@@ -1344,6 +1344,16 @@ vpmulld %ymm16, %ymm17, %ymm19 {z}{k1}
1344
1344
vpmulld (%rax ), %ymm17, %ymm19 {z}{k1}
1345
1345
vpmulld (%rax ){1to8}, %ymm17, %ymm19 {z}{k1}
1346
1346
1347
+ vpscatterdd %xmm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1348
+ vpscatterdq %xmm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1349
+ vpscatterqd %xmm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1350
+ vpscatterqq %xmm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1351
+
1352
+ vpscatterdd %ymm1 , (%rdx ,%ymm0 ,4 ) {%k1 }
1353
+ vpscatterdq %ymm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1354
+ vpscatterqd %xmm1 , (%rdx ,%ymm0 ,4 ) {%k1 }
1355
+ vpscatterqq %ymm1 , (%rdx ,%ymm0 ,4 ) {%k1 }
1356
+
1347
1357
vpshufd $0 , %xmm16, %xmm19
1348
1358
vpshufd $0 , (%rax ), %xmm19
1349
1359
vpshufd $0 , (%rax ){1to4}, %xmm19
@@ -1500,6 +1510,16 @@ vpunpckldq %ymm16, %ymm17, %ymm19 {z}{k1}
1500
1510
vpunpckldq (%rax ), %ymm17, %ymm19 {z}{k1}
1501
1511
vpunpckldq (%rax ){1to8}, %ymm17, %ymm19 {z}{k1}
1502
1512
1513
+ vscatterdps %xmm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1514
+ vscatterdpd %xmm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1515
+ vscatterqps %xmm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1516
+ vscatterqpd %xmm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1517
+
1518
+ vscatterdps %ymm1 , (%rdx ,%ymm0 ,4 ) {%k1 }
1519
+ vscatterdpd %ymm1 , (%rdx ,%xmm0 ,4 ) {%k1 }
1520
+ vscatterqps %xmm1 , (%rdx ,%ymm0 ,4 ) {%k1 }
1521
+ vscatterqpd %ymm1 , (%rdx ,%ymm0 ,4 ) {%k1 }
1522
+
1503
1523
vshuff32x4 $0 , %ymm16, %ymm17, %ymm19
1504
1524
vshuff32x4 $0 , (%rax ), %ymm17, %ymm19
1505
1525
vshuff32x4 $0 , (%rax ){1to8}, %ymm17, %ymm19
@@ -2897,6 +2917,14 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
2897
2917
# CHECK-NEXT: 2 10 1.00 vpmulld %ymm16, %ymm17, %ymm19 {%k1} {z}
2898
2918
# CHECK-NEXT: 3 17 1.00 * vpmulld (%rax), %ymm17, %ymm19 {%k1} {z}
2899
2919
# CHECK-NEXT: 3 17 1.00 * vpmulld (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
2920
+ # CHECK-NEXT: 12 8 2.00 * vpscatterdd %xmm1, (%rdx,%xmm0,4) {%k1}
2921
+ # CHECK-NEXT: 7 7 1.00 * vpscatterdq %xmm1, (%rdx,%xmm0,4) {%k1}
2922
+ # CHECK-NEXT: 8 8 1.00 * vpscatterqd %xmm1, (%rdx,%xmm0,4) {%k1}
2923
+ # CHECK-NEXT: 7 7 1.00 * vpscatterqq %xmm1, (%rdx,%xmm0,4) {%k1}
2924
+ # CHECK-NEXT: 20 8 4.00 * vpscatterdd %ymm1, (%rdx,%ymm0,4) {%k1}
2925
+ # CHECK-NEXT: 11 7 2.00 * vpscatterdq %ymm1, (%rdx,%xmm0,4) {%k1}
2926
+ # CHECK-NEXT: 8 8 1.00 * vpscatterqd %xmm1, (%rdx,%ymm0,4) {%k1}
