batch_size, seq_len, head_num, size_per_head dataType ### batchCount, n, m, k, algoId, customOption, tile, numSplitsK, swizzle, reductionScheme, workspaceSize, stages, exec_time 64 64 32 128 1 ### 1 12288 4096 4096 6 0 20 0 1 0 0 11 1.444813 64 64 32 128 1 ### 2048 64 64 128 112 -1 -1 -1 -1 -1 -1 -1 0.083370 64 64 32 128 1 ### 2048 128 64 64 100 -1 -1 -1 -1 -1 -1 -1 0.070630 64 64 32 128 1 ### 1 4096 4096 4096 6 0 24 1 0 0 0 9 0.502825 64 64 32 128 1 ### 1 16384 4096 4096 6 0 20 0 1 0 0 11 1.898404 64 64 32 128 1 ### 1 4096 4096 16384 21 0 24 1 0 0 0 12 1.909555 64 1 32 128 1 ### 1 12288 64 4096 6 0 18 0 1 0 0 16 0.080251 64 1 32 128 1 ### 1 4096 64 4096 6 0 15 1 0 0 0 18 0.026583 64 1 32 128 1 ### 1 16384 64 4096 6 0 18 0 1 0 0 15 0.110223 64 1 32 128 1 ### 1 4096 64 16384 31 0 15 1 1 0 0 18 0.109978 64 1 32 128 1 ### 1 250880 64 4096 112 -1 -1 -1 -1 -1 -1 -1 1.602350 32 64 32 128 1 ### 1 12288 2048 4096 6 0 20 0 1 0 0 11 0.750490 32 64 32 128 1 ### 1024 64 64 128 109 -1 -1 -1 -1 -1 -1 -1 0.047020 32 64 32 128 1 ### 1024 128 64 64 108 -1 -1 -1 -1 -1 -1 -1 0.037950 32 64 32 128 1 ### 1 4096 2048 4096 6 0 20 0 0 0 0 11 0.256123 32 64 32 128 1 ### 1 16384 2048 4096 6 0 20 0 1 0 0 11 0.959887 32 64 32 128 1 ### 1 4096 2048 16384 6 0 20 0 1 0 0 11 0.979282 32 1 32 128 1 ### 1 12288 32 4096 6 0 18 0 0 0 0 16 0.078582 32 1 32 128 1 ### 1 4096 32 4096 31 0 15 1 0 0 0 18 0.024535 32 1 32 128 1 ### 1 16384 32 4096 6 0 18 0 0 0 0 12 0.105523 32 1 32 128 1 ### 1 4096 32 16384 109 -1 -1 -1 -1 -1 -1 -1 0.105160 32 1 32 128 1 ### 1 250880 32 4096 114 -1 -1 -1 -1 -1 -1 -1 1.479260 16 64 32 128 1 ### 1 12288 1024 4096 6 0 20 2 1 1 3072 11 0.398694 16 64 32 128 1 ### 512 64 64 128 105 -1 -1 -1 -1 -1 -1 -1 0.015370 16 64 32 128 1 ### 512 128 64 64 114 -1 -1 -1 -1 -1 -1 -1 0.014250 16 64 32 128 1 ### 1 4096 1024 4096 21 0 20 2 0 1 1024 11 0.144855 16 64 32 128 1 ### 1 16384 1024 4096 6 0 20 0 1 0 0 11 0.505098 16 64 32 128 1 ### 1 4096 1024 16384 111 -1 -1 -1 -1 -1 -1 -1 0.545680 16 1 32 128 1 ### 1 12288 16 4096 6 0 18 1 1 0 0 16 0.077865 16 1 32 128 1 ### 1 4096 16 4096 31 0 15 1 1 0 0 18 0.024023 16 1 32 128 1 ### 1 16384 16 4096 6 0 21 1 0 0 0 15 0.104765 16 1 32 128 1 ### 1 4096 16 16384 6 0 15 1 1 0 0 17 0.105298 16 1 32 128 1 ### 1 250880 16 4096 109 -1 -1 -1 -1 -1 -1 -1 1.450620 8 64 32 128 1 ### 1 12288 512 4096 115 -1 -1 -1 -1 -1 -1 -1 0.204910 8 64 32 128 1 ### 256 64 64 128 105 -1 -1 -1 -1 -1 -1 -1 