[rocroller] Fix ASan build when using Ninja instead of Make#1111
Merged
Conversation
Contributor
Generated Documentation |
Contributor
Code Coverage Report for gfx942Summary
Artifacts
Commit Hashes |
Contributor
Contributor
Performance Report for gfx950-perfResults@@ Significant (p-val <0.05) Performance Diffs @@
====================================================================================================
+ 0.44% | p=2.5635e-58
| 3. FloatsGEMM(M: 4096, N: 4096, K: 32768, alpha: 2, beta: 0, types: {'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 256, mac_n: 256, mac_k: 128, wave_m: 16, wave_n: 16, wave_k: 128, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: False, betaInFma: True, direct2LDS_A: True, direct2LDS_B: True, scheduler: Priority, prefetch: True, prefetchInFlight: 4, prefetchLDSFactor: 1, prefetchMixMemOps: True, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: True, prefetchScale: True, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=32768, alpha=2, beta=0, types={'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=256, mac_n=256, mac_k=128, wave_m=16, wave_n=16, wave_k=128, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=False, betaInFma=True, direct2LDS_A=True, direct2LDS_B=True, scheduler='Priority', prefetch=True, prefetchInFlight=4, prefetchLDSFactor=1, prefetchMixMemOps=True, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=True, prefetchScale=True, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.16% | p=3.5139e-15
| 3. FloatsGEMM(M: 4096, N: 4096, K: 32768, alpha: 2, beta: 0, types: {'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 256, mac_n: 256, mac_k: 128, wave_m: 32, wave_n: 32, wave_k: 64, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: False, betaInFma: True, direct2LDS_A: True, direct2LDS_B: True, scheduler: Priority, prefetch: True, prefetchInFlight: 4, prefetchLDSFactor: 1, prefetchMixMemOps: True, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: True, prefetchScale: True, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=32768, alpha=2, beta=0, types={'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=256, mac_n=256, mac_k=128, wave_m=32, wave_n=32, wave_k=64, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=False, betaInFma=True, direct2LDS_A=True, direct2LDS_B=True, scheduler='Priority', prefetch=True, prefetchInFlight=4, prefetchLDSFactor=1, prefetchMixMemOps=True, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=True, prefetchScale=True, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.06% | p=1.7418e-03
| 3. FloatsGEMM(M: 4096, N: 4096, K: 32768, alpha: 2, beta: 0, types: {'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: 0, workgroupMappingValue: -1)| GEMM(mac_m: 256, mac_n: 256, mac_k: 128, wave_m: 16, wave_n: 16, wave_k: 128, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: False, betaInFma: True, direct2LDS_A: True, direct2LDS_B: True, scheduler: Priority, prefetch: True, prefetchInFlight: 4, prefetchLDSFactor: 1, prefetchMixMemOps: True, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: True, prefetchScale: True, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=32768, alpha=2, beta=0, types={'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=0, workgroupMappingValue=-1)GEMMSolution(mac_m=256, mac_n=256, mac_k=128, wave_m=16, wave_n=16, wave_k=128, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=False, betaInFma=True, direct2LDS_A=True, direct2LDS_B=True, scheduler='Priority', prefetch=True, prefetchInFlight=4, prefetchLDSFactor=1, prefetchMixMemOps=True, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=True, prefetchScale=True, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 0.06% | p=5.5589e-03
| 3. FloatsGEMM(M: 4096, N: 4096, K: 32768, alpha: 2, beta: 0, types: {'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: 0, workgroupMappingValue: -1)| GEMM(mac_m: 256, mac_n: 256, mac_k: 128, wave_m: 32, wave_n: 32, wave_k: 64, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: False, betaInFma: True, direct2LDS_A: True, direct2LDS_B: True, scheduler: Priority, prefetch: True, prefetchInFlight: 4, prefetchLDSFactor: 1, prefetchMixMemOps: True, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: True, prefetchScale: True, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=32768, alpha=2, beta=0, types={'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=0, workgroupMappingValue=-1)GEMMSolution(mac_m=256, mac_n=256, mac_k=128, wave_m=32, wave_n=32, wave_k=64, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=False, betaInFma=True, direct2LDS_A=True, direct2LDS_B=True, scheduler='Priority', prefetch=True, prefetchInFlight=4, prefetchLDSFactor=1, prefetchMixMemOps=True, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=True, prefetchScale=True, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.24% | p=1.6058e-34
| 3. FloatsGEMM(M: 4096, N: 4096, K: 32768, alpha: 2, beta: 0, types: {'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': True, 'scaleShuffleTileA': [64, 4, 2], 'scaleShuffleTileB': [64, 4, 2]}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: 0, workgroupMappingValue: -1)| GEMM(mac_m: 256, mac_n: 256, mac_k: 128, wave_m: 32, wave_n: 32, wave_k: 64, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: False, betaInFma: True, direct2LDS_A: True, direct2LDS_B: True, scheduler: Priority, prefetch: True, prefetchInFlight: 4, prefetchLDSFactor: 1, prefetchMixMemOps: True, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: True, prefetchScale: True, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=32768, alpha=2, beta=0, types={'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': True, 'scaleShuffleTileA': [64, 4, 2], 'scaleShuffleTileB': [64, 4, 2]}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=0, workgroupMappingValue=-1)GEMMSolution(mac_m=256, mac_n=256, mac_k=128, wave_m=32, wave_n=32, wave_k=64, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=False, betaInFma=True, direct2LDS_A=True, direct2LDS_B=True, scheduler='Priority', prefetch=True, prefetchInFlight=4, prefetchLDSFactor=1, prefetchMixMemOps=True, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=True, prefetchScale=True, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 0.36% | p=1.2777e-45
