← all results

CPU benchmark results

How to read

1.02x
this case's median divided by the baseline median. Below 1.00x is faster, above is slower. We never average cases: a 32x32 batched matmul and a 3072 matmul are different workloads.
baseline
PyTorch wherever the suite measured it; otherwise the case's declared reference arm. The dashed line is that baseline at 1.00x.
bars
tenferro eager (blue) and trace (purple) are filled; PyTorch, JAX and cuTENSOR bars are outlined white so the reference arms stay quiet.
whisker
estimated 95% CI of the median from the case's measured runs (IQR and sample count). A difference smaller than the whisker is not established by this run.
regimes
compute/bandwidth-bound results and latency/overhead results are on separate pages and are never mixed. A small workload is not evidence of slow compute.

Compute- and bandwidth-bound CPU cases only. Per-call latency results are on the latency page.

AMD Ryzen AI 9 HX 470

AMD Ryzen AI 9 HX 470zen5-cpuhardware: x86_64date: 2026-10-28 01:14:23Ztenferro: 8fec1ca70852
cpu/cpu_ops 897 cases · 186 flagged

baseline PyTorch · best first · 897 cases · case table, absolute times and raw report →

bar = case median / PyTorch median · dashed line = baseline (1.00x) · log scale · shorter (left) is faster · whisker = estimated 95% CIsolveeager 0.00xtrace 0.02xJAX 0.10xsolveeager 0.01xtrace 0.01xJAX 0.02xsolveeager 0.01xtrace 0.03xJAX 0.10xsolveeager 0.01xtrace 0.03xJAX 0.06xsolveeager 0.01xtrace 0.05xJAX 1.29xsolveeager 0.01xtrace 0.04xJAX 0.15xsolveeager 0.01xtrace 0.02xJAX 1.51xsolveeager 0.01xtrace 0.03xJAX 0.20xsolveeager 0.01xtrace 0.05xJAX 1.80xgrad_sum_solve_backwardtrace 0.01xJAX 0.11xsolveeager 0.02xtrace 0.05xJAX 0.05xsolveeager 0.02xtrace 0.04xJAX 0.39xsolveeager 0.02xtrace 0.06xJAX 0.56xsolveeager 0.02xtrace 0.04xJAX 2.73xsolveeager 0.03xtrace 0.08xJAX 0.33xsolveeager 0.04xtrace 0.28xJAX 1.28xsolveeager 0.05xtrace 0.05xJAX 1.73xgrad_sum_solve_jvptrace 0.06xJAX 0.04xgrad_sum_solve_jvptrace 0.06xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xsolveeager 0.07xtrace 0.13xJAX 1.40xgrad_sum_solve_jvptrace 0.07xJAX 0.05xgrad_sum_solve_backwardtrace 0.07xJAX 0.11xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.08xJAX 0.05xgrad_sum_solve_jvptrace 0.08xJAX 0.05xgrad_sum_solve_backwardtrace 0.08xJAX 0.04xgrad_sum_solve_jvptrace 0.09xJAX 0.09xgrad_sum_solve_backwardtrace 0.09xJAX 0.08xgrad_sum_svd_s_jvptrace 0.10xJAX 0.07xgrad_sum_solve_jvptrace 0.10xJAX 0.05xgrad_sum_solve_jvptrace 0.10xJAX 0.05xgrad_sum_svd_s_jvptrace 0.11xJAX 0.07xgrad_sum_svd_s_jvptrace 0.11xJAX 0.07xgrad_sum_svd_s_jvptrace 0.12xJAX 0.09xgrad_sum_batched_matmul_backwardtrace 0.12xJAX 0.15xgrad_sum_batched_matmul_backwardtrace 0.12xJAX 0.15xgrad_sum_batched_matmul_backwardtrace 0.13xJAX 0.17xgrad_sum_batched_matmul_backwardtrace 0.13xJAX 0.21xgrad_sum_batched_matmul_backwardtrace 0.13xJAX 0.19xgrad_sum_svd_s_jvptrace 0.13xJAX 0.09xgrad_sum_batched_matmul_backwardtrace 0.13xJAX 0.25xgrad_sum_batched_matmul_backwardtrace 0.13xJAX 0.20xgrad_sum_svd_s_jvptrace 0.13xJAX 0.09xgrad_sum_batched_matmul_backwardtrace 0.13xJAX 0.27xgrad_sum_batched_matmul_backwardtrace 0.13xJAX 0.18xgrad_sum_batched_matmul_backwardtrace 0.13xJAX 0.17xgrad_sum_batched_matmul_backwardtrace 0.14xJAX 0.15xgrad_sum_solve_vjptrace 0.14xJAX 0.10xgrad_sum_batched_matmul_backwardtrace 0.14xJAX 0.14xgrad_sum_solve_vjptrace 0.15xJAX 0.11xgrad_sum_solve_vjptrace 0.15xJAX 0.11xeigheager 0.25xtrace 0.15xJAX 0.13xgrad_sum_solve_vjptrace 0.15xJAX 0.13xgrad_sum_svd_s_vjptrace 0.15xJAX 0.12xgrad_sum_lu_jvptrace 0.16xJAX 0.07xgrad_sum_lu_jvptrace 0.16xJAX 0.07xgrad_sum_solve_vjptrace 0.16xJAX 0.11xgrad_sum_solve_vjptrace 0.16xJAX 0.10xgrad_sum_solve_vjptrace 0.16xJAX 0.11xgrad_sum_svd_s_vjptrace 0.16xJAX 0.12xgrad_sum_lu_jvptrace 0.17xJAX 0.10xgrad_sum_solve_vjptrace 0.17xJAX 0.11xgrad_sum_svd_s_vjptrace 0.17xJAX 0.13xgrad_sum_solve_vjptrace 0.17xJAX 0.11xgrad_sum_solve_vjptrace 0.17xJAX 0.12xgrad_sum_solve_vjptrace 0.17xJAX 0.14xgrad_sum_lu_jvptrace 0.17xJAX 0.07xgrad_sum_lu_jvptrace 0.18xJAX 0.07xgrad_sum_solve_vjptrace 0.18xJAX 0.14xgrad_sum_lu_jvptrace 0.18xJAX 0.07xgrad_sum_lu_jvptrace 0.18xJAX 0.10xgrad_sum_svd_s_jvptrace 0.18xJAX 0.16xgrad_sum_lu_jvptrace 0.18xJAX 0.12xgrad_sum_svd_s_vjptrace 0.18xJAX 0.15xgrad_sum_lu_jvptrace 0.18xJAX 0.07xgrad_sum_lu_jvptrace 0.18xJAX 0.11xgrad_sum_svd_s_jvptrace 0.19xJAX 0.11xgrad_sum_lu_vjptrace 0.19xJAX 0.09xgrad_sum_svd_s_jvptrace 0.19xJAX 0.11xgrad_sum_lu_vjptrace 0.19xJAX 0.09xgrad_sum_svd_s_vjptrace 0.19xJAX 0.15xgrad_sum_lu_vjptrace 0.19xJAX 0.13xgrad_sum_lu_jvptrace 0.19xJAX 0.09xgrad_sum_lu_jvptrace 0.19xJAX 0.09xgrad_sum_solve_vjptrace 0.20xJAX 0.21xgrad_sum_svd_s_vjptrace 0.20xJAX 0.16xgrad_sum_batched_matmul_backwardtrace 0.20xJAX 0.14xgrad_sum_solve_backwardtrace 0.20xJAX 1.35xgrad_sum_batched_matmul_backwardtrace 0.20xJAX 0.25xgrad_sum_lu_vjptrace 0.20xJAX 0.09xgrad_sum_lu_vjptrace 0.21xJAX 0.08xgrad_sum_lu_vjptrace 0.21xJAX 0.12xgrad_sum_lu_vjptrace 0.21xJAX 0.09xgrad_sum_eigh_jvptrace 0.21xJAX 0.10xgrad_sum_batched_matmul_backwardtrace 0.21xJAX 0.22xgrad_sum_batched_matmul_backwardtrace 0.21xJAX 0.16xgrad_sum_batched_matmul_backwardtrace 0.21xJAX 0.15xgrad_sum_lu_vjptrace 0.21xJAX 0.09xgrad_sum_batched_matmul_backwardtrace 0.22xJAX 0.19xgrad_sum_solve_vjptrace 0.22xJAX 0.13xgrad_sum_lu_vjptrace 0.22xJAX 0.13xeigheager 0.22xtrace 0.22xJAX 0.99xgrad_sum_solve_vjptrace 0.22xJAX 0.12xgrad_sum_lu_vjptrace 0.22xJAX 0.16xgrad_sum_eigh_jvptrace 0.23xJAX 0.11xgrad_sum_lu_vjptrace 0.23xJAX 0.11xgrad_sum_eigh_jvptrace 0.24xJAX 0.10xgrad_sum_batched_matmul_backwardtrace 0.24xJAX 0.29xgrad_sum_eigh_vjptrace 0.24xJAX 0.12xsolveeager 0.24xtrace 0.25xJAX 0.58xgrad_sum_lu_vjptrace 0.24xJAX 0.11xgrad_sum_lu_jvptrace 0.24xJAX 0.29xgrad_sum_eigh_jvptrace 0.25xJAX 0.10xgrad_sum_eigh_jvptrace 0.25xJAX 0.18xgrad_sum_eigh_jvptrace 0.25xJAX 0.12xgrad_sum_lu_jvptrace 0.25xJAX 0.19xgrad_sum_batched_matmul_backwardtrace 0.26xJAX 0.28xgrad_sum_eigh_vjptrace 0.26xJAX 0.14xgrad_sum_eigh_jvptrace 0.26xJAX 0.12xgrad_sum_svd_s_vjptrace 0.27xJAX 0.26xgrad_sum_svd_s_vjptrace 0.27xJAX 0.16xgrad_sum_batched_matmul_backwardtrace 0.27xJAX 0.25xgrad_sum_batched_matmul_backwardtrace 0.27xJAX 0.29xgrad_sum_eigh_vjptrace 0.27xJAX 0.11xgrad_sum_eigh_jvptrace 0.28xJAX 0.12xgrad_sum_svd_s_jvptrace 0.28xJAX 0.30xgrad_sum_lu_jvptrace 0.28xJAX 0.19xgrad_sum_eigh_vjptrace 0.28xJAX 0.12xgrad_sum_eigh_vjptrace 0.28xJAX 0.14xgrad_sum_svd_s_vjptrace 0.28xJAX 0.17xgrad_sum_batched_matmul_backwardtrace 0.28xJAX 0.30xgrad_sum_svd_s_jvptrace 0.28xJAX 0.26xgrad_sum_svd_s_jvptrace 0.28xJAX 0.26xgrad_sum_batched_matmul_backwardtrace 0.29xJAX 0.35xgrad_sum_eigh_vjptrace 0.29xJAX 0.24xgrad_sum_eigh_jvptrace 0.29xJAX 0.12xgrad_sum_batched_matmul_backwardtrace 0.30xJAX 0.37xgrad_sum_eigh_vjptrace 0.30xJAX 0.15xeigheager 0.30xtrace 0.42xJAX 0.62xgrad_sum_qr_vjptrace 0.31xJAX 0.09xeigheager 0.72xtrace 0.31xJAX 1.21xgrad_sum_eigh_vjptrace 0.31xJAX 0.15xgrad_sum_batched_matmul_backwardtrace 0.31xJAX 0.46xgrad_sum_qr_vjptrace 0.32xJAX 0.10xgrad_sum_batched_matmul_backwardtrace 0.32xJAX 0.57xgrad_sum_batched_matmul_backwardtrace 0.32xJAX 0.34xgrad_sum_qr_vjptrace 0.32xJAX 0.14xsolveeager 0.32xtrace 0.65xJAX 6.83xgrad_sum_batched_matmul_backwardtrace 0.33xJAX 0.45xgrad_sum_matmul_backwardtrace 0.33xJAX 0.37xgrad_sum_eigh_vjptrace 0.33xJAX 0.15xgrad_sum_qr_jvptrace 0.33xJAX 0.10xgrad_sum_qr_vjptrace 0.33xJAX 0.08xmatmuleager 0.37xtrace 0.34xJAX 0.99xgrad_sum_eigh_jvptrace 0.34xJAX 0.34xgrad_sum_qr_vjptrace 0.34xJAX 0.10xgrad_sum_qr_jvptrace 0.34xJAX 0.15xgrad_sum_batched_matmul_backwardtrace 0.34xJAX 0.45xgrad_sum_lu_vjptrace 0.34xJAX 0.24xgrad_sum_qr_jvptrace 0.34xJAX 0.10xsolveeager 0.34xtrace 0.68xJAX 1.83xgrad_sum_matmul_backwardtrace 0.34xJAX 0.36xgrad_sum_qr_vjptrace 0.34xJAX 0.11xgrad_sum_solve_backwardtrace 0.34xJAX 0.68xgrad_sum_batched_matmul_backwardtrace 0.34xJAX 0.46xgrad_sum_batched_matmul_backwardtrace 0.34xJAX 0.49xgrad_sum_qr_vjptrace 0.34xJAX 0.19xgrad_sum_matmul_backwardtrace 0.35xJAX 0.58xgrad_sum_qr_vjptrace 0.35xJAX 0.09xgrad_sum_eigh_jvptrace 0.35xJAX 0.27xgrad_sum_matmul_backwardtrace 0.35xJAX 0.37xgrad_sum_qr_vjptrace 0.35xJAX 0.10xgrad_sum_batched_matmul_backwardtrace 0.35xJAX 0.24xgrad_sum_batched_matmul_backwardtrace 0.36xJAX 0.49xgrad_sum_qr_vjptrace 0.36xJAX 0.12xgrad_sum_qr_jvptrace 0.36xJAX 0.09xgrad_sum_matmul_backwardtrace 0.36xJAX 0.36xgrad_sum_matmul_backwardtrace 0.36xJAX 0.43xgrad_sum_qr_jvptrace 0.36xJAX 0.20xgrad_sum_qr_jvptrace 0.36xJAX 0.14xgrad_sum_qr_vjptrace 0.36xJAX 0.11xgrad_sum_qr_jvptrace 0.36xJAX 0.09xgrad_sum_batched_matmul_backwardtrace 0.36xJAX 0.17xgrad_sum_eigh_jvptrace 0.36xJAX 0.27xgrad_sum_matmul_backwardtrace 0.36xJAX 0.43xgrad_sum_batched_matmul_backwardtrace 0.37xJAX 0.18xgrad_sum_qr_jvptrace 0.37xJAX 0.11xgrad_sum_qr_jvptrace 0.37xJAX 0.11xgrad_sum_matmul_backwardtrace 0.37xJAX 0.51xgrad_sum_svd_s_backwardtrace 0.37xJAX 0.28xgrad_sum_qr_jvptrace 0.37xJAX 0.14xgrad_sum_qr_jvptrace 0.37xJAX 0.12xgrad_sum_matmul_backwardtrace 0.38xJAX 0.41xgrad_sum_matmul_backwardtrace 0.38xJAX 0.41xgrad_sum_qr_vjptrace 0.38xJAX 0.11xmatmul_recteager 0.38xtrace 0.38xJAX 0.90xsolveeager 0.38xtrace 0.40xJAX 0.90xgrad_sum_solve_backwardtrace 0.38xJAX 0.30xgrad_sum_eigh_vjptrace 0.38xJAX 0.40xsolveeager 0.38xtrace 0.48xJAX 0.74xmatmuleager 0.55xtrace 0.39xJAX 1.10xgrad_sum_qr_jvptrace 0.39xJAX 0.12xsolveeager 0.39xtrace 0.47xJAX 0.77xgrad_sum_solve_backwardtrace 0.39xJAX 0.32xgrad_sum_solve_vjptrace 0.39xJAX 0.42xgrad_sum_svd_s_backwardtrace 