@@ -417,7 +417,8 @@ body: |
417
417
# GCN-LABEL: name: xdl_sgemm16x16_mfma_write_agpr_mfma_srca_read_overlap
418
418
# GCN: V_MFMA
419
419
# GCN-NEXT: S_NOP 7
420
- # GCN-NEXT: S_NOP 2
420
+ # GFX940-NEXT: S_NOP 2
421
+ # GFX950-NEXT: S_NOP 3
421
422
# GCN-NEXT: V_MFMA
422
423
name : xdl_sgemm16x16_mfma_write_agpr_mfma_srca_read_overlap
423
424
body : |
@@ -439,7 +440,8 @@ body: |
439
440
# GCN-LABEL: name: smfmac32x32_write_agpr_mfma_srca_read_overlap
440
441
# GCN: V_SMFMAC
441
442
# GCN-NEXT: S_NOP 7
442
- # GCN-NEXT: S_NOP 2
443
+ # GFX940-NEXT: S_NOP 2
444
+ # GFX950-NEXT: S_NOP 3
443
445
# GCN-NEXT: V_MFMA
444
446
name : smfmac32x32_write_agpr_mfma_srca_read_overlap
445
447
body : |
@@ -450,7 +452,8 @@ body: |
450
452
# GCN-LABEL: name: smfmac32x32_write_agpr_smfmac_srcc_read_overlap
451
453
# GCN: V_SMFMAC
452
454
# GCN-NEXT: S_NOP 7
453
- # GCN-NEXT: S_NOP 2
455
+ # GFX940-NEXT: S_NOP 2
456
+ # GFX950-NEXT: S_NOP 3
454
457
# GCN-NEXT: V_SMFMAC
455
458
name : smfmac32x32_write_agpr_smfmac_srcc_read_overlap
456
459
body : |
@@ -462,7 +465,8 @@ body: |
462
465
# GCN: V_MFMA
463
466
# GCN-NEXT: S_NOP 7
464
467
# GCN-NEXT: S_NOP 7
465
- # GCN-NEXT: S_NOP 2
468
+ # GFX940-NEXT: S_NOP 2
469
+ # GFX950-NEXT: S_NOP 3
466
470
# GCN-NEXT: V_MFMA
467
471
name : xdl_sgemm32x32_mfma_write_agpr_mfma_srca_read_overlap
468
472
body : |
@@ -1715,7 +1719,8 @@ body: |
1715
1719
...
1716
1720
# GCN-LABEL: name: xdl_sgemm16X16X16_mfma_write_agpr_mfma_srca_read_overlap
1717
1721
# GCN: V_MFMA
1718
- # GCN-NEXT: S_NOP 6
1722
+ # GFX940-NEXT: S_NOP 6
1723
+ # GFX950-NEXT: S_NOP 7
1719
1724
# GCN-NEXT: V_MFMA
1720
1725
name : xdl_sgemm16X16X16_mfma_write_agpr_mfma_srca_read_overlap
1721
1726
body : |
@@ -1725,7 +1730,8 @@ body: |
1725
1730
...
1726
1731
# GCN-LABEL: name: xdl_sgemm16X16X32_mfma_write_agpr_mfma_srcb_read_overlap
1727
1732
# GCN: V_MFMA
1728
- # GCN-NEXT: S_NOP 6
1733
+ # GFX940-NEXT: S_NOP 6
1734
+ # GFX950-NEXT: S_NOP 7
1729
1735
# GCN-NEXT: V_MFMA
1730
1736
name : xdl_sgemm16X16X32_mfma_write_agpr_mfma_srcb_read_overlap
1731
1737
body : |
@@ -1735,7 +1741,8 @@ body: |
1735
1741
...
1736
1742
# GCN-LABEL: name: xdl_sgemm16X16X16_mfma_write_vgpr_dmfma16x16_srca_read_overlap
1737
1743
# GCN: V_MFMA
1738
- # GCN-NEXT: S_NOP 6
1744
+ # GFX940-NEXT: S_NOP 6
1745
+ # GFX950-NEXT: S_NOP 7
1739
1746
# GCN-NEXT: V_MFMA
1740
1747
name : xdl_sgemm16X16X16_mfma_write_vgpr_dmfma16x16_srca_read_overlap
1741
1748
body : |
@@ -1826,7 +1833,8 @@ body: |
1826
1833
...
