mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-22 02:33:06 +01:00
[ARM] Make remaining MVE instruction predictable
The unpredictable/hasSideEffects flag is usually inferred by tablegen from whether the instruction has a tablegen pattern (and that pattern only has a single output instruction). Now that the MVE intrinsics are all committed and producing code, the remaining instructions still marked as unpredictable need to be specially handled. This adds the flag directly to instructions that need it, notably the V*MLAL instructions and some of the MOV's. Differential Revision: https://reviews.llvm.org/D76910
This commit is contained in:
David Green
parent
fb0145ca24
commit
ed90d278a6
@ -439,6 +439,8 @@ class MVE_ScalarShiftDoubleReg<string iname, dag iops, string asm,
|
||||
|
||||
let Inst{19-17} = RdaLo{3-1};
|
||||
let Inst{11-9} = RdaHi{3-1};
|
||||
|
||||
let hasSideEffects = 0;
|
||||
}
|
||||
|
||||
class MVE_ScalarShiftDRegImm<string iname, bits<2> op5_4, bit op16,
|
||||
@ -756,6 +758,7 @@ class MVE_VMINMAXNMV<string iname, string suffix, bit sz,
|
||||
let horizontalReduction = 1;
|
||||
|
||||
let Predicates = [HasMVEFloat];
|
||||
let hasSideEffects = 0;
|
||||
}
|
||||
|
||||
multiclass MVE_VMINMAXNMV_p<string iname, bit notAbs, bit isMin,
|
||||
@ -1065,6 +1068,8 @@ class MVE_VMLALDAVBase<string iname, string suffix, dag iops, string cstr,
|
||||
let Inst{3-1} = Qm{2-0};
|
||||
let Inst{0} = bit_0;
|
||||
let horizontalReduction = 1;
|
||||
|
||||
let hasSideEffects = 0;
|
||||
}
|
||||
|
||||
multiclass MVE_VMLALDAVBase_A<string iname, string x, string suffix,
|
||||
@ -1610,6 +1615,8 @@ class MVE_VMOV_lane<string suffix, bit U, dag indexop,
|
||||
let Inst{11-8} = 0b1011;
|
||||
let Inst{7} = Qd{3};
|
||||
let Inst{4-0} = 0b10000;
|
||||
|
||||
let hasSideEffects = 0;
|
||||
}
|
||||
|
||||
class MVE_VMOV_lane_32<MVE_VMOV_lane_direction dir>
|
||||
@ -5227,6 +5234,7 @@ class MVE_VFMAMLA_qr<string iname, string suffix,
|
||||
let Inst{8} = 0b0;
|
||||
let Inst{5} = 0b0;
|
||||
let validForTailPredication = 1;
|
||||
let hasSideEffects = 0;
|
||||
}
|
||||
|
||||
multiclass MVE_VMLA_qr_multi<string iname, MVEVectorVTInfo VTI,
|
||||
@ -5375,6 +5383,7 @@ class MVE_VxDUP<string iname, string suffix, bits<2> size, bit bit_12,
|
||||
let Inst{6-1} = 0b110111;
|
||||
let Inst{0} = imm{0};
|
||||
let validForTailPredication = 1;
|
||||
let hasSideEffects = 0;
|
||||
}
|
||||
|
||||
def MVE_VIDUPu8 : MVE_VxDUP<"vidup", "u8", 0b00, 0b0>;
|
||||
@ -5410,6 +5419,7 @@ class MVE_VxWDUP<string iname, string suffix, bits<2> size, bit bit_12,
|
||||
let Inst{3-1} = Rm{3-1};
|
||||
let Inst{0} = imm{0};
|
||||
let validForTailPredication = 1;
|
||||
let hasSideEffects = 0;
|
||||
}
|
||||
|
||||
def MVE_VIWDUPu8 : MVE_VxWDUP<"viwdup", "u8", 0b00, 0b0>;
|
||||
@ -5478,6 +5488,8 @@ class MVE_VMOV_64bit<dag oops, dag iops, bit to_qreg, string ops, string cstr>
|
||||
let Inst{12-5} = 0b01111000;
|
||||
let Inst{4} = idx2;
|
||||
let Inst{3-0} = Rt{3-0};
|
||||
|
||||
let hasSideEffects = 0;
|
||||
}
|
||||
|
||||
// The assembly syntax for these instructions mentions the vector
|
||||
|
||||
@ -84,17 +84,17 @@ define void @vld3_v8i32(<24 x i32> *%src, <8 x i32> *%dst) {
|
||||
; CHECK-NEXT: .vsave {d8, d9, d10, d11, d12, d13}
|
||||
; CHECK-NEXT: vpush {d8, d9, d10, d11, d12, d13}
|
||||
; CHECK-NEXT: vldrw.u32 q1, [r0, #48]
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #64]
|
||||
; CHECK-NEXT: vldrw.u32 q2, [r0, #80]
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #64]
|
||||
; CHECK-NEXT: vmov.f32 s12, s5
|
||||
; CHECK-NEXT: vmov.f32 s13, s0
|
||||
; CHECK-NEXT: vmov.32 r2, q2[2]
|
||||
; CHECK-NEXT: vmov.f32 s13, s0
|
||||
; CHECK-NEXT: vdup.32 q4, r2
|
||||
; CHECK-NEXT: vmov.f32 s14, s3
|
||||
; CHECK-NEXT: vmov.f32 s15, s19
|
||||
; CHECK-NEXT: vmov.32 r2, q2[1]
|
||||
; CHECK-NEXT: vmov.f64 d8, d2
|
||||
; CHECK-NEXT: vmov.f32 s15, s19
|
||||
; CHECK-NEXT: vdup.32 q5, r2
|
||||
; CHECK-NEXT: vmov.f64 d8, d2
|
||||
; CHECK-NEXT: vmov.f32 s17, s7
|
||||
; CHECK-NEXT: vmov.f32 s18, s2
|
||||
; CHECK-NEXT: vmov.f32 s0, s6
|
||||
@ -108,6 +108,7 @@ define void @vld3_v8i32(<24 x i32> *%src, <8 x i32> *%dst) {
|
||||
; CHECK-NEXT: vadd.i32 q0, q3, q0
|
||||
; CHECK-NEXT: vldrw.u32 q3, [r0, #32]
|
||||
; CHECK-NEXT: vmov.f32 s16, s9
|
||||
; CHECK-NEXT: vstrw.32 q0, [r1, #16]
|
||||
; CHECK-NEXT: vmov.f32 s17, s4