2927
+ # CHECK-NEXT: 11 7 2.00 * vpscatterqq %ymm1, (%rdx,%ymm0,4) {%k1}
2900
2928
# CHECK-NEXT: 1 1 0.50 vpshufd $0, %xmm16, %xmm19
2901
2929
# CHECK-NEXT: 2 7 0.50 * vpshufd $0, (%rax), %xmm19
2902
2930
# CHECK-NEXT: 2 7 0.50 * vpshufd $0, (%rax){1to4}, %xmm19
@@ -3035,6 +3063,14 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
3035
3063
# CHECK-NEXT: 1 1 0.50 vpunpckldq %ymm16, %ymm17, %ymm19 {%k1} {z}
3036
3064
# CHECK-NEXT: 2 8 0.50 * vpunpckldq (%rax), %ymm17, %ymm19 {%k1} {z}
3037
3065
# CHECK-NEXT: 2 8 0.50 * vpunpckldq (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
3066
+ # CHECK-NEXT: 12 8 2.00 * vscatterdps %xmm1, (%rdx,%xmm0,4) {%k1}
3067
+ # CHECK-NEXT: 7 7 1.00 * vscatterdpd %xmm1, (%rdx,%xmm0,4) {%k1}
3068
+ # CHECK-NEXT: 8 8 1.00 * vscatterqps %xmm1, (%rdx,%xmm0,4) {%k1}
3069
+ # CHECK-NEXT: 7 7 1.00 * vscatterqpd %xmm1, (%rdx,%xmm0,4) {%k1}
3070
+ # CHECK-NEXT: 20 8 4.00 * vscatterdps %ymm1, (%rdx,%ymm0,4) {%k1}
3071
+ # CHECK-NEXT: 11 7 2.00 * vscatterdpd %ymm1, (%rdx,%xmm0,4) {%k1}
3072
+ # CHECK-NEXT: 8 8 1.00 * vscatterqps %xmm1, (%rdx,%ymm0,4) {%k1}
3073
+ # CHECK-NEXT: 11 7 2.00 * vscatterqpd %ymm1, (%rdx,%ymm0,4) {%k1}
3038
3074
# CHECK-NEXT: 1 3 1.00 vshuff32x4 $0, %ymm16, %ymm17, %ymm19
3039
3075
# CHECK-NEXT: 2 10 1.00 * vshuff32x4 $0, (%rax), %ymm17, %ymm19
3040
3076
# CHECK-NEXT: 2 10 1.00 * vshuff32x4 $0, (%rax){1to8}, %ymm17, %ymm19
@@ -3232,7 +3268,7 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
3232
3268
3233
3269
# CHECK: Resource pressure per iteration:
3234
3270
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]
3235
- # CHECK-NEXT: - 423.00 438 .33 413 .33 492.50 492.50 16 .00 722 .33 4 .00 16 .00 16 .00 16 .00
3271
+ # CHECK-NEXT: - 423.00 462 .33 421 .33 492.50 492.50 44 .00 738 .33 12 .00 44 .00 44 .00 44 .00
3236
3272
3237
3273
# CHECK: Resource pressure by instruction:
3238
3274
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] Instructions:
@@ -4424,6 +4460,14 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
4424
4460
# CHECK-NEXT: - - 1.00 1.00 - - - - - - - - vpmulld %ymm16, %ymm17, %ymm19 {%k1} {z}
4425
4461
# CHECK-NEXT: - - 1.00 1.00 0.50 0.50 - - - - - - vpmulld (%rax), %ymm17, %ymm19 {%k1} {z}
4426
4462
# CHECK-NEXT: - - 1.00 1.00 0.50 0.50 - - - - - - vpmulld (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
4463
+ # CHECK-NEXT: - - 1.50 0.50 - - 2.00 1.50 0.50 2.00 2.00 2.00 vpscatterdd %xmm1, (%rdx,%xmm0,4) {%k1}
4464
+ # CHECK-NEXT: - - 1.50 0.50 - - 1.00 0.50 0.50 1.00 1.00 1.00 vpscatterdq %xmm1, (%rdx,%xmm0,4) {%k1}