0.39xJAX 0.27xgrad_sum_svd_s_vjptrace 0.39xJAX 0.44xgrad_sum_solve_vjptrace 0.40xJAX 0.48xgrad_sum_matmul_backwardtrace 0.40xJAX 0.47xgrad_sum_solve_backwardtrace 0.40xJAX 0.24xgrad_sum_svd_s_vjptrace 0.40xJAX 0.38xgrad_sum_eigh_vjptrace 0.41xJAX 0.32xgrad_sum_svd_s_backwardtrace 0.41xJAX 0.31xgrad_sum_svd_s_backwardtrace 0.41xJAX 0.31xgrad_sum_matmul_backwardtrace 0.41xJAX 0.44xeinsum_ij_jk_ikeager 0.41xtrace 0.68xJAX 9.83xgrad_sum_svd_s_vjptrace 0.41xJAX 0.39xgrad_sum_solve_backwardtrace 0.41xJAX 0.32xeinsum_ij_jk_ikeager 0.42xtrace 0.67xJAX 10.77xgrad_sum_eigh_vjptrace 0.42xJAX 0.32xsolveeager 0.42xtrace 0.54xJAX 0.77xeinsum_ij_jk_ikeager 0.42xtrace 0.71xJAX 10.14xsolveeager 0.42xtrace 0.54xJAX 0.76xgrad_sum_lu_jvptrace 0.42xJAX 0.40xeinsum_ij_jk_ikeager 0.43xtrace 0.72xJAX 10.28xmatmuleager 0.45xtrace 0.43xJAX 1.05xsolveeager 0.43xtrace 0.51xJAX 0.72xgrad_sum_lu_jvptrace 0.43xJAX 0.74xgrad_sum_qr_vjptrace 0.43xJAX 0.32xgrad_sum_svd_s_backwardtrace 0.44xJAX 0.33xgrad_sum_lu_vjptrace 0.44xJAX 0.30xgrad_sum_solve_backwardtrace 0.44xJAX 0.29xgrad_sum_solve_backwardtrace 0.44xJAX 0.31xgrad_sum_matmuleager 0.67xtrace 0.44xJAX 1.20xgrad_sum_svd_s_backwardtrace 0.44xJAX 0.34xsolveeager 0.44xtrace 0.50xJAX 0.81xeinsum_ij_jk_ikeager 0.45xtrace 0.72xJAX 10.36xgrad_sum_batched_solve_backwardtrace 0.45xJAX 0.37xeinsum_ij_jk_ikeager 0.45xtrace 0.68xJAX 9.74xgrad_sum_batched_solve_backwardtrace 0.45xJAX 0.51xgrad_sum_solve_vjptrace 0.45xJAX 0.47xgrad_sum_solve_backwardtrace 0.45xJAX 0.23xbatched_matmul_ikb_kjb_ijbeager 0.47xtrace 0.45xJAX 9.14xgrad_sum_solve_backwardtrace 0.45xJAX 0.31xgrad_sum_lu_vjptrace 0.45xJAX 0.25xsolveeager 0.45xtrace 0.49xJAX 0.81xmatmuleager 0.57xtrace 0.46xJAX 1.18xgrad_sum_solve_backwardtrace 0.46xJAX 0.29xgrad_sum_qr_jvptrace 0.46xJAX 0.32xsolveeager 0.46xtrace 0.50xJAX 0.67xgrad_sum_lu_jvptrace 0.46xJAX 0.46xeinsum_ij_jk_ikeager 0.46xtrace 0.70xJAX 9.98xsolveeager 0.46xtrace 0.57xJAX 0.73xsolveeager 0.46xtrace 0.54xJAX 0.68xeigheager 0.46xtrace 0.46xJAX 18.34xgrad_sum_svd_s_jvptrace 0.46xJAX 0.48xgrad_sum_qr_vjptrace 0.46xJAX 0.31xsolveeager 0.47xtrace 0.59xJAX 0.80xsolveeager 0.47xtrace 0.57xJAX 0.84xbatched_matmul_ikb_kjb_ijbeager 0.47xtrace 0.48xJAX 8.84xgrad_sum_solve_backwardtrace 0.47xJAX 0.29xsolveeager 0.47xtrace 0.57xJAX 0.83xsolveeager 0.47xtrace 0.59xJAX 0.77xgrad_sum_solve_backwardtrace 0.48xJAX 0.31xeinsum_ij_jk_ikeager 0.48xtrace 0.68xJAX 9.02xgrad_sum_qr_jvptrace 0.48xJAX 1.92xbatched_matmul_ikb_kjb_ijbeager 0.49xtrace 0.48xJAX 8.18xbatched_matmul_ikb_kjb_ijbeager 0.48xtrace 0.49xJAX 8.66xgrad_sum_batched_solve_backwardtrace 0.48xJAX 0.38xgrad_sum_solve_backwardtrace 0.48xJAX 0.38xeinsum_ij_jk_ikeager 0.49xtrace 0.72xJAX 10.05xgrad_sum_qr_vjptrace 0.49xJAX 0.30xgrad_sum_batched_solve_backwardtrace 0.49xJAX 0.36xbatched_matmul_ikb_kjb_ijbeager 0.50xtrace 0.49xJAX 9.30xeinsum_ij_jk_ikeager 0.49xtrace 0.68xJAX 10.16xmatmul_recteager 0.49xtrace 0.50xJAX 1.17xgrad_sum_svd_s_jvptrace 0.49xJAX 0.45xgrad_sum_batched_solve_backwardtrace 0.49xJAX 0.37xgrad_sum_solve_backwardtrace 0.49xJAX 0.52xmatmuleager 2.77xtrace 0.49xJAX 1.28xgrad_sum_svd_s_jvptrace 0.49xJAX 0.45xgrad_sum_matmul_backwardtrace 0.49xJAX 0.71xeinsum_ij_jk_ikeager 0.49xtrace 0.70xJAX 9.95xsolveeager 0.50xtrace 0.57xJAX 0.73xsolveeager 0.50xtrace 0.62xJAX 0.92xgrad_sum_batched_solve_backwardtrace 0.50xJAX 0.40xsolveeager 0.50xtrace 0.56xJAX 0.69xgrad_sum_lu_jvptrace 0.50xJAX 0.48xsolveeager 0.50xtrace 0.57xJAX 0.72xeinsum_ij_jk_ikeager 0.50xtrace 0.68xJAX 9.10xsolveeager 0.50xtrace 0.60xJAX 0.69xsolveeager 0.50xtrace 0.52xJAX 0.79xsolveeager 0.50xtrace 0.60xJAX 0.85xgrad_sum_matmuleager 1.03xtrace 0.51xJAX 0.71xgrad_sum_qr_jvptrace 0.51xJAX 0.59xgrad_sum_batched_solve_backwardtrace 0.51xJAX 0.62xgrad_sum_batched_solve_backwardtrace 0.51xJAX 0.81xbatched_matmul_ikb_kjb_ijbeager 0.51xtrace 0.53xJAX 7.96xgrad_sum_eigh_jvptrace 0.51xJAX 0.57xsolveeager 0.51xtrace 0.62xJAX 0.81xbatched_matmul_ikb_kjb_ijbeager 0.54xtrace 0.51xJAX 7.89xgrad_sum_matmul_backwardtrace 0.52xJAX 0.45xbatched_matmul_ikb_kjb_ijbeager 0.54xtrace 0.52xJAX 9.75xgrad_sum_matmul_backwardtrace 0.52xJAX 0.39xgrad_sum_matmuleager 0.99xtrace 0.52xJAX 1.66xbatched_matmul_ikb_kjb_ijbeager 0.52xtrace 0.53xJAX 9.53xmatmuleager 0.53xtrace 0.56xJAX 0.82xgrad_sum_lu_jvptrace 0.53xJAX 0.80xeinsum_ij_jk_ikeager 0.53xtrace 0.67xJAX 7.61xgrad_sum_svd_s_backwardtrace 0.53xJAX 0.55xgrad_sum_solve_backwardtrace 0.54xJAX 0.38xgrad_sum_solve_vjptrace 0.54xJAX 0.99xeinsum_ij_jk_ikeager 0.54xtrace 0.67xJAX 6.50xsolveeager 0.54xtrace 0.57xJAX 0.79xgrad_sum_svd_s_backwardtrace 0.55xJAX 0.41xgrad_sum_eigh_jvptrace 0.55xJAX 0.52xgrad_sum_eigh_jvptrace 0.55xJAX 0.54xsolveeager 0.55xtrace 0.60xJAX 0.80xgrad_sum_solve_backwardtrace 0.55xJAX 0.34xgrad_sum_batched_solve_backwardtrace 0.56xJAX 0.50xeinsum_ij_jk_ikeager 0.56xtrace 0.69xJAX 6.79xgrad_sum_svd_s_backwardtrace 0.56xJAX 0.36xgrad_sum_batched_solve_backwardtrace 0.57xJAX 0.54xbatched_matmul_ikb_kjb_ijbeager 0.57xtrace 0.60xJAX 6.54xgrad_sum_matmul_backwardtrace 0.58xJAX 0.90xmatmuleager 0.58xtrace 0.58xJAX 0.91xgrad_sum_eigh_vjptrace 0.58xJAX 0.66xgrad_sum_batched_solve_backwardtrace 0.58xJAX 0.84xbatched_matmul_ikb_kjb_ijbeager 0.63xtrace 0.59xJAX 5.07xgrad_sum_batched_matmul_backwardtrace 0.59xJAX 0.99xgrad_sum_batched_solve_backwardtrace 0.59xJAX 0.95xbatched_matmul_ikb_kjb_ijbeager 0.59xtrace 0.61xJAX 6.04xgrad_sum_batched_matmul_backwardtrace 0.59xJAX 0.75xgrad_sum_solve_backwardtrace 0.59xJAX 0.75xgrad_sum_batched_solve_backwardtrace 0.60xJAX 0.82xgrad_sum_svd_s_vjptrace 0.60xJAX 0.63xgrad_sum_batched_solve_backwardtrace 0.60xJAX 0.82xgrad_sum_batched_solve_backwardtrace 0.60xJAX 1.01xbatched_solveeager 0.64xtrace 0.60xJAX 1.00xgrad_sum_batched_matmul_backwardtrace 0.60xJAX 0.98xbatched_matmul_ikb_kjb_ijbeager 0.61xtrace 0.61xJAX 6.17xgrad_sum_solve_backwardtrace 0.62xJAX 0.83xbatched_solveeager 0.62xtrace 0.62xJAX 0.98xbatched_solveeager 0.62xtrace 0.63xJAX 0.78xbatched_solveeager 0.64xtrace 0.62xJAX 0.82xgrad_sum_batched_solve_backwardtrace 0.62xJAX 0.92xgrad_sum_eigh_vjptrace 0.63xJAX 0.59xgrad_sum_solve_backwardtrace 0.63xJAX 0.66xgrad_sum_solve_backwardtrace 0.63xJAX 0.71xsolveeager 0.63xtrace 0.65xJAX 0.78xgrad_sum_eigh_vjptrace 0.63xJAX 0.60xsolveeager 0.63xtrace 0.65xJAX 0.77xgrad_sum_svd_s_vjptrace 0.63xJAX 0.58xgrad_sum_batched_solve_backwardtrace 0.64xJAX 0.71xgrad_sum_batched_solve_backwardtrace 0.64xJAX 0.89xgrad_sum_svd_s_vjptrace 0.64xJAX 0.59xgrad_sum_batched_solve_backwardtrace 0.64xJAX 0.73xsolveeager 0.64xtrace 0.66xJAX 0.78xbatched_solveeager 0.70xtrace 0.64xJAX 1.00xeigheager 0.65xtrace 0.67xJAX 4.60xbatched_solveeager 0.69xtrace 0.65xJAX 1.33xgrad_sum_solve_jvptrace 0.65xJAX 0.68xsolveeager 0.65xtrace 0.69xJAX 1.10xgrad_sum_matmul_backwardtrace 0.66xJAX 1.03xsolveeager 0.66xtrace 0.70xJAX 1.13xgrad_sum_solve_jvptrace 0.66xJAX 0.79xgrad_sum_matmul_backwardtrace 0.66xJAX 0.80xgrad_sum_svd_s_backwardtrace 0.66xJAX 0.75xgrad_sum_svd_s_backwardtrace 0.67xJAX 0.70xgrad_sum_batched_solve_backwardtrace 0.67xJAX 1.12xgrad_sum_solve_jvptrace 0.67xJAX 0.78xsolveeager 0.67xtrace 0.72xJAX 0.97xgrad_sum_matmul_backwardtrace 0.68xJAX 1.18xbatched_matmul_ikb_kjb_ijbeager 0.72xtrace 0.68xJAX 4.98xmatmuleager 0.75xtrace 0.68xJAX 1.02xmatmuleager 0.78xtrace 0.68xJAX 1.40xgrad_sum_matmul_backwardtrace 0.68xJAX 0.77xbatched_solveeager 0.69xtrace 0.68xJAX 0.87xgrad_sum_svd_s_backwardtrace 0.68xJAX 0.70xsolveeager 0.68xtrace 0.74xJAX 0.76xgrad_sum_batched_solve_backwardtrace 0.69xJAX 1.14xgrad_sum_lu_jvptrace 0.69xJAX 0.74xbatched_matmul_ikb_kjb_ijbeager 0.69xtrace 0.71xJAX 5.71xgrad_sum_solve_backwardtrace 0.70xJAX 0.73xmatmul_recteager 0.70xtrace 0.70xJAX 0.85xbatched_matmul_ikb_kjb_ijbeager 0.71xtrace 0.70xJAX 1.27xgrad_sum_batched_solve_backwardtrace 0.71xJAX 0.95xbatched_solveeager 0.75xtrace 0.71xJAX 0.91xgrad_sum_svd_s_jvptrace 0.72xJAX 0.75xgrad_sum_svd_s_jvptrace 0.72xJAX 1.12xgrad_sum_svd_s_backwardtrace 0.72xJAX 1.20xgrad_sum_solve_backwardtrace 0.72xJAX 0.91xsolveeager 0.72xtrace 0.80xJAX 0.75xbatched_solveeager 0.74xtrace 0.72xJAX 1.50xsolveeager 0.72xtrace 0.75xJAX 0.67xgrad_sum_svd_s_jvptrace 0.72xJAX 1.35xgrad_sum_eigh_jvptrace 0.72xJAX 0.74xgrad_sum_svd_s_jvptrace 0.72xJAX 0.75xsolveeager 0.73xtrace 0.74xJAX 1.06xqreager 0.80xtrace 0.73xJAX 1.12xmatmuleager 0.77xtrace 0.73xJAX 1.03xbatched_matmul_ikb_kjb_ijbeager 0.74xtrace 0.80xJAX 9.48xgrad_sum_eigh_jvptrace 0.74xJAX 1.18xgrad_sum_batched_solve_backwardtrace 0.74xJAX 1.43xbatched_matmul_ikb_kjb_ijbeager 0.75xtrace 0.74xJAX 1.51xgrad_sum_solve_backwardtrace 0.74xJAX 0.85xsolveeager 0.75xtrace 0.75xJAX 1.03xgrad_sum_svd_s_jvptrace 0.75xJAX 1.25xgrad_sum_solve_backwardtrace 0.75xJAX 0.92xbatched_solveeager 0.76xtrace 0.75xJAX 1.05xgrad_sum_eigh_jvptrace 0.76xJAX 0.78xgrad_sum_solve_backwardtrace 0.76xJAX 0.83xbatched_matmul_ikb_kjb_ijbeager 0.77xtrace 0.76xJAX 2.74xgrad_sum_batched_solve_backwardtrace 0.76xJAX 0.90xbatched_matmul_ikb_kjb_ijbeager 0.76xtrace 0.78xJAX 5.25xgrad_sum_qr_vjptrace 0.76xJAX 1.43xqreager 0.92xtrace 0.77xJAX 1.42xsolveeager 0.77xtrace 1.25xJAX 20.19xbatched_matmul_ikb_kjb_ijbeager 0.77xtrace 0.82xJAX 9.03xgrad_sum_svd_s_backwardtrace 0.77xJAX 0.84xsolveeager 0.77xtrace 0.78xJAX 1.05xqreager 0.85xtrace 0.77xJAX 1.27xqreager 0.86xtrace 0.78xJAX 1.59xgrad_sum_svd_s_jvptrace 0.78xJAX 0.80xsolveeager 0.78xtrace 0.80xJAX 1.60xgrad_sum_matmuleager 0.81xtrace 0.78xJAX 0.91xgrad_sum_batched_solve_backwardtrace 0.78xJAX 0.97xbatched_matmul_ikb_kjb_ijbeager 0.78xtrace 0.79xJAX 3.17xsolveeager 