|
||||
; CHECK-NEXT: vmov.32 r0, q3[2]
|
||||
; CHECK-NEXT: vdup.32 q5, r0
|
||||
@ -116,7 +117,6 @@ define void @vld3_v8i32(<24 x i32> *%src, <8 x i32> *%dst) {
|
||||
; CHECK-NEXT: vmov.32 r0, q3[1]
|
||||
; CHECK-NEXT: vmov.f64 d10, d4
|
||||
; CHECK-NEXT: vdup.32 q6, r0
|
||||
; CHECK-NEXT: vstrw.32 q0, [r1, #16]
|
||||
; CHECK-NEXT: vmov.f32 s21, s11
|
||||
; CHECK-NEXT: vmov.f32 s22, s6
|
||||
; CHECK-NEXT: vmov.f32 s4, s10
|
||||
@ -148,17 +148,17 @@ define void @vld3_v16i32(<48 x i32> *%src, <16 x i32> *%dst) {
|
||||
; CHECK-NEXT: .pad #16
|
||||
; CHECK-NEXT: sub sp, #16
|
||||
; CHECK-NEXT: vldrw.u32 q1, [r0, #48]
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #64]
|
||||
; CHECK-NEXT: vldrw.u32 q2, [r0, #80]
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #64]
|
||||
; CHECK-NEXT: vmov.f32 s12, s5
|
||||
; CHECK-NEXT: vmov.f32 s13, s0
|
||||
; CHECK-NEXT: vmov.32 r2, q2[2]
|
||||
; CHECK-NEXT: vmov.f32 s13, s0
|
||||
; CHECK-NEXT: vdup.32 q4, r2
|
||||
; CHECK-NEXT: vmov.f32 s14, s3
|
||||
; CHECK-NEXT: vmov.f32 s15, s19
|
||||
; CHECK-NEXT: vmov.32 r2, q2[1]
|
||||
; CHECK-NEXT: vmov.f64 d8, d2
|
||||
; CHECK-NEXT: vmov.f32 s15, s19
|
||||
; CHECK-NEXT: vdup.32 q5, r2
|
||||
; CHECK-NEXT: vmov.f64 d8, d2
|
||||
; CHECK-NEXT: vmov.f32 s17, s7
|
||||
; CHECK-NEXT: vmov.f32 s18, s2
|
||||
; CHECK-NEXT: vmov.f32 s0, s6
|
||||
@ -181,6 +181,7 @@ define void @vld3_v16i32(<48 x i32> *%src, <16 x i32> *%dst) {
|
||||
; CHECK-NEXT: vmov.32 r2, q3[1]
|
||||
; CHECK-NEXT: vmov.f64 d10, d4
|
||||
; CHECK-NEXT: vdup.32 q6, r2
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #128]
|
||||
; CHECK-NEXT: vmov.f32 s21, s11
|
||||
; CHECK-NEXT: vmov.f32 s22, s6
|
||||
; CHECK-NEXT: vmov.f32 s4, s10
|
||||
@ -202,7 +203,6 @@ define void @vld3_v16i32(<48 x i32> *%src, <16 x i32> *%dst) {
|
||||
; CHECK-NEXT: vmov.32 r2, q4[1]
|
||||
; CHECK-NEXT: vmov.f64 d12, d6
|
||||
; CHECK-NEXT: vdup.32 q7, r2
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #128]
|
||||
; CHECK-NEXT: vmov.f32 s25, s15
|
||||
; CHECK-NEXT: vmov.f32 s26, s10
|
||||
; CHECK-NEXT: vmov.f32 s8, s14
|
||||
@ -1125,17 +1125,17 @@ define void @vld3_v8f32(<24 x float> *%src, <8 x float> *%dst) {
|
||||
; CHECK-NEXT: .vsave {d8, d9, d10, d11, d12, d13}
|
||||
; CHECK-NEXT: vpush {d8, d9, d10, d11, d12, d13}
|
||||
; CHECK-NEXT: vldrw.u32 q1, [r0, #48]
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #64]
|
||||
; CHECK-NEXT: vldrw.u32 q2, [r0, #80]
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #64]
|
||||
; CHECK-NEXT: vmov.f32 s12, s5
|
||||
; CHECK-NEXT: vmov.f32 s13, s0
|
||||
; CHECK-NEXT: vmov.32 r2, q2[2]
|
||||
; CHECK-NEXT: vmov.f32 s13, s0
|
||||
; CHECK-NEXT: vdup.32 q4, r2
|
||||
; CHECK-NEXT: vmov.f32 s14, s3
|
||||
; CHECK-NEXT: vmov.f32 s15, s19
|
||||
; CHECK-NEXT: vmov.32 r2, q2[1]
|
||||
; CHECK-NEXT: vmov.f64 d8, d2
|
||||
; CHECK-NEXT: vmov.f32 s15, s19
|
||||
; CHECK-NEXT: vdup.32 q5, r2
|
||||
; CHECK-NEXT: vmov.f64 d8, d2
|
||||
; CHECK-NEXT: vmov.f32 s17, s7
|
||||
; CHECK-NEXT: vmov.f32 s18, s2
|
||||
; CHECK-NEXT: vmov.f32 s0, s6
|
||||
@ -1149,6 +1149,7 @@ define void @vld3_v8f32(<24 x float> *%src, <8 x float> *%dst) {
|
||||
; CHECK-NEXT: vadd.f32 q0, q3, q0
|
||||
; CHECK-NEXT: vldrw.u32 q3, [r0, #32]
|
||||
; CHECK-NEXT: vmov.f32 s16, s9
|
||||
; CHECK-NEXT: vstrw.32 q0, [r1, #16]
|
||||
; CHECK-NEXT: vmov.f32 s17, s4
|
||||
; CHECK-NEXT: vmov.32 r0, q3[2]
|
||||
; CHECK-NEXT: vdup.32 q5, r0
|
||||
@ -1157,7 +1158,6 @@ define void @vld3_v8f32(<24 x float> *%src, <8 x float> *%dst) {
|
||||
; CHECK-NEXT: vmov.32 r0, q3[1]
|
||||
; CHECK-NEXT: vmov.f64 d10, d4
|
||||
; CHECK-NEXT: vdup.32 q6, r0
|
||||
; CHECK-NEXT: vstrw.32 q0, [r1, #16]
|
||||
; CHECK-NEXT: vmov.f32 s21, s11
|
||||
; CHECK-NEXT: vmov.f32 s22, s6
|
||||
; CHECK-NEXT: vmov.f32 s4, s10
|
||||
@ -1189,17 +1189,17 @@ define void @vld3_v16f32(<48 x float> *%src, <16 x float> *%dst) {
|
||||
; CHECK-NEXT: .pad #16
|
||||
; CHECK-NEXT: sub sp, #16
|
||||
; CHECK-NEXT: vldrw.u32 q1, [r0, #48]
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #64]
|
||||
; CHECK-NEXT: vldrw.u32 q2, [r0, #80]
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #64]
|
||||
; CHECK-NEXT: vmov.f32 s12, s5
|
||||
; CHECK-NEXT: vmov.f32 s13, s0
|
||||
; CHECK-NEXT: vmov.32 r2, q2[2]
|
||||
; CHECK-NEXT: vmov.f32 s13, s0
|
||||
; CHECK-NEXT: vdup.32 q4, r2
|
||||