4465
+ # CHECK-NEXT: - - 1.50 0.50 - - 1.00 1.50 0.50 1.00 1.00 1.00 vpscatterqd %xmm1, (%rdx,%xmm0,4) {%k1}
4466
+ # CHECK-NEXT: - - 1.50 0.50 - - 1.00 0.50 0.50 1.00 1.00 1.00 vpscatterqq %xmm1, (%rdx,%xmm0,4) {%k1}
4467
+ # CHECK-NEXT: - - 1.50 0.50 - - 4.00 1.50 0.50 4.00 4.00 4.00 vpscatterdd %ymm1, (%rdx,%ymm0,4) {%k1}
4468
+ # CHECK-NEXT: - - 1.50 0.50 - - 2.00 0.50 0.50 2.00 2.00 2.00 vpscatterdq %ymm1, (%rdx,%xmm0,4) {%k1}
4469
+ # CHECK-NEXT: - - 1.50 0.50 - - 1.00 1.50 0.50 1.00 1.00 1.00 vpscatterqd %xmm1, (%rdx,%ymm0,4) {%k1}
4470
+ # CHECK-NEXT: - - 1.50 0.50 - - 2.00 0.50 0.50 2.00 2.00 2.00 vpscatterqq %ymm1, (%rdx,%ymm0,4) {%k1}
4427
4471
# CHECK-NEXT: - - - 0.50 - - - 0.50 - - - - vpshufd $0, %xmm16, %xmm19
4428
4472
# CHECK-NEXT: - - - 0.50 0.50 0.50 - 0.50 - - - - vpshufd $0, (%rax), %xmm19
4429
4473
# CHECK-NEXT: - - - 0.50 0.50 0.50 - 0.50 - - - - vpshufd $0, (%rax){1to4}, %xmm19
@@ -4562,6 +4606,14 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
4562
4606
# CHECK-NEXT: - - - 0.50 - - - 0.50 - - - - vpunpckldq %ymm16, %ymm17, %ymm19 {%k1} {z}
4563
4607
# CHECK-NEXT: - - - 0.50 0.50 0.50 - 0.50 - - - - vpunpckldq (%rax), %ymm17, %ymm19 {%k1} {z}
4564
4608
# CHECK-NEXT: - - - 0.50 0.50 0.50 - 0.50 - - - - vpunpckldq (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
4609
+ # CHECK-NEXT: - - 1.50 0.50 - - 2.00 1.50 0.50 2.00 2.00 2.00 vscatterdps %xmm1, (%rdx,%xmm0,4) {%k1}
4610
+ # CHECK-NEXT: - - 1.50 0.50 - - 1.00 0.50 0.50 1.00 1.00 1.00 vscatterdpd %xmm1, (%rdx,%xmm0,4) {%k1}
4611
+ # CHECK-NEXT: - - 1.50 0.50 - - 1.00 1.50 0.50 1.00 1.00 1.00 vscatterqps %xmm1, (%rdx,%xmm0,4) {%k1}
4612
+ # CHECK-NEXT: - - 1.50 0.50 - - 1.00 0.50 0.50 1.00 1.00 1.00 vscatterqpd %xmm1, (%rdx,%xmm0,4) {%k1}
4613
+ # CHECK-NEXT: - - 1.50 0.50 - - 4.00 1.50 0.50 4.00 4.00 4.00 vscatterdps %ymm1, (%rdx,%ymm0,4) {%k1}
4614
+ # CHECK-NEXT: - - 1.50 0.50 - - 2.00 0.50 0.50 2.00 2.00 2.00 vscatterdpd %ymm1, (%rdx,%xmm0,4) {%k1}
4615
+ # CHECK-NEXT: - - 1.50 0.50 - - 1.00 1.50 0.50 1.00 1.00 1.00 vscatterqps %xmm1, (%rdx,%ymm0,4) {%k1}
4616
+ # CHECK-NEXT: - - 1.50 0.50 - - 2.00 0.50 0.50 2.00 2.00 2.00 vscatterqpd %ymm1, (%rdx,%ymm0,4) {%k1}
4565
4617
# CHECK-NEXT: - - - - - - - 1.00 - - - - vshuff32x4 $0, %ymm16, %ymm17, %ymm19
4566
4618
# CHECK-NEXT: - - - - 0.50 0.50 - 1.00 - - - - vshuff32x4 $0, (%rax), %ymm17, %ymm19
4567
4619
# CHECK-NEXT: - - - - 0.50 0.50 - 1.00 - - - - vshuff32x4 $0, (%rax){1to8}, %ymm17, %ymm19
0 commit comments