0.79xtrace 0.80xJAX 0.94xgrad_sum_batched_solve_backwardtrace 0.79xJAX 1.45xgrad_sum_svd_s_backwardtrace 0.79xJAX 0.79xgrad_sum_lu_jvptrace 0.79xJAX 1.14xqreager 0.91xtrace 0.80xJAX 1.58xbatched_matmul_ikb_kjb_ijbeager 0.81xtrace 0.80xJAX 2.92xgrad_sum_eigh_jvptrace 0.80xJAX 0.83xgrad_sum_svd_s_backwardtrace 0.80xJAX 1.56xgrad_sum_qr_vjptrace 0.80xJAX 0.79xqreager 0.81xtrace 0.80xJAX 0.92xbatched_matmul_ikb_kjb_ijbeager 0.80xtrace 0.81xJAX 2.85xsolveeager 0.81xtrace 0.83xJAX 1.31xsolveeager 0.81xtrace 0.81xJAX 1.20xsolveeager 0.81xtrace 0.82xJAX 0.97xqreager 0.82xtrace 0.81xJAX 0.82xgrad_sum_matmuleager 1.55xtrace 0.81xJAX 0.76xsolveeager 0.81xtrace 0.87xJAX 1.63xgrad_sum_lu_vjptrace 0.81xJAX 0.77xqreager 0.81xtrace 0.82xJAX 0.86xgrad_sum_batched_matmul_backwardtrace 0.82xJAX 1.00xmatmul_recteager 0.82xtrace 0.82xJAX 1.02xsolveeager 0.82xtrace 0.83xJAX 1.00xgrad_sum_svd_s_backwardtrace 0.82xJAX 0.80xsolveeager 0.82xtrace 0.84xJAX 1.41xbatched_matmul_ikb_kjb_ijbeager 0.82xtrace 0.83xJAX 9.08xsolveeager 0.82xtrace 0.85xJAX 1.17xgrad_sum_matmuleager 1.47xtrace 0.82xJAX 1.88xqreager 0.86xtrace 0.83xJAX 0.94xmatmuleager 0.84xtrace 0.83xJAX 0.92xsolveeager 0.83xtrace 0.83xJAX 1.08xgrad_sum_solve_jvptrace 0.83xJAX 1.46xsvdeager 0.83xtrace 0.87xJAX 1.40xqreager 0.83xtrace 0.83xJAX 0.84xgrad_sum_matmuleager 0.91xtrace 0.83xJAX 1.02xbatched_solveeager 0.84xtrace 0.83xJAX 1.50xqreager 0.88xtrace 0.83xJAX 1.31xbatched_solveeager 1.00xtrace 0.83xJAX 1.83xsolveeager 0.84xtrace 0.86xJAX 1.60xgrad_sum_matmuleager 1.02xtrace 0.84xJAX 1.12xmatmul_recteager 0.84xtrace 0.85xJAX 0.93xgrad_sum_svd_s_backwardtrace 0.84xJAX 1.54xgrad_sum_svd_s_jvptrace 0.84xJAX 1.14xeigheager 0.91xtrace 0.84xJAX 1.52xgrad_sum_svd_s_backwardtrace 0.84xJAX 0.89xbatched_matmul_ikb_kjb_ijbeager 1.03xtrace 0.84xJAX 6.29xsvdeager 0.93xtrace 0.84xJAX 1.52xgrad_sum_qr_jvptrace 0.84xJAX 0.77xgrad_sum_qr_jvptrace 0.84xJAX 0.90xsolveeager 0.84xtrace 0.89xJAX 1.51xgrad_sum_svd_s_backwardtrace 0.85xJAX 1.33xgrad_sum_matmul_backwardtrace 0.85xJAX 0.85xgrad_sum_matmul_backwardtrace 0.85xJAX 0.77xsolveeager 0.85xtrace 0.87xJAX 1.09xgrad_sum_matmuleager 1.48xtrace 0.85xJAX 0.71xgrad_sum_matmuleager 0.90xtrace 0.85xJAX 0.88xsolveeager 0.85xtrace 0.88xJAX 1.19xgrad_sum_batched_matmul_backwardtrace 0.85xJAX 0.25xgrad_sum_batched_matmul_backwardtrace 0.85xJAX 0.21xgrad_sum_svd_s_vjptrace 0.86xJAX 1.40xbatched_solveeager 0.96xtrace 0.86xJAX 1.63xgrad_sum_svd_s_backwardtrace 0.87xJAX 0.97xgrad_sum_matmuleager 1.41xtrace 0.87xJAX 1.86xgrad_sum_lu_vjptrace 0.87xJAX 0.87xbatched_qreager 0.91xtrace 0.87xJAX 1.48xbatched_solveeager 0.94xtrace 0.87xJAX 1.59xgrad_sum_matmul_backwardtrace 0.87xJAX 0.90xbatched_qreager 0.89xtrace 0.87xJAX 1.40xsolveeager 0.87xtrace 7.06xJAX 1.01xeigheager 0.98xtrace 0.87xJAX 1.28xgrad_sum_lu_vjptrace 0.87xJAX 1.57xsvdeager 0.87xtrace 0.88xJAX 0.90xgrad_sum_matmul_backwardtrace 0.87xJAX 0.71xbatched_solveeager 0.89xtrace 0.87xJAX 1.69xgrad_sum_matmul_backwardtrace 0.87xJAX 0.80xbatched_solveeager 0.99xtrace 0.88xJAX 1.57xmatmuleager 0.89xtrace 0.88xJAX 0.83xeigheager 0.98xtrace 0.88xJAX 1.50xgrad_sum_svd_s_vjptrace 0.88xJAX 1.67xsvdeager 0.99xtrace 0.88xJAX 1.51xeigheager 0.94xtrace 0.88xJAX 1.30xsvdeager 0.89xtrace 0.88xJAX 0.96xgrad_sum_qr_vjptrace 0.88xJAX 0.94xsvdeager 0.88xtrace 0.88xJAX 1.24xqreager 1.08xtrace 0.88xJAX 0.87xgrad_sum_matmuleager 1.45xtrace 0.88xJAX 0.89xbatched_qreager 0.90xtrace 0.88xJAX 1.40xsvdeager 0.90xtrace 0.88xJAX 1.03xgrad_sum_batched_matmul_backwardtrace 0.88xJAX 0.22xsolveeager 0.88xtrace 0.91xJAX 1.17xgrad_sum_svd_s_backwardtrace 0.89xJAX 0.89xbatched_solveeager 0.89xtrace 0.92xJAX 1.52xqreager 1.38xtrace 0.89xJAX 1.48xsolveeager 0.89xtrace 0.89xJAX 1.13xsvdeager 0.89xtrace 0.89xJAX 0.93xgrad_sum_svd_s_backwardtrace 0.89xJAX 0.89xgrad_sum_batched_solve_backwardtrace 0.89xJAX 1.74xqreager 0.90xtrace 0.89xJAX 1.38xsvdeager 0.91xtrace 0.89xJAX 1.09xgrad_sum_svd_s_vjptrace 0.89xJAX 0.95xgrad_sum_svd_s_backwardtrace 0.90xJAX 1.00xbatched_qreager 0.92xtrace 0.90xJAX 1.44xqreager 1.03xtrace 0.90xJAX 1.71xbatched_solveeager 1.03xtrace 0.90xJAX 1.58xsvdeager 0.94xtrace 0.90xJAX 1.10xgrad_sum_batched_solve_backwardtrace 0.90xJAX 1.22xmatmuleager 0.90xtrace 0.90xJAX 0.84xeigheager 0.92xtrace 0.90xJAX 1.27xgrad_sum_qr_vjptrace 0.90xJAX 0.87xbatched_solveeager 1.00xtrace 0.90xJAX 1.18xqreager 0.99xtrace 0.90xJAX 1.59xeigheager 0.93xtrace 0.90xJAX 1.33xqreager 1.08xtrace 0.90xJAX 1.63xbatched_qreager 0.96xtrace 0.91xJAX 1.43xgrad_sum_matmul_backwardtrace 0.91xJAX 0.96xsvdeager 0.91xtrace 0.91xJAX 0.93xbatched_qreager 0.91xtrace 0.91xJAX 1.45xqreager 1.02xtrace 0.91xJAX 1.46xeigheager 0.96xtrace 0.91xJAX 1.31xeigheager 0.94xtrace 0.91xJAX 1.20xeigheager 0.99xtrace 0.91xJAX 1.27xbatched_matmul_ikb_kjb_ijbeager 0.94xtrace 0.91xJAX 6.46xbatched_solveeager 0.91xtrace 0.95xJAX 1.63xbatched_matmul_ikb_kjb_ijbeager 0.91xtrace 0.91xJAX 1.23xeigheager 0.92xtrace 0.91xJAX 1.67xmatmuleager 0.96xtrace 0.91xJAX 1.27xsvdeager 0.91xtrace 0.91xJAX 0.97xbatched_svdeager 0.91xtrace 0.93xJAX 0.70xsvdeager 1.39xtrace 0.91xJAX 1.05xsvdeager 0.92xtrace 0.91xJAX 0.96xbatched_svdeager 0.91xtrace 0.92xJAX 1.00xgrad_sum_svd_s_vjptrace 0.91xJAX 1.42xbatched_svdeager 0.91xtrace 0.92xJAX 1.01xgrad_sum_svd_s_backwardtrace 0.92xJAX 0.96xbatched_solveeager 0.98xtrace 0.92xJAX 1.32xbatched_matmul_ikb_kjb_ijbeager 0.92xtrace 0.93xJAX 2.79xqreager 1.05xtrace 0.92xJAX 1.90xgrad_sum_matmul_backwardtrace 0.92xJAX 0.93xsvdeager 0.96xtrace 0.92xJAX 1.33xeigheager 1.01xtrace 0.92xJAX 1.51xgrad_sum_qr_jvptrace 0.92xJAX 0.88xmatmul_recteager 0.92xtrace 0.93xJAX 0.96xbatched_svdeager 0.92xtrace 0.92xJAX 1.04xsvdeager 1.00xtrace 0.92xJAX 1.76xsvdeager 0.95xtrace 0.92xJAX 1.12xbatched_svdeager 0.92xtrace 0.92xJAX 1.00xbatched_matmul_ikb_kjb_ijbeager 0.94xtrace 0.93xJAX 2.38xbatched_qreager 0.94xtrace 0.93xJAX 1.23xbatched_svdeager 0.93xtrace 0.93xJAX 0.26xgrad_sum_svd_s_backwardtrace 0.93xJAX 0.98xsvdeager 0.94xtrace 0.93xJAX 1.08xbatched_qreager 0.94xtrace 0.93xJAX 1.08xbatched_qreager 0.93xtrace 0.93xJAX 1.10xsvdeager 0.93xtrace 0.94xJAX 1.44xgrad_sum_svd_s_vjptrace 0.93xJAX 0.96xbatched_svdeager 0.93xtrace 0.93xJAX 0.27xsvdeager 1.01xtrace 0.93xJAX 1.32xbatched_qreager 0.94xtrace 0.93xJAX 0.55xbatched_svdeager 0.93xtrace 0.93xJAX 0.31xgrad_sum_lu_vjptrace 0.93xJAX 1.45xbatched_eigheager 0.93xtrace 0.94xJAX 1.29xbatched_eigheager 0.93xtrace 0.93xJAX 1.25xsvdeager 0.93xtrace 0.93xJAX 0.96xmatmuleager 0.94xtrace 0.93xJAX 0.95xgrad_sum_solve_vjptrace 0.94xJAX 0.78xeigheager 0.94xtrace 0.94xJAX 1.03xeigheager 0.95xtrace 0.94xJAX 1.65xbatched_svdeager 0.94xtrace 0.94xJAX 0.64xbatched_svdeager 0.94xtrace 0.94xJAX 0.46xbatched_eigheager 0.94xtrace 0.96xJAX 1.06xbatched_eigheager 0.94xtrace 0.94xJAX 1.25xbatched_eigheager 0.94xtrace 0.94xJAX 1.34xbatched_svdeager 0.94xtrace 0.94xJAX 0.46xeigheager 1.06xtrace 0.94xJAX 1.31xeigheager 1.02xtrace 0.94xJAX 1.30xbatched_qreager 0.95xtrace 0.95xJAX 0.52xsvdeager 0.95xtrace 1.39xJAX 1.16xeigheager 0.98xtrace 0.95xJAX 1.37xgrad_sum_lu_vjptrace 0.95xJAX 0.87xbatched_eigheager 0.95xtrace 0.95xJAX 0.80xbatched_eigheager 0.95xtrace 0.95xJAX 0.38xbatched_eigheager 0.95xtrace 0.95xJAX 0.40xsvdeager 0.95xtrace 0.95xJAX 1.06xeigheager 0.97xtrace 0.95xJAX 1.22xqreager 1.04xtrace 0.95xJAX 1.76xeigheager 0.96xtrace 0.95xJAX 1.16xgrad_sum_batched_solve_backwardtrace 0.95xJAX 1.83xgrad_sum_qr_vjptrace 0.96xJAX 1.55xgrad_sum_qr_jvptrace 0.96xJAX 1.67xbatched_eigheager 0.96xtrace 0.96xJAX 0.41xbatched_eigheager 0.96xtrace 0.96xJAX 0.63xsvdeager 1.00xtrace 0.96xJAX 1.11xbatched_qreager 1.02xtrace 0.96xJAX 1.49xgrad_sum_qr_jvptrace 0.96xJAX 1.73xqreager 1.15xtrace 0.96xJAX 1.70xgrad_sum_qr_jvptrace 0.96xJAX 1.52xsvdeager 1.09xtrace 0.96xJAX 1.35xbatched_svdeager 0.97xtrace 0.96xJAX 0.50xbatched_eigheager 0.96xtrace 0.96xJAX 1.00xbatched_qreager 1.02xtrace 0.96xJAX 1.81xsvdeager 1.04xtrace 0.97xJAX 1.08xgrad_sum_batched_solve_backwardtrace 0.97xJAX 1.95xeigheager 0.97xtrace 0.97xJAX 0.94xeigheager 1.06xtrace 0.97xJAX 1.50xbatched_qreager 0.99xtrace 0.97xJAX 0.99xbatched_qreager 1.00xtrace 0.98xJAX 1.63xbatched_qreager 1.01xtrace 0.98xJAX 1.64xgrad_sum_svd_s_vjptrace 0.98xJAX 1.04xbatched_solveeager 1.00xtrace 0.98xJAX 1.77xbatched_qreager 1.00xtrace 0.98xJAX 1.58xqreager 1.23xtrace 0.98xJAX 1.86xbatched_qreager 1.01xtrace 0.98xJAX 1.60xbatched_qreager 0.99xtrace 0.99xJAX 1.57xgrad_sum_eigh_vjptrace 0.99xJAX 1.02xbatched_matmul_ikb_kjb_ijbeager 0.99xtrace 0.99xJAX 3.56xbatched_solveeager 1.09xtrace 0.99xJAX 1.79xgrad_sum_eigh_vjptrace 0.99xJAX 1.03xbatched_matmul_ikb_kjb_ijbeager 1.00xtrace 0.99xJAX 1.46xbatched_eigheager 1.00xtrace 1.00xJAX 1.19xgrad_sum_svd_s_backwardtrace 1.00xJAX 1.37xsvdeager 1.05xtrace 1.00xJAX 1.18xbatched_qreager 1.02xtrace 1.00xJAX 1.64xqreager 1.12xtrace 1.00xJAX 1.76xbatched_eigheager 1.00xtrace 1.01xJAX 1.21xbatched_eigheager 1.01xtrace 1.00xJAX 1.13xbatched_eigheager 1.01xtrace 1.00xJAX 0.69xbatched_eigheager 1.01xtrace 1.01xJAX 1.12xbatched_qreager 1.03xtrace 1.01xJAX 1.58xbatched_eigheager 1.01xtrace 1.01xJAX 1.11xbatched_eigheager 1.02xtrace 1.01xJAX 1.21xgrad_sum_batched_solve_backwardtrace 1.01xJAX 1.38xbatched_solveeager 1.16xtrace 1.01xJAX 1.79xbatched_eigheager 1.01xtrace 1.01xJAX 1.21xbatched_qreager 1.01xtrace 1.02xJAX 1.83xgrad_sum_eigh_vjptrace 1.01xJAX 1.98xgrad_sum_batched_solve_backwardtrace 1.01xJAX 1.69xeigheager 1.02xtrace 1.01xJAX 0.91xgrad_sum_batched_solve_backwardtrace 1.01xJAX 1.43xbatched_eigheager 1.02xtrace 1.02xJAX 0.84xbatched_qreager 1.02xtrace 1.02xJAX 