| 3. FloatsGEMM(M: 4096, N: 4096, K: 32768, alpha: 2, beta: 0.5, types: {'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 256, mac_n: 256, mac_k: 128, wave_m: 32, wave_n: 32, wave_k: 64, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 2, prefetchMixMemOps: False, loadLDSScale_A: True, loadLDSScale_B: True, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=32768, alpha=2, beta=0.5, types={'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=256, mac_n=256, mac_k=128, wave_m=32, wave_n=32, wave_k=64, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=2, prefetchMixMemOps=False, loadLDSScale_A=True, loadLDSScale_B=True, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.56% | p=1.5045e-134
| 3. FloatsGEMM(M: 4096, N: 4096, K: 32768, alpha: 2, beta: 0.5, types: {'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 256, mac_n: 256, mac_k: 128, wave_m: 32, wave_n: 32, wave_k: 64, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: True, direct2LDS_B: True, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 1, prefetchMixMemOps: False, loadLDSScale_A: True, loadLDSScale_B: True, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=32768, alpha=2, beta=0.5, types={'type_A': 'fp4', 'type_B': 'fp4', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'Separate', 'scaleType_A': 'E8M0', 'scale_B': 'Separate', 'scaleType_B': 'E8M0', 'scaleBlockSize': 32, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=256, mac_n=256, mac_k=128, wave_m=32, wave_n=32, wave_k=64, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=True, direct2LDS_B=True, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=1, prefetchMixMemOps=False, loadLDSScale_A=True, loadLDSScale_B=True, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 0.25% | p=2.5347e-02
| 3. FloatsGEMM(M: 4096, N: 4096, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'bf6', 'type_B': 'bf6', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 128, mac_k: 64, wave_m: 32, wave_n: 32, wave_k: 64, wave_b: -1, workgroup_size_x: 256, workgroup_size_y: 1, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=8192, alpha=2, beta=0.5, types={'type_A': 'bf6', 'type_B': 'bf6', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=128, mac_k=64, wave_m=32, wave_n=32, wave_k=64, wave_b=-1, workgroup_size_x=256, workgroup_size_y=1, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.81% | p=2.5347e-02
| 3. FloatsGEMM(M: 4096, N: 4096, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'bf8', 'type_B': 'bf8', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'N', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 128, mac_k: 64, wave_m: 32, wave_n: 32, wave_k: 64, wave_b: 1, workgroup_size_x: 256, workgroup_size_y: 1, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=8192, alpha=2, beta=0.5, types={'type_A': 'bf8', 'type_B': 'bf8', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'N', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=128, mac_k=64, wave_m=32, wave_n=32, wave_k=64, wave_b=1, workgroup_size_x=256, workgroup_size_y=1, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.88% | p=2.5347e-02
| 3. FloatsGEMM(M: 4096, N: 4096, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'bf8', 'type_B': 'bf8', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 128, mac_k: 64, wave_m: 32, wave_n: 32, wave_k: 64, wave_b: 1, workgroup_size_x: 256, workgroup_size_y: 1, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=8192, alpha=2, beta=0.5, types={'type_A': 'bf8', 'type_B': 'bf8', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=128, mac_k=64, wave_m=32, wave_n=32, wave_k=64, wave_b=1, workgroup_size_x=256, workgroup_size_y=1, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 1.13% | p=2.5347e-02
| 3. FloatsGEMM(M: 4096, N: 4096, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'bf8', 'type_B': 'bf8', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 128, wave_m: 16, wave_n: 16, wave_k: 128, wave_b: 1, workgroup_size_x: 256, workgroup_size_y: 1, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=8192, alpha=2, beta=0.5, types={'type_A': 'bf8', 'type_B': 'bf8', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=128, wave_m=16, wave_n=16, wave_k=128, wave_b=1, workgroup_size_x=256, workgroup_size_y=1, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 1.29% | p=1.7451e-03
| 3. FloatsGEMM(M: 4096, N: 4096, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'fp8', 'type_B': 'fp8', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 128, mac_k: 64, wave_m: 32, wave_n: 32, wave_k: 64, wave_b: 1, workgroup_size_x: 256, workgroup_size_y: 1, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=4096, N=4096, K=8192, alpha=2, beta=0.5, types={'type_A': 'fp8', 'type_B': 'fp8', 'type_C': 'float', 'type_D': 'float', 'type_acc': 'float', 'trans_A': 'T', 'trans_B': 'N', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=128, mac_k=64, wave_m=32, wave_n=32, wave_k=64, wave_b=1, workgroup_size_x=256, workgroup_size_y=1, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.43% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 128, mac_k: 32, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: True, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=128, mac_k=32, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=True, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 1.08% | p=5.6994e-05
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 64, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: True, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=64, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=True, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 0.43% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 64, mac_k: 64, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: True, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=64, mac_k=64, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=True, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.48% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 128, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=128, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.53% | p=1.7451e-03
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 128, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: True, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=128, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=True, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.41% | p=1.7451e-03
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 128, mac_k: 32, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=128, mac_k=32, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.71% | p=1.7451e-03