; CHECK-NEXT: vmov.f32 s14, s3
|
||||
; CHECK-NEXT: vmov.f32 s15, s19
|
||||
; CHECK-NEXT: vmov.32 r2, q2[1]
|
||||
; CHECK-NEXT: vmov.f64 d8, d2
|
||||
; CHECK-NEXT: vmov.f32 s15, s19
|
||||
; CHECK-NEXT: vdup.32 q5, r2
|
||||
; CHECK-NEXT: vmov.f64 d8, d2
|
||||
; CHECK-NEXT: vmov.f32 s17, s7
|
||||
; CHECK-NEXT: vmov.f32 s18, s2
|
||||
; CHECK-NEXT: vmov.f32 s0, s6
|
||||
@ -1222,6 +1222,7 @@ define void @vld3_v16f32(<48 x float> *%src, <16 x float> *%dst) {
|
||||
; CHECK-NEXT: vmov.32 r2, q3[1]
|
||||
; CHECK-NEXT: vmov.f64 d10, d4
|
||||
; CHECK-NEXT: vdup.32 q6, r2
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #128]
|
||||
; CHECK-NEXT: vmov.f32 s21, s11
|
||||
; CHECK-NEXT: vmov.f32 s22, s6
|
||||
; CHECK-NEXT: vmov.f32 s4, s10
|
||||
@ -1243,7 +1244,6 @@ define void @vld3_v16f32(<48 x float> *%src, <16 x float> *%dst) {
|
||||
; CHECK-NEXT: vmov.32 r2, q4[1]
|
||||
; CHECK-NEXT: vmov.f64 d12, d6
|
||||
; CHECK-NEXT: vdup.32 q7, r2
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #128]
|
||||
; CHECK-NEXT: vmov.f32 s25, s15
|
||||
; CHECK-NEXT: vmov.f32 s26, s10
|
||||
; CHECK-NEXT: vmov.f32 s8, s14
|
||||
|
||||
@ -22,10 +22,10 @@ define void @vld4_v2i32(<8 x i32> *%src, <2 x i32> *%dst) {
|
||||
; CHECK-NEXT: add r2, r3
|
||||
; CHECK-NEXT: vmov r3, s0
|
||||
; CHECK-NEXT: add r0, r2
|
||||
; CHECK-NEXT: str r0, [r1, #4]
|
||||
; CHECK-NEXT: vmov.32 r2, q0[2]
|
||||
; CHECK-NEXT: vmov r0, s8
|
||||
; CHECK-NEXT: vdup.32 q1, r2
|
||||
; CHECK-NEXT: str r0, [r1, #4]
|
||||
; CHECK-NEXT: vmov r0, s8
|
||||
; CHECK-NEXT: vmov r2, s4
|
||||
; CHECK-NEXT: add r0, r2
|
||||
; CHECK-NEXT: vmov r2, s12
|
||||
|
||||
@ -93,20 +93,21 @@ define void @vst3_v8i32(<8 x i32> *%src, <24 x i32> *%dst) {
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #80]
|
||||
; CHECK-NEXT: vldrw.u32 q5, [r0, #16]
|
||||
; CHECK-NEXT: vldrw.u32 q1, [r0, #48]
|
||||
; CHECK-NEXT: vmov.f64 d6, d1
|
||||
; CHECK-NEXT: vmov.32 r2, q1[3]
|
||||
; CHECK-NEXT: vldrw.u32 q4, [r0]
|
||||
; CHECK-NEXT: vmov.f64 d6, d1
|
||||
; CHECK-NEXT: vldrw.u32 q6, [r0, #32]
|
||||
; CHECK-NEXT: vdup.32 q2, r2
|
||||
; CHECK-NEXT: vmov.32 r2, q1[3]
|
||||
; CHECK-NEXT: vldrw.u32 q7, [r0, #64]
|
||||
; CHECK-NEXT: vdup.32 q2, r2
|
||||
; CHECK-NEXT: vstrw.32 q4, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vstrw.32 q5, [sp] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vmov.f32 s13, s23
|
||||
; CHECK-NEXT: vstrw.32 q6, [sp, #16] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.f32 s15, s3
|
||||
; CHECK-NEXT: vstrw.32 q5, [sp] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.f32 s14, s10
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vmov.f64 d4, d8
|
||||
; CHECK-NEXT: vstrw.32 q3, [r1, #80]
|
||||
; CHECK-NEXT: vmov.f32 s9, s24
|
||||
; CHECK-NEXT: vmov.f32 s11, s17
|
||||
; CHECK-NEXT: vmov q4, q5
|
||||
@ -120,20 +121,19 @@ define void @vst3_v8i32(<8 x i32> *%src, <24 x i32> *%dst) {
|
||||
; CHECK-NEXT: vmov.f32 s3, s6
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp] @ 16-byte Reload
|
||||
; CHECK-NEXT: vdup.32 q6, r0
|
||||
; CHECK-NEXT: vmov.32 r0, q4[3]
|
||||
; CHECK-NEXT: vmov.f32 s22, s26
|
||||
; CHECK-NEXT: vldrw.u32 q6, [sp, #32] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s2, s6
|
||||
; CHECK-NEXT: vmov.32 r0, q4[3]
|
||||
; CHECK-NEXT: vmov.f32 s4, s17
|
||||
; CHECK-NEXT: vstrw.32 q5, [r1, #48]
|
||||
; CHECK-NEXT: vmov.f32 s5, s29
|
||||
; CHECK-NEXT: vmov.f32 s4, s17
|
||||
; CHECK-NEXT: vstrw.32 q0, [r1, #64]
|
||||
; CHECK-NEXT: vmov.f32 s5, s29
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1]
|
||||
; CHECK-NEXT: vmov.f32 s28, s30
|
||||
; CHECK-NEXT: vstrw.32 q3, [r1, #80]
|
||||
; CHECK-NEXT: vmov.f32 s7, s18
|
||||
; CHECK-NEXT: vdup.32 q4, r0
|
||||
; CHECK-NEXT: vmov.f32 s29, s27
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1]
|
||||
; CHECK-NEXT: vmov.f32 s6, s26
|
||||
; CHECK-NEXT: vmov.f32 s30, s18
|
||||
; CHECK-NEXT: vstrw.32 q1, [r1, #16]
|
||||
@ -185,32 +185,32 @@ define void @vst3_v16i32(<16 x i32> *%src, <48 x i32> *%dst) {
|
||||
; CHECK-NEXT: vstrw.32 q2, [sp, #16] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.f64 d4, d15
|
||||
; CHECK-NEXT: vldrw.u32 q4, [r0, #176]
|
||||
; CHECK-NEXT: vstrw.32 q0, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vstrw.32 q1, [r1, #16]
|
||||
; CHECK-NEXT: vmov.32 r0, q6[3]
|
||||
; CHECK-NEXT: vstrw.32 q1, [r1, #16]