0.010500 8 64 32 128 1 ### 256 128 64 64 109 -1 -1 -1 -1 -1 -1 -1 0.010250 8 64 32 128 1 ### 1 4096 512 4096 6 0 20 4 1 1 512 11 0.081009 8 64 32 128 1 ### 1 16384 512 4096 107 -1 -1 -1 -1 -1 -1 -1 0.257450 8 64 32 128 1 ### 1 4096 512 16384 6 0 20 5 1 1 512 11 0.256573 8 1 32 128 1 ### 1 12288 8 4096 6 0 18 1 1 0 0 16 0.077445 8 1 32 128 1 ### 1 4096 8 4096 31 0 15 1 1 0 0 18 0.023245 8 1 32 128 1 ### 1 16384 8 4096 110 -1 -1 -1 -1 -1 -1 -1 0.104450 8 1 32 128 1 ### 1 4096 8 16384 6 0 15 1 1 0 0 17 0.104192 8 1 32 128 1 ### 1 250880 8 4096 108 -1 -1 -1 -1 -1 -1 -1 1.429910 1 64 32 128 1 ### 1 12288 64 4096 109 -1 -1 -1 -1 -1 -1 -1 0.080110 1 64 32 128 1 ### 32 64 64 128 103 -1 -1 -1 -1 -1 -1 -1 0.005320 1 64 32 128 1 ### 32 128 64 64 109 -1 -1 -1 -1 -1 -1 -1 0.005470 1 64 32 128 1 ### 1 4096 64 4096 6 0 15 1 0 0 0 18 0.026429 1 64 32 128 1 ### 1 16384 64 4096 107 -1 -1 -1 -1 -1 -1 -1 0.110100 1 64 32 128 1 ### 1 4096 64 16384 31 0 15 1 1 0 0 18 0.109885 1 1 32 128 1 ### 1 12288 1 4096 6 0 18 1 1 0 0 16 0.076769 1 1 32 128 1 ### 1 4096 1 4096 6 0 15 1 1 0 0 18 0.023040 1 1 32 128 1 ### 1 16384 1 4096 105 -1 -1 -1 -1 -1 -1 -1 0.103720 1 1 32 128 1 ### 1 4096 1 16384 6 0 18 3 0 4 24576 16 0.102124 1 1 32 128 1 ### 1 250880 1 4096 102 -1 -1 -1 -1 -1 -1 -1 1.402680 64 128 32 128 1 ### 1 12288 8192 4096 6 0 20 0 1 0 0 11 2.837852 64 128 32 128 1 ### 2048 128 128 128 111 -1 -1 -1 -1 -1 -1 -1 0.202480 64 128 32 128 1 ### 2048 128 128 128 103 -1 -1 -1 -1 -1 -1 -1 0.156770 64 128 32 128 1 ### 1 4096 8192 4096 6 0 20 0 1 0 0 11 0.955003 64 128 32 128 1 ### 1 16384 8192 4096 6 0 20 0 1 0 0 11 3.772959 64 128 32 128 1 ### 1 4096 8192 16384 6 0 20 0 1 0 0 11 3.703818 64 1 32 128 1 ### 1 12288 64 4096 6 0 18 0 0 0 0 16 0.080015 64 1 32 128 1 ### 1 4096 64 4096 6 0 15 1 0 0 0 18 0.026460 64 1 32 128 1 ### 1 16384 64 4096 105 -1 -1 -1 -1 -1 -1 -1 0.110300 64 1 32 128 1 ### 1 4096 64 16384 31 0 15 1 1 0 0 18 0.109691 64 1 32 128 1 ### 1 250880 64 4096 100 -1 -1 -1 -1 -1 -1 -1 1.603500 32 128 32 128 1 ### 1 12288 4096 4096 6 0 20 0 1 0 0 11 1.444751 32 128 32 128 1 ### 1024 128 128 128 112 -1 -1 -1 -1 -1 -1 -1 0.105780 32 128 32 128 1 ### 1024 128 128 128 113 -1 -1 -1 -1 -1 -1 -1 0.084340 32 128 32 128 1 ### 1 4096 4096 4096 6 0 24 1 0 0 0 9 0.502835 32 128 32 128 1 ### 1 16384 4096 4096 6 0 20 0 1 0 0 11 1.898291 32 128 32 128 1 ### 1 4096 4096 16384 21 0 24 1 0 0 0 12 1.910139 32 1 32 128 1 ### 1 12288 32 4096 