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 256, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=256, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 0.21% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 256, mac_k: 32, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: True, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=256, mac_k=32, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=True, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.35% | p=1.7451e-03
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: True, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=True, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 0.20% | p=1.7451e-03
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 32, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=32, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.39% | p=5.6994e-05
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 32, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: True, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=32, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=True, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.13% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 64, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 256, streamKTwoTile: True, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=64, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=256, streamKTwoTile=True, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 1.61% | p=2.5347e-02
| 3. FloatsGEMM(M: 8448, N: 8448, K: 128, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 256, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Cooperative, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=8448, N=8448, K=128, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=256, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Cooperative', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx950', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 3.03% | p=5.6994e-05
| CodeGen(instCount: 40000, instructions: comments)| CodeGen() | CodeGen(instCount: 40000, instructions: comments)
+ 1.77% | p=1.7451e-03
| CodeGen(instCount: 40000, instructions: simple_mi)| CodeGen() | CodeGen(instCount: 40000, instructions: simple_mi)
Full table of results
|
Contributor
Performance Report for gfx12Results@@ Significant (p-val <0.05) Performance Diffs @@
====================================================================================================
+ 0.95% | p=5.6994e-05
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'bf16', 'type_B': 'bf16', 'type_C': 'bf16', 'type_D': 'bf16', 'type_acc': 'bf16', 'trans_A': 'T', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 64, wave_m: 16, wave_n: 16, wave_k: 16, wave_b: 1, workgroup_size_x: 64, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Cooperative, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx1201', 'Xnack': False, 'Sramecc': False}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8192, alpha=2, beta=0.5, types={'type_A': 'bf16', 'type_B': 'bf16', 'type_C': 'bf16', 'type_D': 'bf16', 'type_acc': 'bf16', 'trans_A': 'T', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=64, wave_m=16, wave_n=16, wave_k=16, wave_b=1, workgroup_size_x=64, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Cooperative', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx1201', 'Xnack': False, 'Sramecc': False}, matchMemoryAccess=True, version='')
+ 0.35% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'bf16', 'type_B': 'bf16', 'type_C': 'bf16', 'type_D': 'bf16', 'type_acc': 'bf16', 'trans_A': 'T', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 64, wave_m: 16, wave_n: 16, wave_k: 16, wave_b: 1, workgroup_size_x: 64, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Sequential, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx1201', 'Xnack': False, 'Sramecc': False}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8192, alpha=2, beta=0.5, types={'type_A': 'bf16', 'type_B': 'bf16', 'type_C': 'bf16', 'type_D': 'bf16', 'type_acc': 'bf16', 'trans_A': 'T', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=64, wave_m=16, wave_n=16, wave_k=16, wave_b=1, workgroup_size_x=64, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Sequential', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx1201', 'Xnack': False, 'Sramecc': False}, matchMemoryAccess=True, version='')
+ 0.45% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'half', 'trans_A': 'T', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 64, wave_m: 16, wave_n: 16, wave_k: 16, wave_b: 1, workgroup_size_x: 64, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Cooperative, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx1201', 'Xnack': False, 'Sramecc': False}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8192, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'half', 'trans_A': 'T', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=64, wave_m=16, wave_n=16, wave_k=16, wave_b=1, workgroup_size_x=64, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Cooperative', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx1201', 'Xnack': False, 'Sramecc': False}, matchMemoryAccess=True, version='')
- 0.94% | p=2.5347e-02
| CodeGen(instCount: 40000, instructions: comments)| CodeGen() | CodeGen(instCount: 40000, instructions: comments)
+ 2.89% | p=5.6994e-05
| CodeGen(instCount: 40000, instructions: simple_mi)| CodeGen() | CodeGen(instCount: 40000, instructions: simple_mi)
Full table of results
|
Contributor
Performance Report for gfx942Results@@ Significant (p-val <0.05) Performance Diffs @@
====================================================================================================
+ 3.84% | p=2.5347e-02
| 3. FloatsGEMM(M: 1024, N: 50304, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 128, mac_k: 32, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Cooperative, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=1024, N=50304, K=8192, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=128, mac_k=32, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Cooperative', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 7.23% | p=1.7451e-03
| 3. FloatsGEMM(M: 1024, N: 50304, K: 8192, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 128, mac_k: 32, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Sequential, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=1024, N=50304, K=8192, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=128, mac_k=32, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Sequential', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 4.89% | p=1.7451e-03