1.70xgrad_sum_svd_s_vjptrace 1.02xJAX 1.26xbatched_eigheager 1.02xtrace 1.02xJAX 1.21xbatched_eigheager 1.03xtrace 1.02xJAX 0.48xbatched_qreager 1.09xtrace 1.03xJAX 1.59xbatched_eigheager 1.03xtrace 1.03xJAX 0.57xbatched_qreager 1.04xtrace 1.03xJAX 1.42xbatched_qreager 1.04xtrace 1.03xJAX 1.45xbatched_qreager 1.04xtrace 1.03xJAX 0.99xbatched_eigheager 1.04xtrace 1.04xJAX 0.53xbatched_qreager 1.04xtrace 1.04xJAX 0.79xbatched_qreager 1.09xtrace 1.05xJAX 1.49xmatmuleager 1.21xtrace 1.05xJAX 2.66xeigheager 1.05xtrace 1.05xJAX 0.86xgrad_sum_batched_solve_backwardtrace 1.05xJAX 1.51xbatched_matmul_ikb_kjb_ijbeager 1.07xtrace 1.05xJAX 2.59xqreager 1.06xtrace 1.06xJAX 0.79xbatched_matmul_ikb_kjb_ijbeager 1.06xtrace 1.06xJAX 2.80xbatched_qreager 1.12xtrace 1.06xJAX 1.56xbatched_qreager 1.07xtrace 1.06xJAX 1.49xgrad_sum_batched_solve_backwardtrace 1.06xJAX 1.67xbatched_solveeager 1.11xtrace 1.06xJAX 1.85xbatched_qreager 1.09xtrace 1.07xJAX 1.51xgrad_sum_eigh_vjptrace 1.07xJAX 1.07xbatched_svdeager 1.07xtrace 1.07xJAX 0.92xbatched_svdeager 1.08xtrace 1.07xJAX 0.88xbatched_svdeager 1.08xtrace 1.09xJAX 0.93xbatched_svdeager 1.08xtrace 1.08xJAX 0.87xbatched_svdeager 1.08xtrace 1.08xJAX 0.88xbatched_qreager 1.15xtrace 1.08xJAX 1.48xgrad_sum_batched_solve_backwardtrace 1.08xJAX 1.79xbatched_solveeager 1.15xtrace 1.08xJAX 2.43xbatched_eigheager 1.10xtrace 1.08xJAX 1.04xqreager 1.18xtrace 1.08xJAX 2.19xbatched_svdeager 1.09xtrace 1.09xJAX 0.82xsvdeager 1.09xtrace 1.09xJAX 1.18xsvdeager 1.09xtrace 1.26xJAX 1.13xbatched_svdeager 1.09xtrace 1.09xJAX 0.55xbatched_svdeager 1.09xtrace 1.09xJAX 0.99xbatched_svdeager 1.09xtrace 1.09xJAX 0.23xbatched_solveeager 1.25xtrace 1.09xJAX 1.85xbatched_svdeager 1.09xtrace 1.09xJAX 0.71xbatched_qreager 1.25xtrace 1.09xJAX 1.59xbatched_svdeager 1.09xtrace 1.09xJAX 0.21xbatched_eigheager 1.09xtrace 1.10xJAX 1.07xgrad_sum_batched_solve_backwardtrace 1.10xJAX 2.29xbatched_eigheager 1.12xtrace 1.10xJAX 1.08xbatched_qreager 1.12xtrace 1.10xJAX 1.65xmatmuleager 1.21xtrace 1.10xJAX 1.52xbatched_qreager 1.11xtrace 1.11xJAX 1.65xbatched_svdeager 1.13xtrace 1.11xJAX 0.98xbatched_qreager 1.12xtrace 1.11xJAX 1.48xbatched_qreager 1.13xtrace 1.12xJAX 1.56xbatched_qreager 1.23xtrace 1.12xJAX 1.50xbatched_qreager 1.12xtrace 1.12xJAX 1.61xbatched_qreager 1.15xtrace 1.12xJAX 1.46xbatched_qreager 1.13xtrace 1.12xJAX 1.36xgrad_sum_eigh_vjptrace 1.12xJAX 1.77xbatched_svdeager 1.13xtrace 1.14xJAX 1.00xbatched_eigheager 1.13xtrace 1.13xJAX 1.03xbatched_svdeager 1.16xtrace 1.13xJAX 1.19xbatched_svdeager 1.13xtrace 1.14xJAX 0.40xbatched_eigheager 1.14xtrace 1.13xJAX 1.07xbatched_svdeager 1.16xtrace 1.14xJAX 1.01xbatched_eigheager 1.14xtrace 1.14xJAX 1.08xbatched_eigheager 1.14xtrace 1.14xJAX 0.98xbatched_eigheager 1.14xtrace 1.15xJAX 0.74xbatched_eigheager 1.15xtrace 1.15xJAX 0.98xgrad_sum_eigh_vjptrace 1.15xJAX 2.06xbatched_svdeager 1.15xtrace 1.15xJAX 0.96xbatched_eigheager 1.15xtrace 1.15xJAX 1.08xbatched_eigheager 1.15xtrace 1.15xJAX 0.79xbatched_svdeager 1.16xtrace 1.15xJAX 0.94xbatched_svdeager 1.16xtrace 1.16xJAX 0.94xbatched_svdeager 1.16xtrace 1.16xJAX 1.02xbatched_svdeager 1.16xtrace 1.18xJAX 1.02xbatched_svdeager 1.17xtrace 1.17xJAX 1.03xbatched_svdeager 1.17xtrace 1.17xJAX 1.00xbatched_eigheager 1.17xtrace 1.17xJAX 1.09xbatched_svdeager 1.17xtrace 1.17xJAX 0.49xbatched_svdeager 1.17xtrace 1.17xJAX 0.40xbatched_svdeager 1.21xtrace 1.18xJAX 1.26xbatched_solveeager 1.34xtrace 1.20xJAX 2.30xbatched_qreager 1.27xtrace 1.20xJAX 1.63xbatched_svdeager 1.23xtrace 1.20xJAX 1.25xgrad_sum_batched_solve_backwardtrace 1.20xJAX 2.57xbatched_eigheager 1.29xtrace 1.21xJAX 1.08xbatched_qreager 1.26xtrace 1.22xJAX 1.51xbatched_eigheager 1.28xtrace 1.22xJAX 1.07xbatched_svdeager 1.24xtrace 1.23xJAX 1.13xbatched_eigheager 1.27xtrace 1.23xJAX 1.10xgrad_sum_batched_solve_backwardtrace 1.24xJAX 2.24xbatched_solveeager 1.59xtrace 1.24xJAX 2.53xgrad_sum_svd_s_backwardtrace 1.24xJAX 1.20xbatched_qreager 1.26xtrace 1.28xJAX 1.56xbatched_svdeager 1.27xtrace 1.27xJAX 1.15xbatched_matmul_ikb_kjb_ijbeager 1.28xtrace 1.27xJAX 3.23xbatched_svdeager 1.28xtrace 1.27xJAX 1.21xbatched_svdeager 1.28xtrace 1.29xJAX 1.19xbatched_svdeager 1.28xtrace 1.29xJAX 1.11xeigheager 1.29xtrace 1.28xJAX 6.53xbatched_qreager 1.31xtrace 1.29xJAX 1.55xgrad_sum_solve_vjptrace 1.30xJAX 0.92xqreager 1.30xtrace 1.31xJAX 1.41xeigheager 3.09xtrace 1.31xJAX 2.23xbatched_qreager 1.32xtrace 1.31xJAX 1.57xbatched_svdeager 1.31xtrace 1.31xJAX 1.25xbatched_qreager 1.32xtrace 1.35xJAX 1.68xbatched_svdeager 1.32xtrace 1.32xJAX 1.26xbatched_matmul_ikb_kjb_ijbeager 1.35xtrace 1.33xJAX 2.95xgrad_sum_batched_solve_backwardtrace 1.33xJAX 3.06xgrad_sum_solve_vjptrace 1.33xJAX 1.13xgrad_sum_matmuleager 1.33xtrace 3.95xJAX 1.70xbatched_svdeager 1.34xtrace 1.34xJAX 1.36xgrad_sum_matmuleager 2.31xtrace 1.36xJAX 1.35xbatched_matmul_ikb_kjb_ijbeager 1.62xtrace 1.36xJAX 3.61xbatched_solveeager 1.46xtrace 1.37xJAX 2.73xbatched_eigheager 1.41xtrace 1.37xJAX 0.86xgrad_sum_batched_solve_backwardtrace 1.37xJAX 3.21xbatched_svdeager 1.38xtrace 1.39xJAX 1.39xbatched_solveeager 1.68xtrace 1.38xJAX 2.76xbatched_solveeager 1.59xtrace 1.40xJAX 2.38xmatmuleager 1.56xtrace 1.40xJAX 2.58xgrad_sum_svd_s_backwardtrace 1.40xJAX 1.35xgrad_sum_lu_jvptrace 1.41xJAX 1.52xbatched_eigheager 1.41xtrace 1.41xJAX 0.89xqreager 1.50xtrace 1.42xJAX 1.74xbatched_eigheager 1.43xtrace 1.42xJAX 1.12xbatched_solveeager 1.59xtrace 1.43xJAX 2.65xmatmuleager 1.69xtrace 1.44xJAX 2.70xbatched_solveeager 1.54xtrace 1.45xJAX 3.77xgrad_sum_svd_s_jvptrace 1.45xJAX 1.51xmatmuleager 1.67xtrace 1.45xJAX 5.46xgrad_sum_batched_matmul_backwardtrace 1.46xJAX 2.60xbatched_solveeager 1.74xtrace 1.46xJAX 3.06xgrad_sum_batched_solve_backwardtrace 1.46xJAX 2.83xgrad_sum_batched_solve_backwardtrace 1.46xJAX 3.40xbatched_solveeager 1.82xtrace 1.48xJAX 2.61xbatched_eigheager 1.50xtrace 1.49xJAX 0.87xgrad_sum_svd_s_vjptrace 1.49xJAX 1.64xgrad_sum_batched_solve_backwardtrace 1.49xJAX 2.44xbatched_eigheager 1.50xtrace 1.50xJAX 0.89xgrad_sum_solve_vjptrace 1.50xJAX 0.88xgrad_sum_batched_solve_backwardtrace 1.51xJAX 2.58xgrad_sum_solve_jvptrace 1.51xJAX 1.62xbatched_solveeager 2.02xtrace 1.51xJAX 3.09xbatched_eigheager 1.52xtrace 1.53xJAX 0.88xbatched_eigheager 1.53xtrace 1.53xJAX 0.90xgrad_sum_lu_vjptrace 1.53xJAX 1.67xgrad_sum_solve_jvptrace 1.53xJAX 1.01xbatched_eigheager 1.54xtrace 1.54xJAX 0.90xgrad_sum_qr_vjptrace 1.54xJAX 2.02xbatched_matmul_ikb_kjb_ijbeager 1.57xtrace 1.60xJAX 6.70xqreager 1.57xtrace 2.09xJAX 1.74xbatched_matmul_ikb_kjb_ijbeager 1.57xtrace 1.60xJAX 7.11xgrad_sum_solve_jvptrace 1.58xJAX 1.60xbatched_eigheager 1.59xtrace 1.59xJAX 0.85xbatched_solveeager 1.76xtrace 1.59xJAX 2.61xgrad_sum_batched_matmul_backwardtrace 1.59xJAX 2.88xbatched_matmul_ikb_kjb_ijbeager 1.66xtrace 1.60xJAX 3.97xgrad_sum_svd_s_jvptrace 1.60xJAX 1.34xgrad_sum_lu_vjptrace 1.60xJAX 1.89xgrad_sum_batched_solve_backwardtrace 1.62xJAX 2.98xbatched_matmul_ikb_kjb_ijbeager 1.63xtrace 1.66xJAX 6.79xgrad_sum_eigh_jvptrace 1.64xJAX 1.48xqreager 1.65xtrace 3.13xJAX 1.62xbatched_matmul_ikb_kjb_ijbeager 1.68xtrace 1.66xJAX 3.30xgrad_sum_svd_s_backwardtrace 1.67xJAX 1.15xmatmuleager 1.71xtrace 1.76xJAX 4.21xbatched_solveeager 2.11xtrace 1.72xJAX 3.83xgrad_sum_batched_solve_backwardtrace 1.74xJAX 3.72xgrad_sum_matmul_backwardtrace 1.76xJAX 0.94xmatmuleager 1.82xtrace 1.77xJAX 3.61xmatmuleager 2.12xtrace 1.77xJAX 6.68xbatched_solveeager 1.98xtrace 1.78xJAX 4.18xgrad_sum_lu_jvptrace 1.79xJAX 1.06xgrad_sum_svd_s_vjptrace 1.79xJAX 1.39xgrad_sum_batched_solve_backwardtrace 1.82xJAX 3.54xbatched_solveeager 2.05xtrace 1.83xJAX 3.84xbatched_solveeager 1.97xtrace 1.83xJAX 3.17xbatched_solveeager 2.20xtrace 1.88xJAX 3.11xmatmuleager 2.10xtrace 1.90xJAX 4.09xgrad_sum_solve_jvptrace 1.92xJAX 0.38xbatched_solveeager 2.09xtrace 1.94xJAX 4.22xgrad_sum_batched_matmul_backwardtrace 1.95xJAX 0.18xmatmuleager 1.97xtrace 2.02xJAX 3.87xgrad_sum_batched_matmul_backwardtrace 1.98xJAX 0.28xmatmuleager 1.99xtrace 2.14xJAX 3.88xmatmuleager 2.21xtrace 1.99xJAX 4.54xbatched_matmul_ikb_kjb_ijbeager 2.28xtrace 2.00xJAX 3.79xmatmuleager 2.07xtrace 2.01xJAX 4.37xgrad_sum_matmul_backwardtrace 2.02xJAX 1.08xgrad_sum_batched_matmul_backwardtrace 2.03xJAX 0.24xbatched_solveeager 2.40xtrace 2.03xJAX 4.36xbatched_solveeager 2.21xtrace 2.04xJAX 3.36xqreager 2.10xtrace 2.05xJAX 1.90xmatmuleager 2.14xtrace 2.06xJAX 4.28xbatched_solveeager 2.76xtrace 2.15xJAX 3.79xgrad_sum_qr_jvptrace 2.19xJAX 2.30xgrad_sum_eigh_jvptrace 2.21xJAX 1.54xsvdeager 2.24xtrace 2.27xJAX 1.60xgrad_sum_qr_vjptrace 2.25xJAX 1.76xbatched_matmul_ikb_kjb_ijbeager 2.27xtrace 2.26xJAX 5.13xgrad_sum_lu_vjptrace 2.33xJAX 1.70xmatmuleager 3.76xtrace 2.40xJAX 1.59xbatched_matmul_ikb_kjb_ijbeager 2.50xtrace 2.58xJAX 5.87xbatched_solveeager 2.87xtrace 2.53xJAX 4.49xgrad_sum_eigh_vjptrace 2.53xJAX 2.04xbatched_matmul_ikb_kjb_ijbeager 2.82xtrace 2.85xJAX 3.52xbatched_matmul_ikb_kjb_ijbeager 2.89xtrace 2.89xJAX 3.87xgrad_sum_eigh_jvptrace 2.93xJAX 2.44xbatched_matmul_ikb_kjb_ijbeager 3.03xtrace 3.03xJAX 3.42xgrad_sum_solve_vjptrace 3.19xJAX 0.64xgrad_sum_matmul_backwardtrace 3.48xJAX 0.99xgrad_sum_matmuleager 4.07xtrace 3.68xJAX 1.41xgrad_sum_lu_vjptrace 3.96xJAX 2.76xgrad_sum_qr_jvptrace 4.33xJAX 1.99xgrad_sum_qr_vjptrace 4.64xJAX 1.90xgrad_sum_eigh_jvptrace 4.76xJAX 1.83xgrad_sum_solve_jvptrace 5.29xJAX 1.59xgrad_sum_eigh_vjptrace 5.33xJAX 2.92xgrad_sum_qr_vjptrace 5.51xJAX 3.02xgrad_sum_qr_jvptrace 5.53xJAX 3.14xgrad_sum_eigh_jvptrace 7.90xJAX 2.49xgrad_sum_eigh_vjptrace 9.02xJAX 2.83x
cpu/einsum 144 cases