1827
1834
# GCN-LABEL: name: smfmac16x16x32_mfma_write_vgpr_smfmac_read_idx
1828
1835
# GCN: V_SMFMAC
1829
- # GCN-NEXT: S_NOP 6
1836
+ # GFX940-NEXT: S_NOP 6
1837
+ # GFX950-NEXT: S_NOP 7
1830
1838
# GCN-NEXT: V_SMFMAC
1831
1839
name : smfmac16x16x32_mfma_write_vgpr_smfmac_read_idx
1832
1840
body : |
@@ -2188,7 +2196,8 @@ body: |
2188
2196
# 4 pass source
2189
2197
# GCN-LABEL: name: xdl_mfma_4pass_write_vgpr_xdl_mfma_read_overlap_srca
2190
2198
# GCN: V_MFMA
2191
- # GCN-NEXT: S_NOP 6
2199
+ # GFX940-NEXT: S_NOP 6
2200
+ # GFX950-NEXT: S_NOP 7
2192
2201
# GCN-NEXT: V_MFMA
2193
2202
name : xdl_mfma_4pass_write_vgpr_xdl_mfma_read_overlap_srca
2194
2203
body : |
@@ -2202,7 +2211,8 @@ body: |
2202
2211
# 4 pass source
2203
2212
# GCN-LABEL: name: xdl_mfma_4pass_write_vgpr_xdl_mfma_read_overlap_srcb
2204
2213
# GCN: V_MFMA
2205
- # GCN-NEXT: S_NOP 6
2214
+ # GFX940-NEXT: S_NOP 6
2215
+ # GFX950-NEXT: S_NOP 7
2206
2216
# GCN-NEXT: V_MFMA
2207
2217
name : xdl_mfma_4pass_write_vgpr_xdl_mfma_read_overlap_srcb
2208
2218
body : |
@@ -2276,7 +2286,8 @@ body: |
2276
2286
# 4 pass source
2277
2287
# GCN-LABEL: name: xdl_mfma_4pass_write_vgpr_sgemm_mfma_read_overlap_srca
2278
2288
# GCN: V_MFMA
2279
- # GCN-NEXT: S_NOP 6
2289
+ # GFX940-NEXT: S_NOP 6
2290
+ # GFX950-NEXT: S_NOP 7
2280
2291
# GCN-NEXT: V_MFMA
2281
2292
name : xdl_mfma_4pass_write_vgpr_sgemm_mfma_read_overlap_srca
2282
2293
body : |
@@ -2290,7 +2301,8 @@ body: |
2290
2301
# 4 pass source
2291
2302
# GCN-LABEL: name: xdl_mfma_4pass_write_vgpr_sgemm_mfma_read_overlap_srcb
2292
2303
# GCN: V_MFMA
2293
- # GCN-NEXT: S_NOP 6
2304
+ # GFX940-NEXT: S_NOP 6
2305
+ # GFX950-NEXT: S_NOP 7
2294
2306
# GCN-NEXT: V_MFMA
2295
2307
name : xdl_mfma_4pass_write_vgpr_sgemm_mfma_read_overlap_srcb
2296
2308
body : |
@@ -2321,7 +2333,8 @@ body: |
2321
2333
# GCN-LABEL: name: xdl_mfma_8pass_write_vgpr_nonxdl_sgemm_mfma_read_overlap_srca
2322
2334
# GCN: V_MFMA
2323
2335
# GCN-NEXT: S_NOP 7
2324
- # GCN-NEXT: S_NOP 2
2336
+ # GFX940-NEXT: S_NOP 2
2337
+ # GFX950-NEXT: S_NOP 3
2325
2338
# GCN-NEXT: V_MFMA
2326
2339
name : xdl_mfma_8pass_write_vgpr_nonxdl_sgemm_mfma_read_overlap_srca
2327
2340
body : |
@@ -2336,7 +2349,8 @@ body: |
2336
2349
# GCN-LABEL: name: xdl_mfma_8pass_write_vgpr_nonxdl_sgemm_mfma_read_overlap_srcb
2337
2350