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8192, alpha: 2, beta: 0, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'N', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 256, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 64, workgroup_size_y: 4, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8192, alpha=2, beta=0, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'N', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=256, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=64, workgroup_size_y=4, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 4.18% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 256, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 64, workgroup_size_y: 4, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=256, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=64, workgroup_size_y=4, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 1.57% | p=1.7451e-03
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 64, mac_k: 64, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 304, streamKTwoTile: True, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=64, mac_k=64, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=304, streamKTwoTile=True, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 0.27% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 128, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 304, streamKTwoTile: False, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=128, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=304, streamKTwoTile=False, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 2.27% | p=5.6994e-05
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 128, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 304, streamKTwoTile: True, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=128, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=304, streamKTwoTile=True, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 0.75% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 128, mac_k: 32, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 304, streamKTwoTile: False, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=128, mac_k=32, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=304, streamKTwoTile=False, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 1.00% | p=1.7451e-03
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 128, mac_k: 32, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 304, streamKTwoTile: True, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=128, mac_k=32, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=304, streamKTwoTile=True, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 1.11% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 128, mac_k: 64, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 304, streamKTwoTile: True, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=128, mac_k=64, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=304, streamKTwoTile=True, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 1.18% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 256, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 304, streamKTwoTile: False, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=256, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=304, streamKTwoTile=False, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 0.49% | p=1.7451e-03
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 304, streamKTwoTile: True, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=304, streamKTwoTile=True, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 7.32% | p=2.5347e-02
| 3. FloatsGEMM(M: 7680, N: 8448, K: 8448, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 64, mac_n: 64, mac_k: 64, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: False, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: True, numWGs: 304, streamKTwoTile: False, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=7680, N=8448, K=8448, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=64, mac_n=64, mac_k=64, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=False, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=True, numWGs=304, streamKTwoTile=False, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
+ 3.96% | p=5.6994e-05
| 3. FloatsGEMM(M: 8448, N: 8448, K: 128, alpha: 2, beta: 0.5, types: {'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A: 1, scaleValue_B: 1, workgroupMappingDim: -1, workgroupMappingValue: -1)| GEMM(mac_m: 128, mac_n: 256, mac_k: 16, wave_m: 32, wave_n: 32, wave_k: 8, wave_b: 1, workgroup_size_x: 128, workgroup_size_y: 2, workgroupRemapXCC: False, workgroupRemapXCCValue: -1, unroll_x: 0, unroll_y: 0, loadLDS_A: True, loadLDS_B: True, storeLDS_D: True, betaInFma: True, direct2LDS_A: False, direct2LDS_B: False, scheduler: Priority, prefetch: True, prefetchInFlight: 2, prefetchLDSFactor: 0, prefetchMixMemOps: False, loadLDSScale_A: False, loadLDSScale_B: False, swizzleScale: False, prefetchScale: False, streamK: False, numWGs: 0, streamKTwoTile: False, architecture: {'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess: True, version: ) | GEMMProblem(M=8448, N=8448, K=128, alpha=2, beta=0.5, types={'type_A': 'half', 'type_B': 'half', 'type_C': 'half', 'type_D': 'half', 'type_acc': 'float', 'trans_A': 'N', 'trans_B': 'T', 'scale_A': 'None', 'scaleType_A': 'None', 'scale_B': 'None', 'scaleType_B': 'None', 'scaleBlockSize': -1, 'scaleSkipPermlane': False, 'scaleShuffleTileA': [], 'scaleShuffleTileB': []}, scaleValue_A=1, scaleValue_B=1, workgroupMappingDim=-1, workgroupMappingValue=-1)GEMMSolution(mac_m=128, mac_n=256, mac_k=16, wave_m=32, wave_n=32, wave_k=8, wave_b=1, workgroup_size_x=128, workgroup_size_y=2, workgroupRemapXCC=False, workgroupRemapXCCValue=-1, unroll_x=0, unroll_y=0, loadLDS_A=True, loadLDS_B=True, storeLDS_D=True, betaInFma=True, direct2LDS_A=False, direct2LDS_B=False, scheduler='Priority', prefetch=True, prefetchInFlight=2, prefetchLDSFactor=0, prefetchMixMemOps=False, loadLDSScale_A=False, loadLDSScale_B=False, swizzleScale=False, prefetchScale=False, streamK=False, numWGs=0, streamKTwoTile=False, architecture={'ArchString': 'gfx942', 'Xnack': False, 'Sramecc': True}, matchMemoryAccess=True, version='')
- 14.46% | p=5.6994e-05
| CodeGen(instCount: 40000, instructions: comments)| CodeGen() | CodeGen(instCount: 40000, instructions: comments)
+ 3.20% | p=2.5347e-02
| CodeGen(instCount: 40000, instructions: complex_mi_with_coop)| CodeGen() | CodeGen(instCount: 40000, instructions: complex_mi_with_coop)
Full table of results
|
ThanHenderson
approved these changes
Aug 8, 2025
Contributor
ThanHenderson
left a comment
There was a problem hiding this comment.