baseline PyTorch · best first · 144 cases · case table, absolute times and raw report →

bar = case median / PyTorch median · dashed line = baseline (1.00x) · log scale · shorter (left) is faster · whisker = estimated 95% CIbin_elementwise_mul_2048x2048eager 0.24xtrace 0.25xJAX 1.45xbin_elementwise_mul_2048x2048eager 0.24xtrace 0.25xJAX 1.45xbin_omeinsum_batched_matmul_8x8_batch_4eager 0.25xtrace 0.42xJAX 5.33xbin_omeinsum_batched_matmul_8x8_batch_4eager 0.25xtrace 0.42xJAX 5.33xbin_omeinsum_batched_matmul_8x8_batch_4eager 0.27xtrace 0.45xJAX 5.82xbin_omeinsum_batched_matmul_8x8_batch_4eager 0.27xtrace 0.45xJAX 5.82xbin_omeinsum_batched_matmul_8x8_batch_4eager 0.28xtrace 0.44xJAX 5.72xbin_omeinsum_batched_matmul_8x8_batch_4eager 0.28xtrace 0.44xJAX 5.72xbin_omeinsum_high_d_12x12_contract_4_batch_4eager 0.30xtrace 0.33xJAX 1.61xbin_omeinsum_high_d_12x12_contract_4_batch_4eager 0.30xtrace 0.33xJAX 1.61xbin_omeinsum_high_d_12x12_contract_4_batch_4eager 0.31xtrace 0.33xJAX 1.67xbin_omeinsum_high_d_12x12_contract_4_batch_4eager 0.31xtrace 0.33xJAX 1.67xbin_omeinsum_high_d_12x12_contract_4_batch_4eager 0.31xtrace 0.33xJAX 1.91xbin_omeinsum_high_d_12x12_contract_4_batch_4eager 0.31xtrace 0.33xJAX 1.91xbin_matmul_256eager 0.40xtrace 6.22xJAX 0.78xbin_matmul_256eager 0.40xtrace 6.22xJAX 0.78xbin_matmul_256eager 0.40xtrace 2.56xJAX 0.65xbin_matmul_256eager 0.40xtrace 2.56xJAX 0.65xbin_omeinsum_matmul_10x10eager 0.45xtrace 0.91xJAX 4.09xbin_omeinsum_matmul_10x10eager 0.45xtrace 0.91xJAX 4.09xbin_elementwise_mul_2048x2048eager 0.47xtrace 0.61xJAX 1.65xbin_elementwise_mul_2048x2048eager 0.47xtrace 0.61xJAX 1.65xbin_omeinsum_matmul_10x10eager 0.53xtrace 2.71xJAX 3.71xbin_omeinsum_matmul_10x10eager 0.53xtrace 2.71xJAX 3.71xbin_omeinsum_matmul_10x10eager 0.55xtrace 0.91xJAX 3.55xbin_omeinsum_matmul_10x10eager 0.55xtrace 0.91xJAX 3.55xtensornetwork_permutation_focus_step409_316eager 0.73xtrace 0.56xJAX 0.89xtensornetwork_permutation_focus_step409_316eager 0.73xtrace 0.56xJAX 0.89xtensornetwork_permutation_light_415eager 0.59xtrace 0.57xJAX 0.94xtensornetwork_permutation_light_415eager 0.59xtrace 0.57xJAX 0.94xnary_matmul_chain_64eager 0.64xtrace 2.52xJAX 2.04xnary_matmul_chain_64eager 0.64xtrace 2.52xJAX 2.04xbin_matmul_256eager 0.66xtrace 0.82xJAX 0.76xbin_matmul_256eager 0.66xtrace 0.82xJAX 0.76xlm_batch_likelihood_sentence_3_12deager 1.31xtrace 0.70xJAX 0.90xbin_elementwise_mul_2048x2048eager 0.75xtrace 0.73xJAX 1.84xbin_elementwise_mul_2048x2048eager 0.75xtrace 0.73xJAX 1.84xnary_matmul_chain_64eager 0.74xtrace 0.90xJAX 2.35xnary_matmul_chain_64eager 0.74xtrace 0.90xJAX 2.35xlm_batch_likelihood_sentence_4_4deager 1.20xtrace 0.75xJAX 0.84xlm_batch_likelihood_brackets_4_4deager 1.12xtrace 0.75xJAX 0.82xlm_batch_likelihood_sentence_4_4deager 1.31xtrace 0.75xJAX 0.77xlm_batch_likelihood_brackets_4_4deager 1.13xtrace 0.77xJAX 0.88xbin_matmul_1024eager 0.86xtrace 1.04xJAX 0.88xbin_matmul_1024eager 0.86xtrace 1.04xJAX 0.88xstr_mps_varying_inner_product_200eager 1.02xtrace 0.86xJAX 1.43xstr_matrix_chain_multiplication_100eager 0.93xtrace 0.89xJAX 1.45xbin_permuted_r4_abcd_dbef_acef_d32eager 0.89xtrace 1.11xJAX 1.00xbin_permuted_r4_abcd_dbef_acef_d32eager 0.89xtrace 1.11xJAX 1.00xstr_nw_mera_closed_120eager 0.93xtrace 0.90xJAX 1.06xbin_batched_matmul_b32_m128_n128_k128eager 0.90xtrace 1.69xJAX 0.90xbin_batched_matmul_b32_m128_n128_k128eager 0.90xtrace 1.69xJAX 0.90xstr_matrix_chain_multiplication_100eager 0.95xtrace 0.91xJAX 1.17xbin_batched_matmul_b32_m64_n64_k64eager 0.92xtrace 1.15xJAX 1.02xbin_batched_matmul_b32_m64_n64_k64eager 0.92xtrace 1.15xJAX 1.02xstr_matrix_chain_multiplication_100eager 0.95xtrace 0.93xJAX 1.17xstr_matrix_chain_multiplication_100eager 0.94xtrace 1.81xJAX 1.47xbin_elementwise_mul_512x512eager 0.96xtrace 0.98xJAX 2.38xbin_elementwise_mul_512x512eager 0.96xtrace 0.98xJAX 2.38xstr_nw_mera_closed_120eager 1.00xtrace 1.02xJAX 1.06xgm_queen5_5_3.wcspeager 1.08xtrace 1.01xJAX 0.73xlm_batch_likelihood_sentence_3_12deager 4.05xtrace 1.02xJAX 0.93xbin_outer_product_4096eager 1.03xtrace 1.03xJAX 1.03xbin_outer_product_4096eager 1.03xtrace 1.03xJAX 1.03xstr_mps_varying_inner_product_200eager 1.05xtrace 1.04xJAX 1.48xgm_queen5_5_3.wcspeager 1.07xtrace 1.28xJAX 0.78xlm_batch_likelihood_sentence_3_12deager 1.97xtrace 1.07xJAX 1.54xstr_nw_mera_closed_120eager 1.14xtrace 1.08xJAX 0.92xbin_matmul_1024eager 1.10xtrace 1.21xJAX 0.60xbin_matmul_1024eager 1.10xtrace 1.21xJAX 0.60xstr_mps_varying_inner_product_200eager 1.11xtrace 1.56xJAX 1.27xbin_matmul_1024eager 1.11xtrace 1.46xJAX 0.70xbin_matmul_1024eager 1.11xtrace 1.46xJAX 0.70xstr_nw_mera_open_26eager 1.19xtrace 1.12xJAX 1.39xstr_nw_mera_open_26eager 1.24xtrace 1.16xJAX 1.41xstr_nw_mera_closed_120eager 1.27xtrace 1.17xJAX 0.92xlm_batch_likelihood_sentence_3_12deager 4.51xtrace 1.18xJAX 0.81xstr_mps_varying_inner_product_200eager 1.20xtrace 1.25xJAX 1.21xbin_outer_product_4096eager 1.21xtrace 1.26xJAX 1.68xbin_outer_product_4096eager 1.21xtrace 1.26xJAX 1.68xstr_nw_mera_closed_120eager 1.26xtrace 1.31xJAX 0.95xstr_nw_mera_closed_120eager 1.37xtrace 1.46xJAX 0.89xstr_mps_varying_inner_product_200eager 1.37xtrace 1.58xJAX 1.46xbin_permuted_r4_abcd_dbef_acef_d32eager 1.38xtrace 2.08xJAX 0.90xbin_permuted_r4_abcd_dbef_acef_d32eager 1.38xtrace 2.08xJAX 0.90xbin_permuted_r4_abcd_dbef_acef_d32eager 1.43xtrace 1.38xJAX 0.96xbin_permuted_r4_abcd_dbef_acef_d32eager 1.43xtrace 1.38xJAX 0.96xtensornetwork_permutation_light_415eager 1.44xtrace 1.43xJAX 1.58xtensornetwork_permutation_light_415eager 1.44xtrace 1.43xJAX 1.58xtensornetwork_permutation_focus_step409_316eager 1.62xtrace 1.44xJAX 1.50xtensornetwork_permutation_focus_step409_316eager 1.62xtrace 1.44xJAX 1.50xbin_outer_product_4096eager 1.52xtrace 1.48xJAX 1.72xbin_outer_product_4096eager 1.52xtrace 1.48xJAX 1.72xstr_mps_varying_inner_product_200eager 1.80xtrace 1.57xJAX 1.44xlm_batch_likelihood_sentence_3_12deager 9.60xtrace 1.60xJAX 1.71xlm_batch_likelihood_sentence_3_12deager 3.60xtrace 1.62xJAX 1.18xbin_batched_outer_product_compact_j16_k16_o64_t64eager 1.62xtrace 1.81xJAX 2.74xbin_batched_outer_product_compact_j16_k16_o64_t64eager 1.62xtrace 1.81xJAX 2.74xbin_batched_outer_product_noncompact_j16_k16_o64_t64eager 1.62xtrace 6.55xJAX 2.65xbin_batched_outer_product_noncompact_j16_k16_o64_t64eager 1.62xtrace 6.55xJAX 2.65xgm_queen5_5_3.wcspeager 1.64xtrace 1.94xJAX 1.11xstr_nw_mera_open_26eager 1.87xtrace 1.68xJAX 1.38xtensornetwork_permutation_light_415eager 1.82xtrace 1.73xJAX 2.27xtensornetwork_permutation_light_415eager 1.82xtrace 1.73xJAX 2.27xgm_queen5_5_3.wcspeager 1.73xtrace 2.65xJAX 1.26xstr_nw_mera_open_26eager 1.98xtrace 1.76xJAX 1.41xgm_queen5_5_3.wcspeager 1.78xtrace 3.22xJAX 1.26xtensornetwork_permutation_focus_step409_316eager 2.14xtrace 1.82xJAX 2.16xtensornetwork_permutation_focus_step409_316eager 2.14xtrace 1.82xJAX 2.16xgm_queen5_5_3.wcspeager 1.84xtrace 2.40xJAX 1.13xstr_nw_mera_open_26eager 1.92xtrace 2.03xJAX 1.29xstr_nw_mera_open_26eager 1.95xtrace 2.02xJAX 1.25xstr_matrix_chain_multiplication_100eager 2.10xtrace 1.99xJAX 1.38xlm_batch_likelihood_sentence_4_4deager 4.63xtrace 2.08xJAX 1.63xstr_matrix_chain_multiplication_100eager 2.10xtrace 2.10xJAX 1.37xnary_matmul_chain_64eager 2.48xtrace 3.00xJAX 2.74xnary_matmul_chain_64eager 2.48xtrace 3.00xJAX 2.74xlm_batch_likelihood_brackets_4_4deager 3.54xtrace 2.75xJAX 1.87xlm_batch_likelihood_sentence_4_4deager 5.87xtrace 3.24xJAX 1.99xbin_batched_matmul_b32_m128_n128_k128eager 3.29xtrace 7.08xJAX 1.86xbin_batched_matmul_b32_m128_n128_k128eager 3.29xtrace 7.08xJAX 1.86xlm_batch_likelihood_brackets_4_4deager 6.27xtrace 3.37xJAX 2.13xlm_batch_likelihood_brackets_4_4deager 4.65xtrace 3.51xJAX 1.92xlm_batch_likelihood_sentence_4_4deager 6.19xtrace 3.67xJAX 1.78xbin_batched_matmul_b32_m128_n128_k128eager 3.76xtrace 6.07xJAX 1.93xbin_batched_matmul_b32_m128_n128_k128eager 3.76xtrace 6.07xJAX 1.93xlm_batch_likelihood_sentence_4_4deager 6.25xtrace 4.46xJAX 1.96xlm_batch_likelihood_brackets_4_4deager 7.24xtrace 4.53xJAX 2.20xbin_batched_matmul_b32_m64_n64_k64eager 8.03xtrace 17.32xJAX 4.57xbin_batched_matmul_b32_m64_n64_k64eager 8.03xtrace 17.32xJAX 4.57xbin_batched_outer_product_noncompact_j16_k16_o64_t64eager 20.07xtrace 11.81xJAX 7.81xbin_batched_outer_product_noncompact_j16_k16_o64_t64eager 20.07xtrace 11.81xJAX 7.81xbin_batched_outer_product_noncompact_j16_k16_o64_t64eager 13.50xtrace 27.97xJAX 15.80xbin_batched_outer_product_noncompact_j16_k16_o64_t64eager 13.50xtrace 27.97xJAX 15.80xbin_batched_matmul_b32_m64_n64_k64eager 13.65xtrace 23.71xJAX 9.54xbin_batched_matmul_b32_m64_n64_k64eager 13.65xtrace 23.71xJAX 9.54xbin_batched_outer_product_compact_j16_k16_o64_t64eager 19.03xtrace 25.55xJAX 5.48xbin_batched_outer_product_compact_j16_k16_o64_t64eager 19.03xtrace 25.55xJAX 5.48xbin_elementwise_mul_512x512eager 24.47xtrace 36.95xJAX 11.68xbin_elementwise_mul_512x512eager 24.47xtrace 36.95xJAX 11.68xbin_batched_outer_product_compact_j16_k16_o64_t64eager 26.37xtrace 39.67xJAX 14.60xbin_batched_outer_product_compact_j16_k16_o64_t64eager 26.37xtrace 39.67xJAX 14.60xbin_elementwise_mul_512x512eager 51.67xtrace 116.57xJAX 12.00xbin_elementwise_mul_512x512eager 51.67xtrace 116.57xJAX 12.00x
cpu/linalg_jvp_vjp 240 cases

