Files
kernbench2/docs/diagrams/allreduce_latency_plots/summary.csv
T
ywkang a44f832be5 Regenerate latency plots/diagrams for post-Phase-2c model
Allreduce + pe2pe + ipcq + pe_view auto-regenerated by test sweeps
running against the new chunk-streaming wire timing (per-flit
wormhole) — absolute numbers shift upward to reflect bottleneck-link
transit charged once per flit (instead of the previous cut-through
subtraction at HBM CTRL).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-14 23:24:01 -07:00

38 lines
2.5 KiB
CSV

algorithm,sip_topology,n_sips,n_elem,bytes_per_pe,bytes_per_sip,latency_ns
intercube_allreduce,mesh_2d_no_wrap,6,8,16,256,2666.5524999999725
intercube_allreduce,mesh_2d_no_wrap,6,32,64,1024,2747.7399999999725
intercube_allreduce,mesh_2d_no_wrap,6,64,128,2048,2855.98999999998
intercube_allreduce,mesh_2d_no_wrap,6,128,256,4096,3072.4899999999725
intercube_allreduce,mesh_2d_no_wrap,6,512,1024,16384,3336.579999999951
intercube_allreduce,mesh_2d_no_wrap,6,1024,2048,32768,3707.49999999992
intercube_allreduce,mesh_2d_no_wrap,6,2048,4096,65536,4449.339999999875
intercube_allreduce,mesh_2d_no_wrap,6,4096,8192,131072,5933.020000000055
intercube_allreduce,mesh_2d_no_wrap,6,8192,16384,262144,8900.380000000157
intercube_allreduce,mesh_2d_no_wrap,6,16384,32768,524288,14835.099999997583
intercube_allreduce,mesh_2d_no_wrap,6,32768,65536,1048576,26704.540000017492
intercube_allreduce,mesh_2d_no_wrap,6,49152,98304,1572864,38573.980000026335
intercube_allreduce,ring_1d,6,8,16,256,2365.2558333333036
intercube_allreduce,ring_1d,6,32,64,1024,2436.9433333333036
intercube_allreduce,ring_1d,6,64,128,2048,2532.526666666643
intercube_allreduce,ring_1d,6,128,256,4096,2723.6933333333036
intercube_allreduce,ring_1d,6,512,1024,16384,3042.0349999999544
intercube_allreduce,ring_1d,6,1024,2048,32768,3390.201666666597
intercube_allreduce,ring_1d,6,2048,4096,65536,4079.7349999998714
intercube_allreduce,ring_1d,6,4096,8192,131072,5458.801666666721
intercube_allreduce,ring_1d,6,8192,16384,262144,8216.93500000014
intercube_allreduce,ring_1d,6,16384,32768,524288,13733.201666664638
intercube_allreduce,ring_1d,6,32768,65536,1048576,24765.735000014545
intercube_allreduce,ring_1d,6,49152,98304,1572864,35798.268333355256
intercube_allreduce,torus_2d,6,8,16,256,1700.6024999999754
intercube_allreduce,torus_2d,6,32,64,1024,1753.2899999999754
intercube_allreduce,torus_2d,6,64,128,2048,1823.539999999979
intercube_allreduce,torus_2d,6,128,256,4096,1964.0399999999754
intercube_allreduce,torus_2d,6,512,1024,16384,2196.2849999999653
intercube_allreduce,torus_2d,6,1024,2048,32768,2476.74499999995
intercube_allreduce,torus_2d,6,2048,4096,65536,3037.664999999919
intercube_allreduce,torus_2d,6,4096,8192,131072,4159.50500000003
intercube_allreduce,torus_2d,6,8192,16384,262144,6403.185000000081
intercube_allreduce,torus_2d,6,16384,32768,524288,10890.544999998769
intercube_allreduce,torus_2d,6,32768,65536,1048576,19865.265000008738
intercube_allreduce,torus_2d,6,49152,98304,1572864,28839.985000013185