|
||||
; CHECK-NEXT: vdup.32 q1, r0
|
||||
; CHECK-NEXT: vmov.32 r0, q3[3]
|
||||
; CHECK-NEXT: vstrw.32 q0, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.f32 s9, s23
|
||||
; CHECK-NEXT: vmov.f32 s11, s31
|
||||
; CHECK-NEXT: vmov.f32 s10, s6
|
||||
; CHECK-NEXT: vmov.f64 d2, d9
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1, #32]
|
||||
; CHECK-NEXT: vmov.32 r0, q3[3]
|
||||
; CHECK-NEXT: vmov.f32 s5, s3
|
||||
; CHECK-NEXT: vmov.f32 s7, s19
|
||||
; CHECK-NEXT: vmov q2, q1
|
||||
; CHECK-NEXT: vdup.32 q1, r0
|
||||
; CHECK-NEXT: vmov.f32 s10, s6
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vmov.f64 d2, d10
|
||||
; CHECK-NEXT: vstrw.32 q2, [sp, #80] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vldrw.u32 q2, [sp, #32] @ 16-byte Reload
|
||||
; CHECK-NEXT: vdup.32 q0, r0
|
||||
; CHECK-NEXT: vmov.32 r0, q4[0]
|
||||
; CHECK-NEXT: vmov.f32 s5, s24
|
||||
; CHECK-NEXT: vmov.f32 s7, s21
|
||||
; CHECK-NEXT: vmov.f32 s6, s2
|
||||
; CHECK-NEXT: vmov.f64 d0, d4
|
||||
; CHECK-NEXT: vstrw.32 q1, [sp, #64] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r0, q4[0]
|
||||
; CHECK-NEXT: vmov.f32 s1, s12
|
||||
; CHECK-NEXT: vmov.f32 s3, s9
|
||||
; CHECK-NEXT: vmov q1, q0
|
||||
@ -227,9 +227,9 @@ define void @vst3_v16i32(<16 x i32> *%src, <48 x i32> *%dst) {
|
||||
; CHECK-NEXT: vmov.f32 s27, s6
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #16] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s18, s10
|
||||
; CHECK-NEXT: vmov.32 r0, q3[3]
|
||||
; CHECK-NEXT: vmov.f64 d4, d1
|
||||
; CHECK-NEXT: vstrw.32 q4, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r0, q3[3]
|
||||
; CHECK-NEXT: vmov q5, q1
|
||||
; CHECK-NEXT: vldrw.u32 q4, [sp, #144] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s9, s7
|
||||
@ -242,6 +242,8 @@ define void @vst3_v16i32(<16 x i32> *%src, <48 x i32> *%dst) {
|
||||
; CHECK-NEXT: vmov.f64 d6, d1
|
||||
; CHECK-NEXT: vmov.32 r0, q4[3]
|
||||
; CHECK-NEXT: vmov q7, q0
|
||||
; CHECK-NEXT: vstrw.32 q6, [r1, #112]
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1, #128]
|
||||
; CHECK-NEXT: vmov.f32 s13, s7
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #96] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s15, s3
|
||||
@ -249,18 +251,16 @@ define void @vst3_v16i32(<16 x i32> *%src, <48 x i32> *%dst) {
|
||||
; CHECK-NEXT: vmov.f32 s14, s2
|
||||
; CHECK-NEXT: vmov q0, q5
|
||||
; CHECK-NEXT: vmov.f32 s21, s4
|
||||
; CHECK-NEXT: vstrw.32 q3, [r1, #80]
|
||||
; CHECK-NEXT: vmov.f32 s23, s1
|
||||
; CHECK-NEXT: vldrw.u32 q0, [sp, #112] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.32 r0, q0[0]
|
||||
; CHECK-NEXT: vdup.32 q0, r0
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vmov.f32 s22, s2
|
||||
; CHECK-NEXT: vldrw.u32 q0, [sp, #128] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vstrw.32 q3, [r1, #80]
|
||||
; CHECK-NEXT: vmov.f64 d2, d0
|
||||
; CHECK-NEXT: vstrw.32 q5, [r1, #96]
|
||||
; CHECK-NEXT: vstrw.32 q6, [r1, #112]
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1, #128]
|
||||
; CHECK-NEXT: vmov.f64 d2, d0
|
||||
; CHECK-NEXT: vmov.f32 s5, s16
|
||||
; CHECK-NEXT: vmov q4, q0
|
||||
; CHECK-NEXT: vmov.f32 s7, s1
|
||||
@ -503,13 +503,14 @@ define void @vst3_v16i16(<16 x i16> *%src, <48 x i16> *%dst) {
|
||||
; CHECK-NEXT: vmov q7, q1
|
||||
; CHECK-NEXT: vmov.f32 s9, s0
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #80]
|
||||
; CHECK-NEXT: vstrw.32 q2, [sp, #96] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.u16 r2, q2[3]
|
||||
; CHECK-NEXT: vmov q3, q2
|
||||
; CHECK-NEXT: vmov.32 r3, q0[0]
|
||||
; CHECK-NEXT: vmov q1, q0
|
||||
; CHECK-NEXT: vdup.32 q0, r3
|
||||
; CHECK-NEXT: vmov q3, q2
|
||||
; CHECK-NEXT: vstrw.32 q2, [sp, #96] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.u16 r3, q0[2]
|
||||
; CHECK-NEXT: vstrw.32 q7, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.16 q2[2], r3
|
||||
; CHECK-NEXT: vmov.16 q2[3], r2
|
||||
; CHECK-NEXT: vmov.u16 r2, q3[4]
|
||||
@ -578,17 +579,17 @@ define void @vst3_v16i16(<16 x i16> *%src, <48 x i16> *%dst) {
|
||||
; CHECK-NEXT: vmov.16 q0[6], r0
|
||||
; CHECK-NEXT: vmov.u16 r0, q1[7]
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #128] @ 16-byte Reload
|
||||
; CHECK-NEXT: vstrw.32 q2, [sp, #64] @ 16-byte Spill
|
||||
; CHECK-NEXT: vstrw.32 q7, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.16 q0[7], r0
|
||||