Weird problem with ninja/make. LGTM. Thanks for patching the typo.
awhittle3
approved these changes
Aug 8, 2025
Contributor
Author
|
@ROCm/rocm-math-lib-build-infra please review once you have a chance. |
awhittle3
approved these changes
Aug 20, 2025
assistant-librarian Bot
pushed a commit
to ROCm/rocRoller
that referenced
this pull request
Aug 20, 2025
[rocroller] Fix ASan build when using Ninja instead of Make (#1111) Short story: - Fixes a typo in `test/CMakeLists.txt` which prevent ASan build from being configured (`cmake --preset asan`). - Allows ASan builds using Ninja instead of Make (Default) by avoiding `typeinfo` symbol for `rocRoller::Error` from being undefined due to conflicting definitions in objects compiled w/o RTTI. Long story: After fixing a typo in `test/CMakeLists.txt` ASan build would fail when using Ninja instead of Make (RR default) due to linking errors of undefined `typeinfo` symbol for `rocRoller::Error`. Such symbol was undefined because objects compiled with `-fno-rtti` and that use `Error.hpp` had a WEAK definition for said symbol and these objects -- e.g. `InProcessAssembler.cpp.o` and `FastDivision.cpp.o` -- were ordered first in the link command invocation. (It is not clear to me why CMake order these objects differently when using Make & Ninja, but my suspicion is that they do because when Make is used the paths are wrapped with quotes and thus they are ordered after paths that start with `[a-zA-Z]`.) I verified that both Ninja & Make execute exactly the same commands with same flags, except for the ordering mentioned above. I have also verified that manually changing the order produced by CMake when using Ninja resolves the linking issues. Since both `InProcessAssembler.cpp` & `FastDivision.cpp` do not use LLVM headers anymore, then they don't need to be compiled with RTTI disabled (ROCm/rocm-libraries@335a3a7). Once Comgr is used instead of LLVM in `ReadELF.cpp`, then all logic to build objects with RTTI disabled can be safely removed and similar linking issues can be avoided.
BrianHarrisonAMD
pushed a commit
that referenced
this pull request
Mar 25, 2026
…/sphinx (#5767) Bumps [pyjwt](https://github.com/jpadilla/pyjwt) from 2.10.1 to 2.12.0. <details> <summary>Release notes</summary> <p><em>Sourced from <a href="https://github.com/jpadilla/pyjwt/releases">pyjwt's releases</a>.</em></p> <blockquote> <h2>2.12.0</h2> <h2>Security</h2> <ul> <li>Validate the crit (Critical) Header Parameter defined in RFC 7515 §4.1.11. by <a href="https://github.com/dmbs335"><code>@dmbs335</code></a> in <a href="https://github.com/jpadilla/pyjwt/security/advisories/GHSA-752w-5fwx-jx9f">GHSA-752w-5fwx-jx9f</a></li> </ul> <h2>What's Changed</h2> <ul> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1132">jpadilla/pyjwt#1132</a></li> <li>chore(docs): fix docs build by <a href="https://github.com/tamird"><code>@tamird</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1137">jpadilla/pyjwt#1137</a></li> <li>Annotate PyJWKSet.keys for pyright by <a href="https://github.com/tamird"><code>@tamird</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1134">jpadilla/pyjwt#1134</a></li> <li>fix: close HTTPError to prevent ResourceWarning on Python 3.14 by <a href="https://github.com/veeceey"><code>@veeceey</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1133">jpadilla/pyjwt#1133</a></li> <li>chore: remove superfluous constants by <a href="https://github.com/tamird"><code>@tamird</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1136">jpadilla/pyjwt#1136</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1135">jpadilla/pyjwt#1135</a></li> <li>chore(tests): enable mypy by <a href="https://github.com/tamird"><code>@tamird</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1138">jpadilla/pyjwt#1138</a></li> <li>Bump actions/download-artifact from 7 to 8 by <a href="https://github.com/dependabot"><code>@dependabot</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1142">jpadilla/pyjwt#1142</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1141">jpadilla/pyjwt#1141</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1145">jpadilla/pyjwt#1145</a></li> <li>fix: do not store reference to algorithms dict on PyJWK by <a href="https://github.com/akx"><code>@akx</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1143">jpadilla/pyjwt#1143</a></li> <li>Use PyJWK algorithm when encoding without explicit algorithm by <a href="https://github.com/jpadilla"><code>@jpadilla</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1148">jpadilla/pyjwt#1148</a></li> </ul> <h2>New Contributors</h2> <ul> <li><a href="https://github.com/tamird"><code>@tamird</code></a> made their first contribution in <a href="https://github.com/jpadilla/pyjwt/pull/1137">jpadilla/pyjwt#1137</a></li> <li><a