baseline PyTorch · best first · 240 cases · case table, absolute times and raw report →

bar = case median / PyTorch median · dashed line = baseline (1.00x) · log scale · shorter (left) is faster · whisker = estimated 95% CIgrad_sum_solve_jvptrace 0.06xJAX 0.04xgrad_sum_solve_jvptrace 0.06xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.05xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.07xJAX 0.04xgrad_sum_solve_jvptrace 0.08xJAX 0.05xgrad_sum_solve_jvptrace 0.08xJAX 0.05xgrad_sum_solve_jvptrace 0.09xJAX 0.09xgrad_sum_svd_s_jvptrace 0.10xJAX 0.07xgrad_sum_solve_jvptrace 0.10xJAX 0.05xgrad_sum_solve_jvptrace 0.10xJAX 0.05xgrad_sum_svd_s_jvptrace 0.11xJAX 0.07xgrad_sum_svd_s_jvptrace 0.11xJAX 0.07xgrad_sum_svd_s_jvptrace 0.12xJAX 0.09xgrad_sum_svd_s_jvptrace 0.13xJAX 0.09xgrad_sum_svd_s_jvptrace 0.13xJAX 0.09xgrad_sum_solve_vjptrace 0.14xJAX 0.10xgrad_sum_solve_vjptrace 0.15xJAX 0.11xgrad_sum_solve_vjptrace 0.15xJAX 0.11xgrad_sum_solve_vjptrace 0.15xJAX 0.13xgrad_sum_svd_s_vjptrace 0.15xJAX 0.12xgrad_sum_lu_jvptrace 0.16xJAX 0.07xgrad_sum_lu_jvptrace 0.16xJAX 0.07xgrad_sum_solve_vjptrace 0.16xJAX 0.11xgrad_sum_solve_vjptrace 0.16xJAX 0.10xgrad_sum_solve_vjptrace 0.16xJAX 0.11xgrad_sum_svd_s_vjptrace 0.16xJAX 0.12xgrad_sum_lu_jvptrace 0.17xJAX 0.10xgrad_sum_solve_vjptrace 0.17xJAX 0.11xgrad_sum_svd_s_vjptrace 0.17xJAX 0.13xgrad_sum_solve_vjptrace 0.17xJAX 0.11xgrad_sum_solve_vjptrace 0.17xJAX 0.12xgrad_sum_solve_vjptrace 0.17xJAX 0.14xgrad_sum_lu_jvptrace 0.17xJAX 0.07xgrad_sum_lu_jvptrace 0.18xJAX 0.07xgrad_sum_solve_vjptrace 0.18xJAX 0.14xgrad_sum_lu_jvptrace 0.18xJAX 0.07xgrad_sum_lu_jvptrace 0.18xJAX 0.10xgrad_sum_svd_s_jvptrace 0.18xJAX 0.16xgrad_sum_lu_jvptrace 0.18xJAX 0.12xgrad_sum_svd_s_vjptrace 0.18xJAX 0.15xgrad_sum_lu_jvptrace 0.18xJAX 0.07xgrad_sum_lu_jvptrace 0.18xJAX 0.11xgrad_sum_svd_s_jvptrace 0.19xJAX 0.11xgrad_sum_lu_vjptrace 0.19xJAX 0.09xgrad_sum_svd_s_jvptrace 0.19xJAX 0.11xgrad_sum_lu_vjptrace 0.19xJAX 0.09xgrad_sum_svd_s_vjptrace 0.19xJAX 0.15xgrad_sum_lu_vjptrace 0.19xJAX 0.13xgrad_sum_lu_jvptrace 0.19xJAX 0.09xgrad_sum_lu_jvptrace 0.19xJAX 0.09xgrad_sum_solve_vjptrace 0.20xJAX 0.21xgrad_sum_svd_s_vjptrace 0.20xJAX 0.16xgrad_sum_lu_vjptrace 0.20xJAX 0.09xgrad_sum_lu_vjptrace 0.21xJAX 0.08xgrad_sum_lu_vjptrace 0.21xJAX 0.12xgrad_sum_lu_vjptrace 0.21xJAX 0.09xgrad_sum_eigh_jvptrace 0.21xJAX 0.10xgrad_sum_lu_vjptrace 0.21xJAX 0.09xgrad_sum_solve_vjptrace 0.22xJAX 0.13xgrad_sum_lu_vjptrace 0.22xJAX 0.13xgrad_sum_solve_vjptrace 0.22xJAX 0.12xgrad_sum_lu_vjptrace 0.22xJAX 0.16xgrad_sum_eigh_jvptrace 0.23xJAX 0.11xgrad_sum_lu_vjptrace 0.23xJAX 0.11xgrad_sum_eigh_jvptrace 0.24xJAX 0.10xgrad_sum_eigh_vjptrace 0.24xJAX 0.12xgrad_sum_lu_vjptrace 0.24xJAX 0.11xgrad_sum_lu_jvptrace 0.24xJAX 0.29xgrad_sum_eigh_jvptrace 0.25xJAX 0.10xgrad_sum_eigh_jvptrace 0.25xJAX 0.18xgrad_sum_eigh_jvptrace 0.25xJAX 0.12xgrad_sum_lu_jvptrace 0.25xJAX 0.19xgrad_sum_eigh_vjptrace 0.26xJAX 0.14xgrad_sum_eigh_jvptrace 0.26xJAX 0.12xgrad_sum_svd_s_vjptrace 0.27xJAX 0.26xgrad_sum_svd_s_vjptrace 0.27xJAX 0.16xgrad_sum_eigh_vjptrace 0.27xJAX 0.11xgrad_sum_eigh_jvptrace 0.28xJAX 0.12xgrad_sum_svd_s_jvptrace 0.28xJAX 0.30xgrad_sum_lu_jvptrace 0.28xJAX 0.19xgrad_sum_eigh_vjptrace 0.28xJAX 0.12xgrad_sum_eigh_vjptrace 0.28xJAX 0.14xgrad_sum_svd_s_vjptrace 0.28xJAX 0.17xgrad_sum_svd_s_jvptrace 0.28xJAX 0.26xgrad_sum_svd_s_jvptrace 0.28xJAX 0.26xgrad_sum_eigh_vjptrace 0.29xJAX 0.24xgrad_sum_eigh_jvptrace 0.29xJAX 0.12xgrad_sum_eigh_vjptrace 0.30xJAX 0.15xgrad_sum_qr_vjptrace 0.31xJAX 0.09xgrad_sum_eigh_vjptrace 0.31xJAX 0.15xgrad_sum_qr_vjptrace 0.32xJAX 0.10xgrad_sum_qr_vjptrace 0.32xJAX 0.14xgrad_sum_eigh_vjptrace 0.33xJAX 0.15xgrad_sum_qr_jvptrace 0.33xJAX 0.10xgrad_sum_qr_vjptrace 0.33xJAX 0.08xgrad_sum_eigh_jvptrace 0.34xJAX 0.34xgrad_sum_qr_vjptrace 0.34xJAX 0.10xgrad_sum_qr_jvptrace 0.34xJAX 0.15xgrad_sum_lu_vjptrace 0.34xJAX 0.24xgrad_sum_qr_jvptrace 0.34xJAX 0.10xgrad_sum_qr_vjptrace 0.34xJAX 0.11xgrad_sum_qr_vjptrace 0.34xJAX 0.19xgrad_sum_qr_vjptrace 0.35xJAX 0.09xgrad_sum_eigh_jvptrace 0.35xJAX 0.27xgrad_sum_qr_vjptrace 0.35xJAX 0.10xgrad_sum_qr_vjptrace 0.36xJAX 0.12xgrad_sum_qr_jvptrace 0.36xJAX 0.09xgrad_sum_qr_jvptrace 0.36xJAX 0.20xgrad_sum_qr_jvptrace 0.36xJAX 0.14xgrad_sum_qr_vjptrace 0.36xJAX 0.11xgrad_sum_qr_jvptrace 0.36xJAX 0.09xgrad_sum_eigh_jvptrace 0.36xJAX 0.27xgrad_sum_qr_jvptrace 0.37xJAX 0.11xgrad_sum_qr_jvptrace 0.37xJAX 0.11xgrad_sum_qr_jvptrace 0.37xJAX 0.14xgrad_sum_qr_jvptrace 0.37xJAX 0.12xgrad_sum_qr_vjptrace 0.38xJAX 0.11xgrad_sum_eigh_vjptrace 0.38xJAX 0.40xgrad_sum_qr_jvptrace 0.39xJAX 0.12xgrad_sum_solve_vjptrace 0.39xJAX 0.42xgrad_sum_svd_s_vjptrace 0.39xJAX 0.44xgrad_sum_solve_vjptrace 0.40xJAX 0.48xgrad_sum_svd_s_vjptrace 0.40xJAX 0.38xgrad_sum_eigh_vjptrace 0.41xJAX 0.32xgrad_sum_svd_s_vjptrace 0.41xJAX 0.39xgrad_sum_eigh_vjptrace 0.42xJAX 0.32xgrad_sum_lu_jvptrace 0.42xJAX 0.40xgrad_sum_lu_jvptrace 0.43xJAX 0.74xgrad_sum_qr_vjptrace 0.43xJAX 0.32xgrad_sum_lu_vjptrace 0.44xJAX 0.30xgrad_sum_solve_vjptrace 0.45xJAX 0.47xgrad_sum_lu_vjptrace 0.45xJAX 0.25xgrad_sum_qr_jvptrace 0.46xJAX 0.32xgrad_sum_lu_jvptrace 0.46xJAX 0.46xgrad_sum_svd_s_jvptrace 0.46xJAX 0.48xgrad_sum_qr_vjptrace 0.46xJAX 0.31xgrad_sum_qr_jvptrace 0.48xJAX 1.92xgrad_sum_qr_vjptrace 0.49xJAX 0.30xgrad_sum_svd_s_jvptrace 0.49xJAX 0.45xgrad_sum_svd_s_jvptrace 0.49xJAX 0.45xgrad_sum_lu_jvptrace 0.50xJAX 0.48xgrad_sum_qr_jvptrace 0.51xJAX 0.59xgrad_sum_eigh_jvptrace 0.51xJAX 0.57xgrad_sum_lu_jvptrace 0.53xJAX 0.80xgrad_sum_solve_vjptrace 0.54xJAX 0.99xgrad_sum_eigh_jvptrace 0.55xJAX 0.52xgrad_sum_eigh_jvptrace 0.55xJAX 0.54xgrad_sum_eigh_vjptrace 0.58xJAX 0.66xgrad_sum_svd_s_vjptrace 0.60xJAX 0.63xgrad_sum_eigh_vjptrace 0.63xJAX 0.59xgrad_sum_eigh_vjptrace 0.63xJAX 0.60xgrad_sum_svd_s_vjptrace 0.63xJAX 0.58xgrad_sum_svd_s_vjptrace 0.64xJAX 0.59xgrad_sum_solve_jvptrace 0.65xJAX 0.68xgrad_sum_solve_jvptrace 0.66xJAX 0.79xgrad_sum_solve_jvptrace 0.67xJAX 0.78xgrad_sum_lu_jvptrace 0.69xJAX 0.74xgrad_sum_svd_s_jvptrace 0.72xJAX 0.75xgrad_sum_svd_s_jvptrace 0.72xJAX 1.12xgrad_sum_svd_s_jvptrace 0.72xJAX 1.35xgrad_sum_eigh_jvptrace 0.72xJAX 0.74xgrad_sum_svd_s_jvptrace 0.72xJAX 0.75xgrad_sum_eigh_jvptrace 0.74xJAX 1.18xgrad_sum_svd_s_jvptrace 0.75xJAX 1.25xgrad_sum_eigh_jvptrace 0.76xJAX 0.78xgrad_sum_qr_vjptrace 0.76xJAX 1.43xgrad_sum_svd_s_jvptrace 0.78xJAX 0.80xgrad_sum_lu_jvptrace 0.79xJAX 1.14xgrad_sum_eigh_jvptrace 0.80xJAX 0.83xgrad_sum_qr_vjptrace 0.80xJAX 0.79xgrad_sum_lu_vjptrace 0.81xJAX 0.77xgrad_sum_solve_jvptrace 0.83xJAX 1.46xgrad_sum_svd_s_jvptrace 0.84xJAX 1.14xgrad_sum_qr_jvptrace 0.84xJAX 0.77xgrad_sum_qr_jvptrace 0.84xJAX 0.90xgrad_sum_svd_s_vjptrace 0.86xJAX 1.40xgrad_sum_lu_vjptrace 0.87xJAX 0.87xgrad_sum_lu_vjptrace 0.87xJAX 1.57xgrad_sum_svd_s_vjptrace 0.88xJAX 1.67xgrad_sum_qr_vjptrace 0.88xJAX 0.94xgrad_sum_svd_s_vjptrace 0.89xJAX 0.95xgrad_sum_qr_vjptrace 0.90xJAX 0.87xgrad_sum_svd_s_vjptrace 0.91xJAX 1.42xgrad_sum_qr_jvptrace 0.92xJAX 0.88xgrad_sum_svd_s_vjptrace 0.93xJAX 0.96xgrad_sum_lu_vjptrace 0.93xJAX 1.45xgrad_sum_solve_vjptrace 0.94xJAX 0.78xgrad_sum_lu_vjptrace 0.95xJAX 0.87xgrad_sum_qr_vjptrace 0.96xJAX 1.55xgrad_sum_qr_jvptrace 0.96xJAX 1.67xgrad_sum_qr_jvptrace 0.96xJAX 1.73xgrad_sum_qr_jvptrace 0.96xJAX 1.52xgrad_sum_svd_s_vjptrace 0.98xJAX 1.04xgrad_sum_eigh_vjptrace 0.99xJAX 1.02xgrad_sum_eigh_vjptrace 0.99xJAX 1.03xgrad_sum_eigh_vjptrace 1.01xJAX 1.98xgrad_sum_svd_s_vjptrace 1.02xJAX 1.26xgrad_sum_eigh_vjptrace 1.07xJAX 1.07xgrad_sum_eigh_vjptrace 1.12xJAX 1.77xgrad_sum_eigh_vjptrace 1.15xJAX 2.06xgrad_sum_solve_vjptrace 1.30xJAX 0.92xgrad_sum_solve_vjptrace 1.33xJAX 1.13xgrad_sum_lu_jvptrace 1.41xJAX 1.52xgrad_sum_svd_s_jvptrace 1.45xJAX 1.51xgrad_sum_svd_s_vjptrace 1.49xJAX 1.64xgrad_sum_solve_vjptrace 1.50xJAX 0.88xgrad_sum_solve_jvptrace 1.51xJAX 1.62xgrad_sum_lu_vjptrace 1.53xJAX 1.67xgrad_sum_solve_jvptrace 1.53xJAX 1.01xgrad_sum_qr_vjptrace 1.54xJAX 2.02xgrad_sum_solve_jvptrace 1.58xJAX 1.60xgrad_sum_svd_s_jvptrace 1.60xJAX 1.34xgrad_sum_lu_vjptrace 1.60xJAX 1.89xgrad_sum_eigh_jvptrace 1.64xJAX 1.48xgrad_sum_lu_jvptrace 1.79xJAX 1.06xgrad_sum_svd_s_vjptrace 1.79xJAX 1.39xgrad_sum_solve_jvptrace 1.92xJAX 0.38xgrad_sum_qr_jvptrace 2.19xJAX 2.30xgrad_sum_eigh_jvptrace 2.21xJAX 1.54xgrad_sum_qr_vjptrace 2.25xJAX 1.76xgrad_sum_lu_vjptrace 2.33xJAX 1.70xgrad_sum_eigh_vjptrace 2.53xJAX 2.04xgrad_sum_eigh_jvptrace 2.93xJAX 2.44xgrad_sum_solve_vjptrace 3.19xJAX 0.64xgrad_sum_lu_vjptrace 3.96xJAX 2.76xgrad_sum_qr_jvptrace 4.33xJAX 1.99xgrad_sum_qr_vjptrace 4.64xJAX 1.90xgrad_sum_eigh_jvptrace 4.76xJAX 1.83xgrad_sum_solve_jvptrace 5.29xJAX 1.59xgrad_sum_eigh_vjptrace 5.33xJAX 2.92xgrad_sum_qr_vjptrace 5.51xJAX 3.02xgrad_sum_qr_jvptrace 5.53xJAX 3.14xgrad_sum_eigh_jvptrace 7.90xJAX 2.49xgrad_sum_eigh_vjptrace 9.02xJAX 2.83x
cpu/permutation 30 cases