; CHECK-NEXT: vmov.32 r2, q1[3]
|
||||
; CHECK-NEXT: vmov.f32 s2, s19
|
||||
; CHECK-NEXT: vdup.32 q7, r2
|
||||
; CHECK-NEXT: vstrw.32 q2, [sp, #64] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r2, q1[3]
|
||||
; CHECK-NEXT: vmov.u16 r0, q0[3]
|
||||
; CHECK-NEXT: vmov.u16 r2, q7[2]
|
||||
; CHECK-NEXT: vdup.32 q7, r2
|
||||
; CHECK-NEXT: vrev32.16 q3, q3
|
||||
; CHECK-NEXT: vmov.16 q2[2], r2
|
||||
; CHECK-NEXT: vmov.u16 r2, q7[2]
|
||||
; CHECK-NEXT: vstrw.32 q3, [sp] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.16 q2[2], r2
|
||||
; CHECK-NEXT: vstrw.32 q4, [sp, #16] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.16 q2[3], r0
|
||||
; CHECK-NEXT: vmov.u16 r0, q0[4]
|
||||
; CHECK-NEXT: vmov.16 q2[4], r0
|
||||
@ -603,7 +604,6 @@ define void @vst3_v16i16(<16 x i16> *%src, <48 x i16> *%dst) {
|
||||
; CHECK-NEXT: vmov.16 q3[6], r0
|
||||
; CHECK-NEXT: vmov.u16 r0, q1[5]
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #144] @ 16-byte Reload
|
||||
; CHECK-NEXT: vstrw.32 q4, [sp, #16] @ 16-byte Spill
|
||||
; CHECK-NEXT: vldrw.u32 q4, [sp, #48] @ 16-byte Reload
|
||||
; CHECK-NEXT: vrev32.16 q7, q7
|
||||
; CHECK-NEXT: vmov.16 q3[7], r0
|
||||
@ -611,34 +611,35 @@ define void @vst3_v16i16(<16 x i16> *%src, <48 x i16> *%dst) {
|
||||
; CHECK-NEXT: vstrw.32 q7, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.16 q7[0], r0
|
||||
; CHECK-NEXT: vmov.u16 r0, q4[3]
|
||||
; CHECK-NEXT: vmov.f32 s1, s9
|
||||
; CHECK-NEXT: vmov.16 q7[1], r0
|
||||
; CHECK-NEXT: vmov.u16 r0, q1[4]
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #64] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s1, s9
|
||||
; CHECK-NEXT: vmov.f32 s2, s10
|
||||
; CHECK-NEXT: vldrw.u32 q2, [sp, #96] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s25, s5
|
||||
; CHECK-NEXT: vmov.16 q7[6], r0
|
||||
; CHECK-NEXT: vmov.f32 s25, s5
|
||||
; CHECK-NEXT: vmov.u16 r0, q4[5]
|
||||
; CHECK-NEXT: vmov.f32 s26, s6
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #80] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.u16 r0, q4[5]
|
||||
; CHECK-NEXT: vstrw.32 q0, [r1, #80]
|
||||
; CHECK-NEXT: vmov.f32 s21, s5
|
||||
; CHECK-NEXT: vldrw.u32 q0, [sp] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s21, s5
|
||||
; CHECK-NEXT: vmov.16 q7[7], r0
|
||||
; CHECK-NEXT: vmov.f32 s22, s6
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #112] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.16 q7[7], r0
|
||||
; CHECK-NEXT: vmov.u16 r2, q0[2]
|
||||
; CHECK-NEXT: vmov.f32 s9, s5
|
||||
; CHECK-NEXT: vmov.u16 r0, q0[5]
|
||||
; CHECK-NEXT: vmov.f32 s9, s5
|
||||
; CHECK-NEXT: vmov.16 q0[2], r2
|
||||
; CHECK-NEXT: vmov.f32 s10, s6
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #144] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.16 q0[2], r2
|
||||
; CHECK-NEXT: vstrw.32 q5, [r1, #32]
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1, #48]
|
||||
; CHECK-NEXT: vmov.f32 s29, s5
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #16] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s30, s18
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1, #48]
|
||||
; CHECK-NEXT: vstrw.32 q6, [r1]
|
||||
; CHECK-NEXT: vmov.u16 r2, q7[3]
|
||||
; CHECK-NEXT: vmov.f32 s13, s5
|
||||
; CHECK-NEXT: vmov.16 q0[3], r2
|
||||
@ -646,7 +647,6 @@ define void @vst3_v16i16(<16 x i16> *%src, <48 x i16> *%dst) {
|
||||
; CHECK-NEXT: vmov.16 q0[4], r2
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #128] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.16 q0[5], r0
|
||||
; CHECK-NEXT: vstrw.32 q6, [r1]
|
||||
; CHECK-NEXT: vmov.f32 s29, s1
|
||||
; CHECK-NEXT: vmov.f32 s30, s2
|
||||
; CHECK-NEXT: vldrw.u32 q0, [sp, #32] @ 16-byte Reload
|
||||
@ -1251,20 +1251,21 @@ define void @vst3_v8f32(<8 x float> *%src, <24 x float> *%dst) {
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #80]
|
||||
; CHECK-NEXT: vldrw.u32 q5, [r0, #16]
|
||||
; CHECK-NEXT: vldrw.u32 q1, [r0, #48]
|
||||
; CHECK-NEXT: vmov.f64 d6, d1
|
||||
; CHECK-NEXT: vmov.32 r2, q1[3]
|
||||
; CHECK-NEXT: vldrw.u32 q4, [r0]
|
||||
; CHECK-NEXT: vmov.f64 d6, d1
|
||||
; CHECK-NEXT: vldrw.u32 q6, [r0, #32]
|
||||
; CHECK-NEXT: vdup.32 q2, r2
|
||||
; CHECK-NEXT: vmov.32 r2, q1[3]
|
||||
; CHECK-NEXT: vldrw.u32 q7, [r0, #64]
|
||||
; CHECK-NEXT: vdup.32 q2, r2
|
||||
; CHECK-NEXT: vstrw.32 q4, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vstrw.32 q5, [sp] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vmov.f32 s13, s23
