einsum-opt_size 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 0.33%, worst 10.33% among cells of 1 us or more (a sub-microsecond row is timer and scheduler jitter, not kernel time); 2 of 26 engine cells exceed 5% and carry !; median within-process relative SD 0.32%. 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_size (ms) | notes | |---|---| | bin_batched_matmul_b32_m64_n64_k64/n1/b1 | 0.558625 ± 0.0116814 (W 0.00210368) | | | bin_elementwise_mul_2048x2048/n1/b1 | 4.20016 ± 0.306691 (W 0.0110112) | ! between-start range over 5% | | bin_matmul_256/n1/b1 | 1.073 ± 0.00611435 (W 0.00139987) | | | bin_outer_product_4096/n1/b1 | 6.20035 ± 0.00263468 (W 0.00886741) | | | gm_queen5_5_3.wcsp/n1/b1 | 1638.96 ± 0.949016 (W 5.31224) | | | lm_batch_likelihood_brackets_4_4d/n1/b1 | 23.5948 ± 0.203041 (W 0.0689961) | | | lm_batch_likelihood_sentence_3_12d/n1/b1 | 86.2339 ± 0.162658 (W 0.264614) | | | lm_batch_likelihood_sentence_4_4d/n1/b1 | 27.6434 ± 0.179046 (W 0.110436) | | | str_matrix_chain_multiplication_100/n1/b1 | 12.8024 ± 0.036428 (W 0.0296216) | | | str_nw_mera_closed_120/n1/b1 | 1684.38 ± 1.29182 (W 1.11625) | | | str_nw_mera_open_26/n1/b1 | 1170.04 ± 2.5658 (W 0.70401) | | | tensornetwork_permutation_focus_step409_316/n1/b1 | 232.785 ± 0.783139 (W 0.263253) | | | tensornetwork_permutation_light_415/n1/b1 | 232.258 ± 0.0191018 (W 0.423167) | |

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

| case | opt_size (ms) | notes | |---|---| | bin_batched_matmul_b32_m64_n64_k64/n1/b1 | 0.144994 ± 0.000177484 (W 0.00148718) | | | bin_elementwise_mul_2048x2048/n1/b1 | 2.1451 ± 0.0195169 (W 0.0229456) | | | bin_matmul_256/n1/b1 | 1.07171 ± 0.0026566 (W 0.00249767) | | | bin_outer_product_4096/n1/b1 | 6.27782 ± 0.00272095 (W 0.018707) | | | gm_queen5_5_3.wcsp/n1/b1 | 523.801 ± 1.58229 (W 1.29891) | | | lm_batch_likelihood_brackets_4_4d/n1/b1 | 10.6526 ± 0.0796619 (W 0.134083) | | | lm_batch_likelihood_sentence_3_12d/n1/b1 | 26.5571 ± 0.0119565 (W 0.500356) | | | lm_batch_likelihood_sentence_4_4d/n1/b1 | 12.4643 ± 0.117273 (W 0.309349) | | | str_matrix_chain_multiplication_100/n1/b1 | 12.7447 ± 0.00128269 (W 0.0263162) | | | str_nw_mera_closed_120/n1/b1 | 507.512 ± 0.822848 (W 5.17057) | | | str_nw_mera_open_26/n1/b1 | 327.28 ± 0.513942 (W 4.97966) | | | tensornetwork_permutation_focus_step409_316/n1/b1 | 88.4927 ± 0.0945939 (W 0.99491) | | | tensornetwork_permutation_light_415/n1/b1 | 77.18 ± 5.11405 (W 6.23729) | ! between-start range over 5% |

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.