baseline best reference · best first · 30 cases · case table, absolute times and raw report →

Cases without a comparable baseline (30)

cyclic_15d_3, cyclic_15d_3, cyclic_15d_3, memcpy_24d_contiguous, memcpy_24d_contiguous, memcpy_24d_contiguous, reverse_15d_3, reverse_15d_3, reverse_15d_3, reverse_23d_2, reverse_23d_2, reverse_23d_2, rotation_6d_32_32_32_32_16_16, rotation_6d_32_32_32_32_16_16, rotation_6d_32_32_32_32_16_16, tn_light_415_24d_contiguous_same_perm, tn_light_415_24d_contiguous_same_perm, tn_light_415_24d_contiguous_same_perm, tn_light_415_24d_scattered_to_colmajor, tn_light_415_24d_scattered_to_colmajor, tn_light_415_24d_scattered_to_colmajor, transpose_2d_2048, transpose_2d_2048, transpose_2d_2048, transpose_3d_256_102, transpose_3d_256_102, transpose_3d_256_102, transpose_3d_256_201, transpose_3d_256_201, transpose_3d_256_201

cpu/public_api 1014 cases · 244 flagged

baseline PyTorch · best first · 1014 cases · case table, absolute times and raw report →

bar = case median / PyTorch median · dashed line = baseline (1.00x) · log scale · shorter (left) is faster · whisker = estimated 95% CIreshapetrace 0.00xJAX 1.11xbroadcast_in_dimtrace 0.00xJAX 1.02xtransposetrace 0.00xJAX 1.25xtransposetrace 0.00xJAX 0.78xtransposetrace 0.00xJAX 0.53xreshapetrace 0.00xJAX 1.21xbroadcast_in_dimtrace 0.00xJAX 1.74xreshapetrace 0.00xJAX 1.15xbroadcast_in_dimtrace 0.00xJAX 1.92xslicetrace 0.00xJAX 0.88xslicetrace 0.00xJAX 1.08xslicetrace 0.00xJAX 1.21xreduce_min_axis0trace 0.06xJAX 0.07xreduce_max_axis0trace 0.07xJAX 0.07xreduce_min_axis0trace 0.08xJAX 0.20xsigmoideager 0.16xJAX 1.77xerfeager 0.20xJAX 0.87xsilueager 0.22xJAX 1.73xreduce_max_axis0trace 0.23xJAX 0.17xexptrace 0.23xJAX 0.67xsintrace 0.23xJAX 1.84xreduce_prod_axis0trace 0.25xJAX 0.52xreduce_sum_axis0trace 0.28xJAX 0.54xgelueager 0.29xJAX 1.53xcostrace 0.30xJAX 1.92xreduce_max_axis0trace 0.31xJAX 0.20xsintrace 0.33xJAX 3.50xbatched_svdvalsJAX 0.33xlogtrace 0.37xJAX 1.13xbatched_svdvalsJAX 0.37xreduce_min_axis0trace 0.37xJAX 0.18xclamptrace 0.39xJAX 1.21xmasked_softmaxeager 0.39xlogtrace 0.39xJAX 2.00xmasked_softmaxeager 0.40xmasked_softmaxeager 0.40xtanhtrace 0.41xJAX 0.84xtanhtrace 0.42xJAX 0.35xlog1ptrace 0.43xJAX 1.43xtanhtrace 0.43xJAX 0.97xbatched_choleskytrace 0.45xJAX 0.58xbatched_choleskytrace 0.45xJAX 0.67xbatched_svdvalsJAX 0.46xmasked_softmaxeager 0.46xsigmoideager 0.47xJAX 1.76xsigmoideager 0.48xJAX 1.64xsilueager 0.50xJAX 1.66xbatched_lutrace 0.50xJAX 0.91xsilueager 0.50xJAX 1.83xbatched_lutrace 0.50xJAX 1.49xbatched_svdvalsJAX 0.50xexptrace 0.50xJAX 1.21xbatched_solvetrace 0.50xJAX 0.68xbatched_solvetrace 0.51xJAX 0.68xmasked_softmaxeager 0.51xbatched_lutrace 0.52xJAX 0.95xbatched_solvetrace 0.52xJAX 0.74xbatched_solvetrace 0.52xJAX 0.84xbatched_lutrace 0.52xJAX 0.99xbatched_solvetrace 0.53xJAX 0.73xbatched_lutrace 0.53xJAX 1.62xbatched_lutrace 0.53xJAX 2.65xbatched_solvetrace 0.53xJAX 0.69xbatched_solvetrace 0.53xJAX 0.81xbatched_solvetrace 0.54xJAX 1.86xbatched_solvetrace 0.54xJAX 0.82xbatched_lutrace 0.54xJAX 1.73xbatched_lutrace 0.54xJAX 2.82xbatched_lutrace 0.54xJAX 0.99xbatched_lutrace 0.54xJAX 1.83xbatched_lutrace 0.54xJAX 1.06xbatched_lutrace 0.54xJAX 1.06xbatched_lutrace 0.55xJAX 1.04xbatched_lutrace 0.55xJAX 1.95xbatched_solvetrace 0.55xJAX 0.89xbatched_lutrace 0.55xJAX 1.11xlayer_normeager 0.55xJAX 0.97xbatched_solvetrace 0.55xJAX 0.81xgelueager 0.55xJAX 1.55xbatched_lutrace 0.55xJAX 1.07xbatched_lutrace 0.55xJAX 2.08xbatched_lutrace 0.56xJAX 1.77xbatched_lutrace 0.56xJAX 3.18xbatched_lutrace 0.56xJAX 1.57xcompare_lttrace 0.56xJAX 0.98xsqrttrace 0.56xJAX 0.66xbatched_lutrace 0.56xJAX 1.67xbatched_solvetrace 0.57xJAX 0.72xbatched_lutrace 0.57xJAX 1.72xbatched_solvetrace 0.57xJAX 0.91xbatched_solvetrace 0.57xJAX 0.65xmasked_log_softmax_axis1eager 0.57xbatched_lutrace 0.57xJAX 1.08xbatched_lutrace 0.57xJAX 1.96xbatched_solvetrace 0.57xJAX 0.65xbatched_solvetrace 0.57xJAX 1.10xcostrace 0.58xJAX 3.39xbatched_lutrace 0.58xJAX 2.54xbatched_solvetrace 0.58xJAX 2.10xbatched_solvetrace 0.58xJAX 2.31xbatched_lutrace 0.58xJAX 1.13xbatched_solvetrace 0.58xJAX 0.74xbatched_solvetrace 0.58xJAX 0.79xbatched_lutrace 0.59xJAX 1.13xbatched_lutrace 0.59xJAX 1.92xbatched_solvetrace 0.59xJAX 1.93xbatched_solvetrace 0.59xJAX 1.32xreduce_max_axis1trace 0.59xJAX 1.48xbatched_solvetrace 0.59xJAX 0.81xbatched_lutrace 0.59xJAX 1.04xgelueager 0.59xJAX 2.07xbatched_solvetrace 0.59xJAX 0.75xbatched_solvetrace 0.60xJAX 0.70xbatched_solvetrace 0.60xJAX 0.81xbatched_lutrace 0.60xJAX 2.11xbatched_solvetrace 0.60xJAX 1.39xbatched_svdvalsJAX 0.60xbatched_solvetrace 0.60xJAX 0.81xbatched_solvetrace 0.61xJAX 0.74xbatched_lutrace 0.61xJAX 1.27xbatched_solvetrace 0.61xJAX 0.99xbatched_solvetrace 0.61xJAX 0.88xbatched_solvetrace 0.61xJAX 0.84xmasked_softmaxeager 0.61xbatched_lutrace 0.62xJAX 1.35xbatched_svdvalsJAX 0.62xbatched_choleskytrace 0.62xJAX 1.72xbatched_solvetrace 0.62xJAX 1.32xgathertrace 0.63xJAX 0.51xbatched_solvetrace 0.63xJAX 0.91xbatched_lutrace 0.63xJAX 1.20xpadtrace 0.63xJAX 0.81xbatched_lutrace 0.63xJAX 1.84xexpm1trace 0.64xJAX 0.92xbatched_lutrace 0.64xJAX 1.51xrms_normeager 0.65xJAX 3.85xbatched_choleskytrace 0.65xJAX 1.14xbatched_lutrace 0.65xJAX 1.33xrms_normeager 0.65xJAX 1.95xbatched_choleskytrace 0.65xJAX 2.13xbatched_solvetrace 0.66xJAX 1.65xbatched_solvetrace 0.66xJAX 1.56xmasked_log_softmax_axis1eager 0.66xremtrace 0.66xJAX 1.17xgelueager 0.67xJAX 2.50xcholeskytrace 0.68xJAX 1.22xbatched_eigvalshtrace 0.68xJAX 1.07xsqrttrace 0.68xJAX 1.34xbatched_eigvalshtrace 0.69xJAX 1.16xsintrace 0.69xJAX 4.69xlogtrace 0.70xJAX 1.18xbatched_eigvalshtrace 0.70xJAX 1.14xbatched_choleskytrace 0.70xJAX 0.78xbatched_choleskytrace 0.71xJAX 1.24xbatched_solvetrace 0.71xJAX 1.20xerfeager 0.71xJAX 0.78xerfeager 0.72xJAX 0.35xlutrace 0.72xJAX 1.02xsoftpluseager 0.72xJAX 1.79xgelu_tanheager 0.72xJAX 0.71xbatched_eigvalshtrace 0.73xJAX 1.08xlogtrace 0.73xJAX 1.17xbatched_eigvalshtrace 0.73xJAX 1.11xsoftpluseager 0.73xJAX 1.68xbatched_qrtrace 0.73xJAX 1.68xsqrttrace 0.74xJAX 1.38xcostrace 0.74xJAX 4.60xbatched_eigvalshtrace 0.74xJAX 1.11xbatched_eigvalshtrace 0.74xJAX 1.09xgelu_tanheager 0.74xJAX 0.67xlogtrace 0.74xJAX 1.21xbatched_eigvalshtrace 0.74xJAX 1.10xbatched_choleskytrace 0.74xJAX 1.14xbatched_qrtrace 0.74xJAX 1.39xexptrace 0.74xJAX 0.81xwhere_selecteager 0.74xJAX 1.23xextract_diagonaltrace 0.75xJAX 0.80xbatched_qrtrace 0.75xJAX 0.78xbatched_eigvalshtrace 0.75xJAX 1.89xbatched_eigvalshtrace 0.75xJAX 1.15xlayer_normeager 0.75xJAX 2.41xbatched_qrtrace 0.75xJAX 0.98xbatched_eigvalshtrace 0.76xJAX 3.32xselecttrace 0.76xJAX 1.02xbatched_eigvalshtrace 0.76xJAX 1.07xexptrace 0.76xJAX 1.56xbatched_solvetrace 0.76xJAX 0.93xbatched_eigvalshtrace 0.77xJAX 1.13xrms_normeager 0.77xJAX 1.84xbatched_choleskytrace 0.77xJAX 1.14xbatched_eigvalshtrace 0.77xJAX 1.40xbatched_solvetrace 0.77xJAX 1.22xbatched_qrtrace 0.78xJAX 1.15xbatched_eigvalshtrace 0.79xJAX 2.29xbatched_qrtrace 0.79xJAX 2.95xlog1ptrace 0.79xJAX 2.79xbatched_eigvalshtrace 0.79xJAX 1.15xbatched_lutrace 0.80xJAX 1.42xbatched_eigvalshtrace 0.80xJAX 1.35xbatched_qrtrace 0.80xJAX 0.87xclamptrace 0.81xJAX 1.60xbatched_lutrace 0.81xJAX 2.10xreduce_sum_axis0trace 0.81xJAX 0.57xeinsum_ij_jk_ikJAX 0.81xbatched_qrtrace 0.82xJAX 0.92xbatched_eightrace 0.82xJAX 1.47xbatched_qrtrace 0.82xJAX 1.41xbatched_lutrace 0.82xJAX 1.49xbatched_lutrace 0.82xJAX 0.90xbatched_eightrace 0.82xJAX 1.18xbatched_eigvalshtrace 0.82xJAX 1.28xbatched_choleskytrace 0.83xJAX 1.22xeinsum_ij_jk_ikJAX 0.83xbatched_lutrace 0.83xJAX 1.60xsvdtrace 0.83xJAX 0.71xbatched_eigvalshtrace 0.83xJAX 1.30xtriltrace 0.83xJAX 1.59xlogtrace 0.83xJAX 2.37xbatched_eightrace 0.84xJAX 1.60xbatched_lutrace 0.84xJAX 1.77xcholeskytrace 0.84xJAX 3.01xbatched_eigvalshtrace 0.84xJAX 1.60xbatched_eigvalshtrace 0.84xJAX 1.29xwhere_selecteager 0.85xJAX 