|
||||
; CHECK-NEXT: vstrw.32 q6, [sp, #16] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.f32 s15, s3
|
||||
; CHECK-NEXT: vstrw.32 q5, [sp] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.f32 s14, s10
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vmov.f64 d4, d8
|
||||
; CHECK-NEXT: vstrw.32 q3, [r1, #80]
|
||||
; CHECK-NEXT: vmov.f32 s9, s24
|
||||
; CHECK-NEXT: vmov.f32 s11, s17
|
||||
; CHECK-NEXT: vmov q4, q5
|
||||
@ -1278,20 +1279,19 @@ define void @vst3_v8f32(<8 x float> *%src, <24 x float> *%dst) {
|
||||
; CHECK-NEXT: vmov.f32 s3, s6
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp] @ 16-byte Reload
|
||||
; CHECK-NEXT: vdup.32 q6, r0
|
||||
; CHECK-NEXT: vmov.32 r0, q4[3]
|
||||
; CHECK-NEXT: vmov.f32 s22, s26
|
||||
; CHECK-NEXT: vldrw.u32 q6, [sp, #32] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s2, s6
|
||||
; CHECK-NEXT: vmov.32 r0, q4[3]
|
||||
; CHECK-NEXT: vmov.f32 s4, s17
|
||||
; CHECK-NEXT: vstrw.32 q5, [r1, #48]
|
||||
; CHECK-NEXT: vmov.f32 s5, s29
|
||||
; CHECK-NEXT: vmov.f32 s4, s17
|
||||
; CHECK-NEXT: vstrw.32 q0, [r1, #64]
|
||||
; CHECK-NEXT: vmov.f32 s5, s29
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1]
|
||||
; CHECK-NEXT: vmov.f32 s28, s30
|
||||
; CHECK-NEXT: vstrw.32 q3, [r1, #80]
|
||||
; CHECK-NEXT: vmov.f32 s7, s18
|
||||
; CHECK-NEXT: vdup.32 q4, r0
|
||||
; CHECK-NEXT: vmov.f32 s29, s27
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1]
|
||||
; CHECK-NEXT: vmov.f32 s6, s26
|
||||
; CHECK-NEXT: vmov.f32 s30, s18
|
||||
; CHECK-NEXT: vstrw.32 q1, [r1, #16]
|
||||
@ -1343,32 +1343,32 @@ define void @vst3_v16f32(<16 x float> *%src, <48 x float> *%dst) {
|
||||
; CHECK-NEXT: vstrw.32 q2, [sp, #16] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.f64 d4, d15
|
||||
; CHECK-NEXT: vldrw.u32 q4, [r0, #176]
|
||||
; CHECK-NEXT: vstrw.32 q0, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vstrw.32 q1, [r1, #16]
|
||||
; CHECK-NEXT: vmov.32 r0, q6[3]
|
||||
; CHECK-NEXT: vstrw.32 q1, [r1, #16]
|
||||
; CHECK-NEXT: vdup.32 q1, r0
|
||||
; CHECK-NEXT: vmov.32 r0, q3[3]
|
||||
; CHECK-NEXT: vstrw.32 q0, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.f32 s9, s23
|
||||
; CHECK-NEXT: vmov.f32 s11, s31
|
||||
; CHECK-NEXT: vmov.f32 s10, s6
|
||||
; CHECK-NEXT: vmov.f64 d2, d9
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1, #32]
|
||||
; CHECK-NEXT: vmov.32 r0, q3[3]
|
||||
; CHECK-NEXT: vmov.f32 s5, s3
|
||||
; CHECK-NEXT: vmov.f32 s7, s19
|
||||
; CHECK-NEXT: vmov q2, q1
|
||||
; CHECK-NEXT: vdup.32 q1, r0
|
||||
; CHECK-NEXT: vmov.f32 s10, s6
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vmov.f64 d2, d10
|
||||
; CHECK-NEXT: vstrw.32 q2, [sp, #80] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vldrw.u32 q2, [sp, #32] @ 16-byte Reload
|
||||
; CHECK-NEXT: vdup.32 q0, r0
|
||||
; CHECK-NEXT: vmov.32 r0, q4[0]
|
||||
; CHECK-NEXT: vmov.f32 s5, s24
|
||||
; CHECK-NEXT: vmov.f32 s7, s21
|
||||
; CHECK-NEXT: vmov.f32 s6, s2
|
||||
; CHECK-NEXT: vmov.f64 d0, d4
|
||||
; CHECK-NEXT: vstrw.32 q1, [sp, #64] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r0, q4[0]
|
||||
; CHECK-NEXT: vmov.f32 s1, s12
|
||||
; CHECK-NEXT: vmov.f32 s3, s9
|
||||
; CHECK-NEXT: vmov q1, q0
|
||||
@ -1385,9 +1385,9 @@ define void @vst3_v16f32(<16 x float> *%src, <48 x float> *%dst) {
|
||||
; CHECK-NEXT: vmov.f32 s27, s6
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #16] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s18, s10
|
||||
; CHECK-NEXT: vmov.32 r0, q3[3]
|
||||
; CHECK-NEXT: vmov.f64 d4, d1
|
||||
; CHECK-NEXT: vstrw.32 q4, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r0, q3[3]
|
||||
; CHECK-NEXT: vmov q5, q1
|
||||
; CHECK-NEXT: vldrw.u32 q4, [sp, #144] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s9, s7
|
||||
@ -1400,6 +1400,8 @@ define void @vst3_v16f32(<16 x float> *%src, <48 x float> *%dst) {
|
||||
; CHECK-NEXT: vmov.f64 d6, d1
|
||||
; CHECK-NEXT: vmov.32 r0, q4[3]
|
||||
; CHECK-NEXT: vmov q7, q0
|
||||
; CHECK-NEXT: vstrw.32 q6, [r1, #112]
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1, #128]
|
||||
; CHECK-NEXT: vmov.f32 s13, s7
|
||||
; CHECK-NEXT: vldrw.u32 q1, [sp, #96] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.f32 s15, s3
|
||||
@ -1407,18 +1409,16 @@ define void @vst3_v16f32(<16 x float> *%src, <48 x float> *%dst) {