href="https://github.com/veeceey"><code>@veeceey</code></a> made their first contribution in <a href="https://github.com/jpadilla/pyjwt/pull/1133">jpadilla/pyjwt#1133</a></li> </ul> <p><strong>Full Changelog</strong>: <a href="https://github.com/jpadilla/pyjwt/compare/2.11.0...2.12.0">https://github.com/jpadilla/pyjwt/compare/2.11.0...2.12.0</a></p> <h2>2.11.0</h2> <h2>What's Changed</h2> <ul> <li>Fixed type error in comment by <a href="https://github.com/shuhaib-aot"><code>@shuhaib-aot</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1026">jpadilla/pyjwt#1026</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1018">jpadilla/pyjwt#1018</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1033">jpadilla/pyjwt#1033</a></li> <li>Make note of use of leeway with nbf by <a href="https://github.com/djw8605"><code>@djw8605</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1034">jpadilla/pyjwt#1034</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1035">jpadilla/pyjwt#1035</a></li> <li>Fixes <a href="https://github.com/jpadilla/pyjwt/issues/964">#964</a>: Validate key against allowed types for Algorithm family by <a href="https://github.com/pachewise"><code>@pachewise</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/985">jpadilla/pyjwt#985</a></li> <li>Feat <a href="https://github.com/jpadilla/pyjwt/issues/1024">#1024</a>: Add iterator for PyJWKSet by <a href="https://github.com/pachewise"><code>@pachewise</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1041">jpadilla/pyjwt#1041</a></li> <li>Fixes <a href="https://github.com/jpadilla/pyjwt/issues/1039">#1039</a>: Add iss, issuer type checks by <a href="https://github.com/pachewise"><code>@pachewise</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1040">jpadilla/pyjwt#1040</a></li> <li>Fixes <a href="https://github.com/jpadilla/pyjwt/issues/660">#660</a>: Improve typing/logic for <code>options</code> in decode, decode_complete; Improve docs by <a href="https://github.com/pachewise"><code>@pachewise</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1045">jpadilla/pyjwt#1045</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1042">jpadilla/pyjwt#1042</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1052">jpadilla/pyjwt#1052</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1053">jpadilla/pyjwt#1053</a></li> <li>Fix <a href="https://github.com/jpadilla/pyjwt/issues/1022">#1022</a>: Map <code>algorithm=None</code> to "none" by <a href="https://github.com/qqii"><code>@qqii</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1056">jpadilla/pyjwt#1056</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1055">jpadilla/pyjwt#1055</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1058">jpadilla/pyjwt#1058</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1060">jpadilla/pyjwt#1060</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1061">jpadilla/pyjwt#1061</a></li> <li>Fixes <a href="https://github.com/jpadilla/pyjwt/issues/1047">#1047</a>: Correct <code>PyJWKClient.get_signing_key_from_jwt</code> annotation by <a href="https://github.com/khvn26"><code>@khvn26</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1048">jpadilla/pyjwt#1048</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1062">jpadilla/pyjwt#1062</a></li> <li>Fixed doc string typo in _validate_jti() function <a href="https://github.com/jpadilla/pyjwt/issues/1063">#1063</a> by <a href="https://github.com/kuldeepkhatke"><code>@kuldeepkhatke</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1064">jpadilla/pyjwt#1064</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1065">jpadilla/pyjwt#1065</a></li> <li>Update SECURITY.md by <a href="https://github.com/auvipy"><code>@auvipy</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1057">jpadilla/pyjwt#1057</a></li> <li>Typing fix: use <code>float</code> instead of <code>int</code> for <code>lifespan</code> and <code>timeout</code> by <a href="https://github.com/nikitagashkov"><code>@nikitagashkov</code></a> in <a href="https://github.com/jpadilla/pyjwt/pull/1068">jpadilla/pyjwt#1068</a></li> <li>[pre-commit.ci] pre-commit autoupdate by <a href="https://github.com/pre-commit-ci"><code>@pre-commit-ci</code></a>[bot] in <a href="https://github.com/jpadilla/pyjwt/pull/1067">jpadilla/pyjwt#1067</a></li> </ul> <!-- raw HTML omitted --> </blockquote> <p>... (truncated)</p> </details> <details> <summary>Changelog</summary> <p><em>Sourced from <a href="https://github.com/jpadilla/pyjwt/blob/master/CHANGELOG.rst">pyjwt's changelog</a>.