107 -1 -1 -1 -1 -1 -1 -1 0.078600 32 1 32 128 1 ### 1 4096 32 4096 31 0 15 1 0 0 0 18 0.024586 32 1 32 128 1 ### 1 16384 32 4096 6 0 18 0 1 0 0 12 0.105708 32 1 32 128 1 ### 1 4096 32 16384 105 -1 -1 -1 -1 -1 -1 -1 0.105120 32 1 32 128 1 ### 1 250880 32 4096 106 -1 -1 -1 -1 -1 -1 -1 1.480140 16 128 32 128 1 ### 1 12288 2048 4096 6 0 20 0 1 0 0 11 0.750612 16 128 32 128 1 ### 512 128 128 128 108 -1 -1 -1 -1 -1 -1 -1 0.057030 16 128 32 128 1 ### 512 128 128 128 114 -1 -1 -1 -1 -1 -1 -1 0.048080 16 128 32 128 1 ### 1 4096 2048 4096 6 0 20 0 0 0 0 11 0.256000 16 128 32 128 1 ### 1 16384 2048 4096 6 0 20 0 1 0 0 11 0.957215 16 128 32 128 1 ### 1 4096 2048 16384 6 0 20 0 1 0 0 11 0.978862 16 1 32 128 1 ### 1 12288 16 4096 6 0 18 1 1 0 0 16 0.077793 16 1 32 128 1 ### 1 4096 16 4096 31 0 15 1 1 0 0 18 0.023849 16 1 32 128 1 ### 1 16384 16 4096 6 0 21 1 0 0 0 15 0.104858 16 1 32 128 1 ### 1 4096 16 16384 6 0 15 1 1 0 0 17 0.105001 16 1 32 128 1 ### 1 250880 16 4096 108 -1 -1 -1 -1 -1 -1 -1 1.450760 8 128 32 128 1 ### 1 12288 1024 4096 6 0 20 2 1 1 3072 11 0.398592 8 128 32 128 1 ### 256 128 128 128 107 -1 -1 -1 -1 -1 -1 -1 0.018050 8 128 32 128 1 ### 256 128 128 128 104 -1 -1 -1 -1 -1 -1 -1 0.015680 8 128 32 128 1 ### 1 4096 1024 4096 21 0 20 2 0 1 1024 11 0.144763 8 128 32 128 1 ### 1 16384 1024 4096 6 0 20 0 1 0 0 11 0.505160 8 128 32 128 1 ### 1 4096 1024 16384 115 -1 -1 -1 -1 -1 -1 -1 0.545580 8 1 32 128 1 ### 1 12288 8 4096 6 0 18 1 1 0 0 16 0.077445 8 1 32 128 1 ### 1 4096 8 4096 31 0 15 1 1 0 0 18 0.023245 8 1 32 128 1 ### 1 16384 8 4096 110 -1 -1 -1 -1 -1 -1 -1 0.104360 8 1 32 128 1 ### 1 4096 8 16384 6 0 15 1 1 0 0 17 0.104305 8 1 32 128 1 ### 1 250880 8 4096 100 -1 -1 -1 -1 -1 -1 -1 1.430000 1 128 32 128 1 ### 1 12288 128 4096 6 0 18 0 1 0 0 15 0.085402 1 128 32 128 1 ### 32 128 128 128 108 -1 -1 -1 -1 -1 -1 -1 0.007070 1 128 32 128 1 ### 32 128 128 128 114 -1 -1 -1 -1 -1 -1 -1 0.007350 1 128 32 128 1 ### 1 4096 128 4096 104 -1 -1 -1 -1 -1 -1 -1 0.033170 1 128 32 128 1 ### 1 16384 128 4096 6 0 24 0 0 0 0 15 0.115405 1 128 32 128 1 ### 1 4096 128 16384 104 -1 -1 -1 -1 -1 -1 -1 0.118900 1 1 32 128 1 ### 1 12288 1 4096 6 0 18 1 1 0 0 16 0.076872 1 1 32 128 1 ### 1 4096 1 4096 6 0 15 1 1 0 0 18 0.023235 1 1 32 128 1 ### 1 16384 1 4096 107 -1 -1 -1 -1 -1 -1 -1 0.103860 1 1 32 128 1 ### 1 4096 1 16384 6 0 18 3 0 4 24576 16 0.102523 1 1 32 128 1 ### 1 250880 1 4096 103 -1 -1 -1 -1 -1 -1 -1 1.402790