|
||||
; CHECK-NEXT: vmov.f32 s14, s2
|
||||
; CHECK-NEXT: vmov q0, q5
|
||||
; CHECK-NEXT: vmov.f32 s21, s4
|
||||
; CHECK-NEXT: vstrw.32 q3, [r1, #80]
|
||||
; CHECK-NEXT: vmov.f32 s23, s1
|
||||
; CHECK-NEXT: vldrw.u32 q0, [sp, #112] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.32 r0, q0[0]
|
||||
; CHECK-NEXT: vdup.32 q0, r0
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vmov.f32 s22, s2
|
||||
; CHECK-NEXT: vldrw.u32 q0, [sp, #128] @ 16-byte Reload
|
||||
; CHECK-NEXT: vmov.32 r0, q7[0]
|
||||
; CHECK-NEXT: vstrw.32 q3, [r1, #80]
|
||||
; CHECK-NEXT: vmov.f64 d2, d0
|
||||
; CHECK-NEXT: vstrw.32 q5, [r1, #96]
|
||||
; CHECK-NEXT: vstrw.32 q6, [r1, #112]
|
||||
; CHECK-NEXT: vstrw.32 q2, [r1, #128]
|
||||
; CHECK-NEXT: vmov.f64 d2, d0
|
||||
; CHECK-NEXT: vmov.f32 s5, s16
|
||||
; CHECK-NEXT: vmov q4, q0
|
||||
; CHECK-NEXT: vmov.f32 s7, s1
|
||||
@ -1573,11 +1573,11 @@ define void @vst3_v8f16(<8 x half> *%src, <24 x half> *%dst) {
|
||||
; CHECK-NEXT: vmovx.f16 s12, s4
|
||||
; CHECK-NEXT: vmov r2, s4
|
||||
; CHECK-NEXT: vmov.16 q0[0], r3
|
||||
; CHECK-NEXT: vmov.32 r0, q5[0]
|
||||
; CHECK-NEXT: vmov.16 q0[1], r2
|
||||
; CHECK-NEXT: vmov r2, s12
|
||||
; CHECK-NEXT: vdup.32 q4, r0
|
||||
; CHECK-NEXT: vmov.32 r0, q5[0]
|
||||
; CHECK-NEXT: vmov.16 q0[4], r2
|
||||
; CHECK-NEXT: vdup.32 q4, r0
|
||||
; CHECK-NEXT: vmov r2, s9
|
||||
; CHECK-NEXT: vmov.16 q0[6], r2
|
||||
; CHECK-NEXT: vmov r2, s5
|
||||
@ -1606,17 +1606,17 @@ define void @vst3_v8f16(<8 x half> *%src, <24 x half> *%dst) {
|
||||
; CHECK-NEXT: vmovx.f16 s24, s23
|
||||
; CHECK-NEXT: vmov.16 q4[6], r0
|
||||
; CHECK-NEXT: vmov r0, s24
|
||||
; CHECK-NEXT: vstrw.32 q1, [sp] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.16 q4[7], r0
|
||||
; CHECK-NEXT: vmov.32 r2, q2[3]
|
||||
; CHECK-NEXT: vmov.f32 s18, s23
|
||||
; CHECK-NEXT: vmov.16 q4[7], r0
|
||||
; CHECK-NEXT: vdup.32 q7, r2
|
||||
; CHECK-NEXT: vmov.f32 s18, s23
|
||||
; CHECK-NEXT: vmovx.f16 s24, s17
|
||||
; CHECK-NEXT: vmov r2, s29
|
||||
; CHECK-NEXT: vmovx.f16 s28, s30
|
||||
; CHECK-NEXT: vmov r0, s24
|
||||
; CHECK-NEXT: vmov.16 q6[2], r2
|
||||
; CHECK-NEXT: vmov.16 q6[3], r0
|
||||
; CHECK-NEXT: vmovx.f16 s28, s30
|
||||
; CHECK-NEXT: vstrw.32 q1, [sp] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmovx.f16 s4, s10
|
||||
; CHECK-NEXT: vmov.f32 s1, s13
|
||||
; CHECK-NEXT: vmov.f32 s2, s14
|
||||
@ -1696,12 +1696,12 @@ define void @vst3_v16f16(<16 x half> *%src, <48 x half> *%dst) {
|
||||
; CHECK-NEXT: vmov r2, s0
|
||||
; CHECK-NEXT: vldrw.u32 q0, [r0, #64]
|
||||
; CHECK-NEXT: vmov.f32 s5, s12
|
||||
; CHECK-NEXT: vmov q5, q3
|
||||
; CHECK-NEXT: vstrw.32 q3, [sp, #32] @ 16-byte Spill
|
||||
; CHECK-NEXT: vstrw.32 q1, [sp, #128] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r3, q0[0]
|
||||
; CHECK-NEXT: vmov q4, q0
|
||||
; CHECK-NEXT: vdup.32 q0, r3
|
||||
; CHECK-NEXT: vmov q5, q3
|
||||
; CHECK-NEXT: vstrw.32 q1, [sp, #128] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov r3, s1
|
||||
; CHECK-NEXT: vmovx.f16 s0, s2
|
||||
; CHECK-NEXT: vmov.16 q3[2], r3
|
||||
@ -1725,23 +1725,23 @@ define void @vst3_v16f16(<16 x half> *%src, <48 x half> *%dst) {
|
||||
; CHECK-NEXT: vmov.16 q1[6], r2
|
||||
; CHECK-NEXT: vmov r2, s0
|
||||
; CHECK-NEXT: vmov.16 q1[7], r2
|
||||
; CHECK-NEXT: vstrw.32 q3, [sp, #112] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r3, q5[3]
|
||||
; CHECK-NEXT: vmov.f32 s6, s19
|
||||
; CHECK-NEXT: vmovx.f16 s0, s5
|
||||
; CHECK-NEXT: vstrw.32 q4, [sp, #48] @ 16-byte Spill
|
||||
; CHECK-NEXT: vstrw.32 q1, [sp, #96] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.32 r3, q5[3]
|
||||
; CHECK-NEXT: vmov r2, s0
|
||||
; CHECK-NEXT: vdup.32 q0, r3
|
||||
; CHECK-NEXT: vldrw.u32 q5, [r0, #16]
|
||||
; CHECK-NEXT: vmov r3, s1
|
||||
; CHECK-NEXT: vmovx.f16 s0, s2
|
||||
; CHECK-NEXT: vmov.16 q2[2], r3
|
||||
; CHECK-NEXT: vmov r3, s20
|
||||
; CHECK-NEXT: vldrw.u32 q5, [r0, #16]
|
||||
; CHECK-NEXT: vmov.16 q2[3], r2
|
||||
; CHECK-NEXT: vmov.16 q7[0], r3
|
||||
; CHECK-NEXT: vmov r2, s6
|
||||
; CHECK-NEXT: vstrw.32 q3, [sp, #112] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov r3, s20
|
||||
; CHECK-NEXT: vldrw.u32 q3, [r0, #80]
|
||||
; CHECK-NEXT: vmov.16 q7[0], r3
|
||||
; CHECK-NEXT: vstrw.32 q1, [sp, #96] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov r2, s6
|
||||
