einsum-opt_flops on zen5-cpu

What was measured

Hardware

Every timed cell matches a successful verification census, including operation variants. Each cell is median of process medians ± sample SD of those medians; W is the median within-process sample SD. All values are milliseconds. These are descriptive spreads, not confidence intervals. The between-process spread includes the rotating-order schedule. Each process retains 5 chronological calls; 2 complete process sets are recorded. Spread guide (computed from this run's own rows): per-engine between-start range median 0.44%, worst 4.85% among cells of 1 us or more (a sub-microsecond row is timer and scheduler jitter, not kernel time); 0 of 26 engine cells exceed 5% and carry !; median within-process relative SD 0.48%. A cell whose ±, ! or p is as large as the difference you are reading is not evidence of a regression. no row, so no ratio there is position-balanced). ! marks a cell where some engine's own per-start median range exceeds 5% of its median, and p a balanced cell whose estimated position-effect ratio exceeds 1.05. Flagged cells are retained, not discarded; the between-start variation can exceed the difference being compared. ! between-start range over 5%, p position effect over 5%, dispatch-dominated means the row is under 20 us so 4T/8T pay thread-region start-up). Per-start ratios and every chronological sample are in the run directory named at the top of this page (run*.csv). These describe the recorded revisions; they are not a regression claim.

cpueinsum, float64, 1T (CPU 4)

| case | opt_flops (ms) | notes | |---|---| | bin_batched_matmul_b32_m64_n64_k64/n1/b1 | 0.556666 ± 0.00781494 (W 0.0019814) | | | bin_elementwise_mul_2048x2048/n1/b1 | 4.04907 ± 0.0441963 (W 0.0151012) | | | bin_matmul_256/n1/b1 | 1.07433 ± 0.00288429 (W 0.00219764) | | | bin_outer_product_4096/n1/b1 | 6.2375 ± 0.0498913 (W 0.0321347) | | | gm_queen5_5_3.wcsp/n1/b1 | 4505.55 ± 89.7036 (W 16.197) | | | lm_batch_likelihood_brackets_4_4d/n1/b1 | 26.7835 ± 0.919324 (W 0.134745) | | | lm_batch_likelihood_sentence_3_12d/n1/b1 | 73.1295 ± 0.189702 (W 0.133277) | | | lm_batch_likelihood_sentence_4_4d/n1/b1 | 27.3547 ± 0.20067 (W 0.0597421) | | | str_matrix_chain_multiplication_100/n1/b1 | 12.6843 ± 0.197837 (W 0.21469) | | | str_nw_mera_closed_120/n1/b1 | 1770.2 ± 4.46369 (W 0.930345) | | | str_nw_mera_open_26/n1/b1 | 1176.03 ± 1.01757 (W 1.10771) | | | tensornetwork_permutation_focus_step409_316/n1/b1 | 244.561 ± 0.725674 (W 0.221339) | | | tensornetwork_permutation_light_415/n1/b1 | 235.707 ± 4.37737 (W 0.450369) | |

cpueinsum, float64, 4T (CPU 4-7)

| case | opt_flops (ms) | notes | |---|---| | bin_batched_matmul_b32_m64_n64_k64/n1/b1 | 0.144225 ± 0.000361332 (W 0.000975819) | | | bin_elementwise_mul_2048x2048/n1/b1 | 2.17117 ± 0.00455377 (W 0.0172388) | | | bin_matmul_256/n1/b1 | 0.315926 ± 7.77817e-05 (W 0.01058) | | | bin_outer_product_4096/n1/b1 | 4.27968 ± 0.00548786 (W 0.0622154) | | | gm_queen5_5_3.wcsp/n1/b1 | 1834.1 ± 10.9817 (W 3.53733) | | | lm_batch_likelihood_brackets_4_4d/n1/b1 | 12.1717 ± 0.0629353 (W 0.062057) | | | lm_batch_likelihood_sentence_3_12d/n1/b1 | 23.3578 ± 0.502956 (W 0.59669) | | | lm_batch_likelihood_sentence_4_4d/n1/b1 | 15.7073 ± 0.0438236 (W 0.320124) | | | str_matrix_chain_multiplication_100/n1/b1 | 6.73405 ± 0.0216247 (W 0.0825944) | | | str_nw_mera_closed_120/n1/b1 | 518.378 ± 0.0598347 (W 4.83988) | | | str_nw_mera_open_26/n1/b1 | 320.379 ± 0.949285 (W 1.74368) | | | tensornetwork_permutation_focus_step409_316/n1/b1 | 89.6096 ± 0.148932 (W 1.20518) | | | tensornetwork_permutation_light_415/n1/b1 | 76.0641 ± 1.66933 (W 4.80084) | |

Repeated process sets describe observed repeatability under the recorded order schedule; they are not a paired baseline/candidate experiment and do not establish equal performance between engines.