</em></p> <blockquote> <h2><code>v2.12.0 <https://github.com/jpadilla/pyjwt/compare/2.11.0...2.12.0></code>__</h2> <p>Fixed</p> <pre><code> - Annotate PyJWKSet.keys for pyright by @tamird in `[#1134](jpadilla/pyjwt#1134) <https://github.com/jpadilla/pyjwt/pull/1134>`__ - Close ``HTTPError`` response to prevent ``ResourceWarning`` on Python 3.14 by @veeceey in `[#1133](jpadilla/pyjwt#1133) <https://github.com/jpadilla/pyjwt/pull/1133>`__ - Do not keep ``algorithms`` dict in PyJWK instances by @akx in `[#1143](jpadilla/pyjwt#1143) <https://github.com/jpadilla/pyjwt/pull/1143>`__ - Validate the crit (Critical) Header Parameter defined in RFC 7515 §4.1.11. by @dmbs335 in `GHSA-752w-5fwx-jx9f <https://github.com/jpadilla/pyjwt/security/advisories/GHSA-752w-5fwx-jx9f>`__ - Use PyJWK algorithm when encoding without explicit algorithm in `[#1148](jpadilla/pyjwt#1148) <https://github.com/jpadilla/pyjwt/pull/1148>`__ <p>Added </code></pre></p> <ul> <li>Docs: Add <code>PyJWKClient</code> API reference and document the two-tier caching system (JWK Set cache and signing key LRU cache).</li> </ul> <h2><code>v2.11.0 <https://github.com/jpadilla/pyjwt/compare/2.10.1...2.11.0></code>__</h2> <p>Fixed</p> <pre><code> - Enforce ECDSA curve validation per RFC 7518 Section 3.4. - Fix build system warnings by @kurtmckee in `[#1105](jpadilla/pyjwt#1105) <https://github.com/jpadilla/pyjwt/pull/1105>`__ - Validate key against allowed types for Algorithm family in `[#964](jpadilla/pyjwt#964) <https://github.com/jpadilla/pyjwt/pull/964>`__ - Add iterator for JWKSet in `[#1041](jpadilla/pyjwt#1041) <https://github.com/jpadilla/pyjwt/pull/1041>`__ - Validate `iss` claim is a string during encoding and decoding by @pachewise in `[#1040](jpadilla/pyjwt#1040) <https://github.com/jpadilla/pyjwt/pull/1040>`__ - Improve typing/logic for `options` in decode, decode_complete by @pachewise in `[#1045](jpadilla/pyjwt#1045) <https://github.com/jpadilla/pyjwt/pull/1045>`__ - Declare float supported type for lifespan and timeout by @nikitagashkov in `[#1068](jpadilla/pyjwt#1068) <https://github.com/jpadilla/pyjwt/pull/1068>`__ - Fix ``SyntaxWarning``\s/``DeprecationWarning``\s caused by invalid escape sequences by @kurtmckee in `[#1103](jpadilla/pyjwt#1103) <https://github.com/jpadilla/pyjwt/pull/1103>`__ - Development: Build a shared wheel once to speed up test suite setup times by @kurtmckee in `[#1114](jpadilla/pyjwt#1114) <https://github.com/jpadilla/pyjwt/pull/1114>`__ - Development: Test type annotations across all supported Python versions, increase the strictness of the type checking, and remove the mypy pre-commit hook by @kurtmckee in `[#1112](jpadilla/pyjwt#1112) <https://github.com/jpadilla/pyjwt/pull/1112>`__ <p>Added </code></pre></p> <ul> <li>Support Python 3.14, and test against PyPy 3.10 and 3.11 by <a href="https://github.com/kurtmckee"><code>@kurtmckee</code></a> in <code>[#1104](jpadilla/pyjwt#1104) <https://github.com/jpadilla/pyjwt/pull/1104></code>__</li> <li>Development: Migrate to <code>build</code> to test package building in CI by <a href="https://github.com/kurtmckee"><code>@kurtmckee</code></a> in <code>[#1108](jpadilla/pyjwt#1108) <https://github.com/jpadilla/pyjwt/pull/1108></code>__</li> <li>Development: Improve coverage config and eliminate unused test suite code by <a href="https://github.com/kurtmckee"><code>@kurtmckee</code></a> in <code>[#1115](jpadilla/pyjwt#1115) <https://github.com/jpadilla/pyjwt/pull/1115></code>__</li> <li>Docs: Standardize CHANGELOG links to PRs by <a href="https://github.com/kurtmckee"><code>@kurtmckee</code></a> in <code>[#1110](jpadilla/pyjwt#1110) <https://github.com/jpadilla/pyjwt/pull/1110></code>__</li> <li>Docs: Fix Read the Docs builds by <a href="https://github.com/kurtmckee"><code>@kurtmckee</code></a> in <code>[#1111](jpadilla/pyjwt#1111) <https://github.com/jpadilla/pyjwt/pull/1111></code>__</li> <li>Docs: Add example of using leeway with nbf by <a href="https://github.com/djw8605"><code>@djw8605</code></a> in <code>[#1034](jpadilla/pyjwt#1034) <https://github.com/jpadilla/pyjwt/pull/1034></code>__</li> <li>Docs: Refactored docs with <code>autodoc</code>; added <code>PyJWS</code> and <code>jwt.algorithms</code> docs by <a href="https://github.com/pachewise"><code>@pachewise</code></a> in <code>[#1045](jpadilla/pyjwt#1045) <https://github.com/jpadilla/pyjwt/pull/1045></code>__</li> <li>Docs: Documentation improvements for "sub" and "jti" claims by <a href="https://github.com/cleder"><code>@cleder</code></a> in <code>[#1088](jpadilla/pyjwt#1088) <https://github.com/jpadilla/pyjwt/pull/1088></code>__</li> <li>Development: Add pyupgrade as a pre-commit hook by <a href="https://github.com/kurtmckee"><code>@kurtmckee</code></a> in <code>[#1109](jpadilla/pyjwt#1109) <https://github.com/jpadilla/pyjwt/pull/1109></code>__</li> <li>Add minimum key length validation for HMAC and RSA keys (CWE-326). Warns by default via <code>InsecureKeyLengthWarning</code> when keys are below</li> </ul> <!