; CHECK-NEXT: vstrw.32 q4, [sp, #48] @ 16-byte Spill
|
||||
; CHECK-NEXT: vmov.16 q2[4], r2
|
||||
; CHECK-NEXT: vmov r2, s0
|
||||
; CHECK-NEXT: vmov.16 q2[5], r2
|
||||
|
||||
@ -936,59 +936,14 @@ TEST(MachineInstrValidTailPredication, IsCorrect) {
|
||||
|
||||
TEST(MachineInstr, HasSideEffects) {
|
||||
using namespace ARM;
|
||||
unsigned Opcodes[] = {
|
||||
// MVE Loads/Stores
|
||||
MVE_VLDRBS16, MVE_VLDRBS16_post, MVE_VLDRBS16_pre,
|
||||
MVE_VLDRBS16_rq, MVE_VLDRBS32, MVE_VLDRBS32_post,
|
||||
MVE_VLDRBS32_pre, MVE_VLDRBS32_rq, MVE_VLDRBU16,
|
||||
MVE_VLDRBU16_post, MVE_VLDRBU16_pre, MVE_VLDRBU16_rq,
|
||||
MVE_VLDRBU32, MVE_VLDRBU32_post, MVE_VLDRBU32_pre,
|
||||
MVE_VLDRBU32_rq, MVE_VLDRBU8, MVE_VLDRBU8_post,
|
||||
MVE_VLDRBU8_pre, MVE_VLDRBU8_rq, MVE_VLDRDU64_qi,
|
||||
MVE_VLDRDU64_qi_pre, MVE_VLDRDU64_rq, MVE_VLDRDU64_rq_u,
|
||||
MVE_VLDRHS32, MVE_VLDRHS32_post, MVE_VLDRHS32_pre,
|
||||
MVE_VLDRHS32_rq, MVE_VLDRHS32_rq_u, MVE_VLDRHU16,
|
||||
MVE_VLDRHU16_post, MVE_VLDRHU16_pre, MVE_VLDRHU16_rq,
|
||||
MVE_VLDRHU16_rq_u, MVE_VLDRHU32, MVE_VLDRHU32_post,
|
||||
MVE_VLDRHU32_pre, MVE_VLDRHU32_rq, MVE_VLDRHU32_rq_u,
|
||||
MVE_VLDRWU32, MVE_VLDRWU32_post, MVE_VLDRWU32_pre,
|
||||
MVE_VLDRWU32_qi, MVE_VLDRWU32_qi_pre, MVE_VLDRWU32_rq,
|
||||
MVE_VLDRWU32_rq_u, MVE_VLD20_16, MVE_VLD20_16_wb,
|
||||
MVE_VLD20_32, MVE_VLD20_32_wb, MVE_VLD20_8,
|
||||
MVE_VLD20_8_wb, MVE_VLD21_16, MVE_VLD21_16_wb,
|
||||
MVE_VLD21_32, MVE_VLD21_32_wb, MVE_VLD21_8,
|
||||
MVE_VLD21_8_wb, MVE_VLD40_16, MVE_VLD40_16_wb,
|
||||
MVE_VLD40_32, MVE_VLD40_32_wb, MVE_VLD40_8,
|
||||
MVE_VLD40_8_wb, MVE_VLD41_16, MVE_VLD41_16_wb,
|
||||
MVE_VLD41_32, MVE_VLD41_32_wb, MVE_VLD41_8,
|
||||
MVE_VLD41_8_wb, MVE_VLD42_16, MVE_VLD42_16_wb,
|
||||
MVE_VLD42_32, MVE_VLD42_32_wb, MVE_VLD42_8,
|
||||
MVE_VLD42_8_wb, MVE_VLD43_16, MVE_VLD43_16_wb,
|
||||
MVE_VLD43_32, MVE_VLD43_32_wb, MVE_VLD43_8,
|
||||
MVE_VLD43_8_wb, MVE_VSTRB16, MVE_VSTRB16_post,
|
||||
MVE_VSTRB16_pre, MVE_VSTRB16_rq, MVE_VSTRB32,
|
||||
MVE_VSTRB32_post, MVE_VSTRB32_pre, MVE_VSTRB32_rq,
|
||||
MVE_VSTRB8_rq, MVE_VSTRBU8, MVE_VSTRBU8_post,
|
||||
MVE_VSTRBU8_pre, MVE_VSTRD64_qi, MVE_VSTRD64_qi_pre,
|
||||
MVE_VSTRD64_rq, MVE_VSTRD64_rq_u, MVE_VSTRH16_rq,
|
||||
MVE_VSTRH16_rq_u, MVE_VSTRH32, MVE_VSTRH32_post,
|
||||
MVE_VSTRH32_pre, MVE_VSTRH32_rq, MVE_VSTRH32_rq_u,
|
||||
MVE_VSTRHU16, MVE_VSTRHU16_post, MVE_VSTRHU16_pre,
|
||||
MVE_VSTRW32_qi, MVE_VSTRW32_qi_pre, MVE_VSTRW32_rq,
|
||||
MVE_VSTRW32_rq_u, MVE_VSTRWU32, MVE_VSTRWU32_post,
|
||||
MVE_VSTRWU32_pre, MVE_VST20_16, MVE_VST20_16_wb,
|
||||
MVE_VST20_32, MVE_VST20_32_wb, MVE_VST20_8,
|
||||
MVE_VST20_8_wb, MVE_VST21_16, MVE_VST21_16_wb,
|
||||
MVE_VST21_32, MVE_VST21_32_wb, MVE_VST21_8,
|
||||
MVE_VST21_8_wb, MVE_VST40_16, MVE_VST40_16_wb,
|
||||
MVE_VST40_32, MVE_VST40_32_wb, MVE_VST40_8,
|
||||
MVE_VST40_8_wb, MVE_VST41_16, MVE_VST41_16_wb,
|
||||
MVE_VST41_32, MVE_VST41_32_wb, MVE_VST41_8,
|
||||
MVE_VST41_8_wb, MVE_VST42_16, MVE_VST42_16_wb,
|
||||
MVE_VST42_32, MVE_VST42_32_wb, MVE_VST42_8,
|
||||
MVE_VST42_8_wb, MVE_VST43_16, MVE_VST43_16_wb,
|
||||
MVE_VST43_32, MVE_VST43_32_wb, MVE_VST43_8,
|
||||
MVE_VST43_8_wb,
|
||||
std::set<unsigned> UnpredictableOpcodes = {
|
||||
MVE_VCTP8, MVE_VCTP16, MVE_VCTP32, MVE_VCTP64, MVE_VPST,
|
||||
MVE_VPTv16i8, MVE_VPTv8i16, MVE_VPTv4i32, MVE_VPTv16i8r, MVE_VPTv8i16r,
|
||||
MVE_VPTv4i32r, MVE_VPTv16s8, MVE_VPTv8s16, MVE_VPTv4s32, MVE_VPTv16s8r,
|
||||
MVE_VPTv8s16r, MVE_VPTv4s32r, MVE_VPTv16u8, MVE_VPTv8u16, MVE_VPTv4u32,
|
||||
MVE_VPTv16u8r, MVE_VPTv8u16r, MVE_VPTv4u32r, MVE_VPTv8f16, MVE_VPTv4f32,
|
||||
MVE_VPTv8f16r, MVE_VPTv4f32r, MVE_VADC, MVE_VADCI, MVE_VSBC,
|
||||
MVE_VSBCI, MVE_VSHLC,
|
||||
};
|
||||
|
||||
LLVMInitializeARMTargetInfo();
|
||||
@ -1013,8 +968,13 @@ TEST(MachineInstr, HasSideEffects) {
|
||||
const ARMBaseInstrInfo *TII = ST.getInstrInfo();
|
||||
auto MII = TM->getMCInstrInfo();
|
||||
|
||||
for (unsigned Op : Opcodes) {
|
||||
for (unsigned Op = 0; Op < ARM::INSTRUCTION_LIST_END; ++Op) {
|
||||
const MCInstrDesc &Desc = TII->get(Op);
|
||||
if ((Desc.TSFlags & ARMII::DomainMask) != ARMII::DomainMVE)
|
||||
continue;
|
||||
if (UnpredictableOpcodes.count(Op))
|
||||
continue;
|
||||
|
||||
ASSERT_FALSE(Desc.hasUnmodeledSideEffects())
|
||||
<< MII->getName(Op) << " has unexpected side effects";
|
||||
}
|
||||
|
||||
Loading…
Reference in New Issue
Block a user