96)
830 ; X32-NEXT: insertps $0, 12(%{{...}},%{{...}}), {{.*#+}} xmm0 = mem[0],xmm0[1,2,3]
836 ; X64-NEXT: insertps $0, 12(%{{...}},%{{...}}), {{.*#+}} xmm0 = mem[0],xmm0[1,2,3]
840 %3 = tail call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %a, <4 x float> %2, i32 192)
849 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
854 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
862 %7 = tail call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %a, <4 x float> %6, i32 48)
870 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
875 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
883 %7 = tail call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %a, <4 x float> %6, i32 48)
893 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm4[0]
894 ; X32-NEXT: insertps {{.*#+}} xmm1 = xmm1[0,1,2],xmm4[0]
895 ; X32-NEXT: insertps {{.*#+}} xmm2 = xmm2[0,1,2],xmm4[0]
896 ; X32-NEXT: insertps {{.*#+}} xmm3 = xmm3[0,1,2],xmm4[0]
905 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm4[0]
906 ; X64-NEXT: insertps {{.*#+}} xmm1 = xmm1[0,1,2],xmm4[0]
907 ; X64-NEXT: insertps {{.*#+}} xmm2 = xmm2[0,1,2],xmm4[0]
908 ; X64-NEXT: insertps {{.*#+}} xmm3 = xmm3[0,1,2],xmm4[0]
919 %7 = tail call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %a, <4 x float> %6, i32 48)
920 %8 = tail call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %b, <4 x float> %6, i32 48)
921 %9 = tail call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %c, <4 x float> %6, i32 48)
922 %10 = tail call <4 x float> @llvm.x86.sse41.insertps(<4 x float> %d, <4 x float> %6, i32 48)
956 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm0[2],mem[0]
961 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm0[2],mem[0]
968 ; Edge case for insertps where we end up with a shuffle with mask=<0, 7, -1, -1>
993 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm1[2],zero
998 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm1[2],zero
1013 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm1[1],zero,zero
1018 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm1[1],zero,zero
1033 ; X32-NEXT: insertps {{.*#+}} xmm0 = zero,xmm0[1],xmm1[2],zero
1038 ; X64-NEXT: insertps {{.*#+}} xmm0 = zero,xmm0[1],xmm1[2],zero
1052 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm1[1],zero
1057 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm1[1],zero
1072 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],zero,zero
1077 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],zero,zero
1092 ; X32-NEXT: insertps {{.*#+}} xmm1 = zero,xmm0[0],xmm1[2],zero
1098 ; X64-NEXT: insertps {{.*#+}} xmm1 = zero,xmm0[0],xmm1[2],zero
1113 ; X32-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm0[0],zero
1118 ; X64-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],zero,xmm0[0],zero