-- raw HTML omitted --> </blockquote> <p>... (truncated)</p> </details> <details> <summary>Commits</summary> <ul> <li><a href="https://github.com/jpadilla/pyjwt/commit/bd9700cca7f9258fadcc429c1034e508025931f2"><code>bd9700c</code></a> Use PyJWK algorithm when encoding without explicit algorithm (<a href="https://github.com/jpadilla/pyjwt/issues/1148">#1148</a>)</li> <li><a href="https://github.com/jpadilla/pyjwt/commit/051ea341b5573fe3edcd53042f347929b92c2b92"><code>051ea34</code></a> Merge commit from fork</li> <li><a href="https://github.com/jpadilla/pyjwt/commit/1451d70eca2059bc472703692f0bb0777bc0fe93"><code>1451d70</code></a> fix: do not store reference to algorithms dict on PyJWK (<a href="https://github.com/jpadilla/pyjwt/issues/1143">#1143</a>)</li> <li><a href="https://github.com/jpadilla/pyjwt/commit/f3ba74c106df9ce10e272dfaad96acb4ab3ef5a5"><code>f3ba74c</code></a> [pre-commit.ci] pre-commit autoupdate (<a href="https://github.com/jpadilla/pyjwt/issues/1145">#1145</a>)</li> <li><a href="https://github.com/jpadilla/pyjwt/commit/0318ffa7b156b01600376e38952bf961382e0724"><code>0318ffa</code></a> [pre-commit.ci] pre-commit autoupdate (<a href="https://github.com/jpadilla/pyjwt/issues/1141">#1141</a>)</li> <li><a href="https://github.com/jpadilla/pyjwt/commit/a52753db3c1075ac01337fa8b7cc92b13a19ac09"><code>a52753d</code></a> Bump actions/download-artifact from 7 to 8 (<a href="https://github.com/jpadilla/pyjwt/issues/1142">#1142</a>)</li> <li><a href="https://github.com/jpadilla/pyjwt/commit/b85050f1d444c6828bb4618ee764443b0a3f5d18"><code>b85050f</code></a> chore(tests): enable mypy (<a href="https://github.com/jpadilla/pyjwt/issues/1138">#1138</a>)</li> <li><a href="https://github.com/jpadilla/pyjwt/commit/1272b264779717cc481c8341f321a7fc8b3aaba6"><code>1272b26</code></a> [pre-commit.ci] pre-commit autoupdate (<a href="https://github.com/jpadilla/pyjwt/issues/1135">#1135</a>)</li> <li><a href="https://github.com/jpadilla/pyjwt/commit/99a87287c26cb97c94399084ee4186ee52207a7f"><code>99a8728</code></a> chore: remove superfluous constants (<a href="https://github.com/jpadilla/pyjwt/issues/1136">#1136</a>)</li> <li><a href="https://github.com/jpadilla/pyjwt/commit/412cb67a93363812ae4029d6a95f5d4d40ab2609"><code>412cb67</code></a> fix: close HTTPError to prevent ResourceWarning on Python 3.14 (<a href="https://github.com/jpadilla/pyjwt/issues/1133">#1133</a>)</li> <li>Additional commits viewable in <a href="https://github.com/jpadilla/pyjwt/compare/2.10.1...2.12.0">compare view</a></li> </ul> </details> <br /> [](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores) Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`. [//]: # (dependabot-automerge-start) [//]: # (dependabot-automerge-end) --- <details> <summary>Dependabot commands and options</summary> <br /> You can trigger Dependabot actions by commenting on this PR: - `@dependabot rebase` will rebase this PR - `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it - `@dependabot show <dependency name> ignore conditions` will show all of the ignore conditions of the specified dependency - `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself) You can disable automated security fix PRs for this repo from the [Security Alerts page](https://github.com/ROCm/rocm-libraries/network/alerts). </details> Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Short story:
test/CMakeLists.txtwhich prevent ASan build from being configured (cmake --preset asan).typeinfosymbol forrocRoller::Errorfrom being undefined due to conflicting definitions in objects compiled w/o RTTI.Long story:
After fixing a typo in
test/CMakeLists.txtASan build would fail when using Ninja instead of Make (RR default) due to linking errors of undefinedtypeinfosymbol forrocRoller::Error. Such symbol was undefined because objects compiled with-fno-rttiand that useError.hpphad a WEAK definition for said symbol and these objects -- e.g.InProcessAssembler.cpp.oandFastDivision.cpp.o-- were ordered first in the link command invocation.(It is not clear to me why CMake order these objects differently when using Make & Ninja, but my suspicion is that they do because when Make is used the paths are wrapped with quotes and thus they are ordered after paths that start with
[a-zA-Z].)I verified that both Ninja & Make execute exactly the same commands with same flags, except for the ordering mentioned above. I have also verified that manually changing the order produced by CMake when using Ninja resolves the linking issues.
Since both
InProcessAssembler.cpp&FastDivision.cppdo not use LLVM headers anymore, then they don't need to be compiled with RTTI disabled (335a3a7).Once Comgr is used instead of LLVM in
ReadELF.cpp, then all logic to build objects with RTTI disabled can be safely removed and similar linking issues can be avoided.