# GCN: V_MFMA
2338
2351
# GCN-NEXT: S_NOP 7
2339
- # GCN-NEXT: S_NOP 2
2352
+ # GFX940-NEXT: S_NOP 2
2353
+ # GFX950-NEXT: S_NOP 3
2340
2354
# GCN-NEXT: V_MFMA
2341
2355
name : xdl_mfma_8pass_write_vgpr_nonxdl_sgemm_mfma_read_overlap_srcb
2342
2356
body : |
@@ -2370,7 +2384,8 @@ body: |
2370
2384
# GCN: V_MFMA
2371
2385
# GCN-NEXT: S_NOP 7
2372
2386
# GCN-NEXT: S_NOP 7
2373
- # GCN-NEXT: S_NOP 2
2387
+ # GFX940-NEXT: S_NOP 2
2388
+ # GFX950-NEXT: S_NOP 3
2374
2389
# GCN-NEXT: V_MFMA
2375
2390
name : xdl_16pass_write_vgpr_nonxdl_sgemm_mfma_read_overlap_srca
2376
2391
body : |
@@ -2386,7 +2401,8 @@ body: |
2386
2401
# GCN: V_MFMA
2387
2402
# GCN-NEXT: S_NOP 7
2388
2403
# GCN-NEXT: S_NOP 7
2389
- # GCN-NEXT: S_NOP 2
2404
+ # GFX940-NEXT: S_NOP 2
2405
+ # GFX950-NEXT: S_NOP 3
2390
2406
# GCN-NEXT: V_MFMA
2391
2407
name : xdl_16pass_write_vgpr_nonxdl_sgemm_mfma_read_overlap_srcb
2392
2408
body : |
@@ -2456,7 +2472,8 @@ body: |
2456
2472
# GCN-LABEL: name: xdl_mfma_8pass_write_vgpr_xdl_mfma_read_overlap_srca
2457
2473
# GCN: V_MFMA
2458
2474
# GCN-NEXT: S_NOP 7
2459
- # GCN-NEXT: S_NOP 2
2475
+ # GFX940-NEXT: S_NOP 2
2476
+ # GFX950-NEXT: S_NOP 3
2460
2477
# GCN-NEXT: V_MFMA
2461
2478
name : xdl_mfma_8pass_write_vgpr_xdl_mfma_read_overlap_srca
2462
2479
body : |
@@ -2470,7 +2487,8 @@ body: |
2470
2487
# GCN-LABEL: name: xdl_mfma_8pass_write_vgpr_xdl_mfma_read_overlap_srcb
2471
2488
# GCN: V_MFMA
2472
2489
# GCN-NEXT: S_NOP 7
2473
- # GCN-NEXT: S_NOP 2
2490
+ # GFX940-NEXT: S_NOP 2
2491
+ # GFX950-NEXT: S_NOP 3
2474
2492
# GCN-NEXT: V_MFMA
2475
2493
name : xdl_mfma_8pass_write_vgpr_xdl_mfma_read_overlap_srcb
2476
2494
body : |
@@ -2502,7 +2520,8 @@ body: |
2502
2520
# GCN: V_MFMA
2503
2521
# GCN-NEXT: S_NOP 7
2504
2522
# GCN-NEXT: S_NOP 7
2505
- # GCN-NEXT: S_NOP 2
2523
+ # GFX940-NEXT: S_NOP 2
2524
+ # GFX950-NEXT: S_NOP 3
2506
2525
# GCN-NEXT: V_MFMA
2507
2526
name : xdl_16pass_write_vgpr_xdl_mfma_read_overlap_srca
2508
2527
body : |
@@ -2519,7 +2538,8 @@ body: |
2519
2538
# GCN: V_MFMA
2520
2539
# GCN-NEXT: S_NOP 7
2521
2540
# GCN-NEXT: S_NOP 7
2522
- # GCN-NEXT: S_NOP 2
2541
+ # GFX940-NEXT: S_NOP 2
2542
+ # GFX950-NEXT: S_NOP 3
2523
2543
# GCN-NEXT: V_MFMA
2524
2544
name : xdl_16pass_write_vgpr_xdl_mfma_read_overlap_srcb
2525
2545
body : |
0 commit comments