1.55xbatched_qrtrace 0.85xJAX 1.72xbatched_eigvalshtrace 0.85xJAX 1.36xreduce_prod_axis0trace 0.85xJAX 0.51xbatched_eightrace 0.85xJAX 1.52xbatched_choleskytrace 0.85xJAX 2.15xbatched_lutrace 0.85xJAX 1.16xbatched_choleskytrace 0.85xJAX 1.74xbatched_lutrace 0.86xJAX 1.07xbatched_eightrace 0.86xJAX 1.46xbatched_eightrace 0.86xJAX 1.33xtriltrace 0.86xJAX 1.82xbatched_choleskytrace 0.86xJAX 2.24xbatched_eightrace 0.86xJAX 1.55xbatched_eigvalshtrace 0.86xJAX 1.41xcholeskytrace 0.86xJAX 0.86xbatched_svdtrace 0.86xJAX 0.76xbatched_eightrace 0.86xJAX 1.52xselecttrace 0.86xJAX 1.51xbatched_eigvalshtrace 0.87xJAX 1.34xbatched_choleskytrace 0.87xJAX 0.98xqrtrace 0.87xJAX 0.80xbatched_eigvalshtrace 0.87xJAX 1.58xbatched_eightrace 0.87xJAX 1.50xbatched_eightrace 0.87xJAX 1.60xbatched_eightrace 0.87xJAX 1.71xbatched_eightrace 0.87xJAX 4.31xtriutrace 0.88xJAX 1.04xbatched_eightrace 0.88xJAX 1.58xbatched_eightrace 0.88xJAX 1.72xbatched_svdtrace 0.88xJAX 0.75xlutrace 0.88xJAX 2.03xbatched_lutrace 0.88xJAX 0.70xbatched_eightrace 0.88xJAX 2.62xslogdettrace 0.88xJAX 0.90xbatched_choleskytrace 0.88xJAX 2.46xdettrace 0.88xJAX 0.90xbatched_solvetrace 0.89xJAX 1.18xbatched_qrtrace 0.89xJAX 0.92xhouseholder_qreager 0.89xJAX 1.16xbatched_eightrace 0.89xJAX 1.49xrsqrttrace 0.89xJAX 2.79xbatched_eigvalshtrace 0.89xJAX 1.12xbatched_svdtrace 0.89xJAX 0.98xgelu_tanheager 0.89xJAX 1.97xbatched_svdtrace 0.89xJAX 1.00xbatched_svdtrace 0.89xJAX 1.70xbatched_eigvalshtrace 0.89xJAX 1.65xbatched_choleskytrace 0.89xJAX 1.39xsubtrace 0.89xJAX 1.51xbatched_eigvalshtrace 0.89xJAX 5.05xnorm_frotrace 0.89xJAX 12.82xaddtrace 0.89xJAX 1.52xbatched_lutrace 0.90xJAX 1.64xbatched_choleskytrace 0.90xJAX 2.73xbatched_svdtrace 0.90xJAX 1.72xbatched_eightrace 0.90xJAX 1.41xbatched_eightrace 0.90xJAX 2.54xmultrace 0.90xJAX 1.52xreduce_prod_axis0trace 0.90xJAX 0.58xbatched_svdtrace 0.90xJAX 1.07xbatched_svdtrace 0.90xJAX 0.89xbatched_svdtrace 0.90xJAX 0.95xbatched_eightrace 0.90xJAX 1.43xtriltrace 0.90xJAX 1.01xbatched_eigvalshtrace 0.90xJAX 2.72xbatched_svdtrace 0.90xJAX 0.92xbatched_eightrace 0.91xJAX 1.43xbatched_eightrace 0.91xJAX 1.36xbatched_eightrace 0.91xJAX 1.37xbatched_eightrace 0.91xJAX 2.79xbatched_eightrace 0.91xJAX 1.41xbatched_eigvalshtrace 0.91xJAX 1.74xbatched_eightrace 0.91xJAX 1.60xbatched_svdtrace 0.91xJAX 0.24xbatched_eightrace 0.91xJAX 1.70xbatched_eightrace 0.91xJAX 1.54xbatched_eigvalshtrace 0.91xJAX 2.38xbatched_svdtrace 0.91xJAX 0.28xbatched_svdtrace 0.91xJAX 1.02xbatched_svdtrace 0.91xJAX 1.43xbatched_eigvalshtrace 0.91xJAX 2.54xbatched_eigvalshtrace 0.92xJAX 2.33xwhere_selecteager 0.92xJAX 1.69xdivtrace 0.92xJAX 1.56xsubtrace 0.92xJAX 1.58xbatched_choleskytrace 0.92xJAX 2.06xbatched_svdtrace 0.92xJAX 0.93xbatched_svdtrace 0.92xJAX 0.90xbatched_choleskytrace 0.92xJAX 2.27xbatched_eigvalshtrace 0.92xJAX 2.57xmultrace 0.92xJAX 1.61xbatched_choleskytrace 0.92xJAX 3.92xbatched_svdtrace 0.92xJAX 1.15xaddtrace 0.92xJAX 1.59xreduce_min_axis1trace 0.92xJAX 1.49xdynamic_slicetrace 0.92xJAX 1.13xnegtrace 0.92xJAX 1.75xbatched_eightrace 0.92xJAX 1.49xdivtrace 0.92xJAX 1.64xbatched_choleskytrace 0.92xJAX 2.31xbatched_choleskytrace 0.92xJAX 1.36xbatched_eightrace 0.92xJAX 1.36xabstrace 0.92xJAX 1.74xbatched_choleskytrace 0.92xJAX 2.29xbatched_svdtrace 0.92xJAX 1.20xbatched_svdtrace 0.92xJAX 0.98xbatched_svdtrace 0.92xJAX 0.46xpinv_with_rtoltrace 0.92xJAX 0.93xbatched_eightrace 0.92xJAX 4.42xbatched_eigvalshtrace 0.92xJAX 4.53xconjtrace 0.92xJAX 1.64xeigtrace 0.92xJAX 1.03xbatched_eightrace 0.92xJAX 1.47xbatched_eightrace 0.92xJAX 1.41xbatched_eightrace 0.92xJAX 1.49xrsqrttrace 0.92xJAX 2.55xbatched_eightrace 0.92xJAX 1.23xmasked_log_softmax_axis1eager 0.93xbatched_eightrace 0.93xJAX 1.81xbatched_eightrace 0.93xJAX 1.65xsigntrace 0.93xJAX 1.84xbatched_solvetrace 0.93xJAX 1.04xdivtrace 0.93xJAX 1.79xpinvtrace 0.93xJAX 0.94xselecttrace 0.93xJAX 1.68xbatched_svdtrace 0.93xJAX 1.62xbatched_svdtrace 0.93xJAX 0.91xbatched_eightrace 0.93xJAX 1.40xbatched_svdtrace 0.93xJAX 0.87xbatched_choleskytrace 0.93xJAX 2.34xbatched_eigvalshtrace 0.93xJAX 1.75xmaximumtrace 0.93xJAX 1.60xbatched_eightrace 0.93xJAX 1.72xbatched_svdtrace 0.93xJAX 1.33xbatched_svdtrace 0.93xJAX 0.30xhouseholder_qreager 0.93xJAX 1.10xbatched_svdtrace 0.93xJAX 1.42xminimumtrace 0.94xJAX 1.60xbatched_eightrace 0.94xJAX 3.86xbatched_eightrace 0.94xJAX 2.34xbatched_eigvalshtrace 0.94xJAX 1.57xbatched_svdtrace 0.94xJAX 1.76xconjtrace 0.94xJAX 1.57xbatched_svdtrace 0.94xJAX 1.48xabstrace 0.94xJAX 1.77xbatched_lutrace 0.94xJAX 1.14xbatched_eightrace 0.94xJAX 1.41xbatched_svdtrace 0.94xJAX 1.50xbatched_eightrace 0.94xJAX 1.84xbatched_eightrace 0.95xJAX 1.23xdivtrace 0.95xJAX 1.14xbatched_eightrace 0.95xJAX 1.63xcompare_lttrace 0.95xJAX 1.18xbatched_eightrace 0.95xJAX 2.31xembed_diagonaltrace 0.95xJAX 1.94xreduce_sum_alltrace 0.95xJAX 2.97xbatched_choleskytrace 0.95xJAX 4.11xmaximumtrace 0.95xJAX 1.52xrsqrttrace 0.95xJAX 1.61xbatched_eightrace 0.95xJAX 1.42xminimumtrace 0.95xJAX 1.56xbatched_choleskytrace 0.95xJAX 2.50xbatched_eightrace 0.95xJAX 0.83xnegtrace 0.95xJAX 1.72xbatched_eightrace 0.95xJAX 1.59xbatched_choleskytrace 0.96xJAX 1.33xbatched_eightrace 0.96xJAX 1.57xbatched_eigvalshtrace 0.96xJAX 5.34xbatched_eightrace 0.96xJAX 1.66xbatched_svdtrace 0.96xJAX 1.38xbatched_qrtrace 0.96xJAX 3.06xbatched_eightrace 0.96xJAX 1.55xbatched_eightrace 0.96xJAX 1.28xbatched_svdtrace 0.96xJAX 1.74xbatched_eigvalshtrace 0.96xJAX 1.57xbatched_eightrace 0.96xJAX 1.17xsigntrace 0.96xJAX 1.55xbatched_choleskytrace 0.96xJAX 1.37xbatched_svdtrace 0.96xJAX 1.15xbatched_eightrace 0.96xJAX 0.48xcompare_lttrace 0.96xJAX 1.11xeigtrace 0.96xJAX 0.97xbatched_eightrace 0.96xJAX 0.63xsvd_fulleager 0.97xtrace 1.00xJAX 1.54xdivtrace 0.97xJAX 1.89xbatched_eightrace 0.97xJAX 2.77xbatched_svdtrace 0.97xJAX 2.59xaddtrace 0.97xJAX 1.22xbatched_eightrace 0.97xJAX 1.23xbatched_svdtrace 0.97xJAX 1.38xbatched_choleskytrace 0.97xJAX 2.04xbatched_eigvalshtrace 0.97xJAX 1.12xsubtrace 0.97xJAX 1.21xbatched_eightrace 0.97xJAX 1.02xbatched_eigvalshtrace 0.97xJAX 2.18xtriutrace 0.97xJAX 1.87xbatched_eightrace 0.97xJAX 0.91xinvtrace 0.97xJAX 1.04xbatched_eightrace 0.97xJAX 1.28xbatched_eightrace 0.97xJAX 1.38xdot_general_with_conjJAX 0.97xbatched_svdtrace 0.97xJAX 2.90xbatched_eightrace 0.98xJAX 1.03xbatched_choleskytrace 0.98xJAX 1.62xbatched_eightrace 0.98xJAX 3.28xbatched_eightrace 0.98xJAX 2.54xsoftmaxeager 0.98xJAX 2.12xbatched_eightrace 0.98xJAX 1.23xeinsum_ij_jk_ikJAX 0.98xbatched_eightrace 0.98xJAX 1.30xnorm_frotrace 0.98xJAX 0.82xbatched_choleskytrace 0.98xJAX 1.12xbatched_svdvalsJAX 0.98xchain_log1p_exp_multrace 0.98xJAX 0.85xbatched_choleskytrace 0.99xJAX 4.08xbatched_eigvalshtrace 0.99xJAX 2.88xbatched_eightrace 0.99xJAX 1.26xsvd_fulleager 0.99xtrace 0.99xJAX 1.05xbatched_choleskytrace 0.99xJAX 4.09xbatched_eightrace 0.99xJAX 0.71xbatched_eigvalshtrace 0.99xJAX 1.21xbatched_svdtrace 0.99xJAX 1.55xbatched_svdtrace 0.99xJAX 1.38xbatched_svdtrace 0.99xJAX 1.17xreduce_prod_alltrace 0.99xJAX 1.84xbatched_choleskytrace 0.99xJAX 3.67xsvd_fulleager 0.99xtrace 1.00xJAX 1.63xlutrace 0.99xJAX 2.34xbatched_qrtrace 0.99xJAX 2.10xbatched_svdtrace 0.99xJAX 2.58xbatched_eightrace 0.99xJAX 1.56xbatched_eightrace 0.99xJAX 0.82xembed_diagonaltrace 0.99xJAX 1.83xbatched_eightrace 0.99xJAX 1.27xreduce_prod_alltrace 0.99xJAX 3.18xbatched_qrtrace 0.99xJAX 1.56xbatched_eigvalshtrace 0.99xJAX 2.64xbatched_eightrace 0.99xJAX 1.31xbatched_svdtrace 1.00xJAX 1.53xbatched_choleskytrace 1.00xJAX 1.41xeigtrace 1.00xJAX 1.34xbatched_eightrace 1.00xJAX 1.77xmultrace 1.00xJAX 1.25xsolvetrace 1.00xJAX 0.98xbatched_choleskytrace 1.00xJAX 1.38xbatched_choleskytrace 1.00xJAX 1.70xbatched_svdtrace 1.00xJAX 3.38xbatched_eightrace 1.00xJAX 1.31xmultrace 1.00xJAX 1.61xbatched_svdvalsJAX 1.00xconjtrace 1.00xJAX 1.18xbatched_eightrace 1.00xJAX 3.84xbatched_eightrace 1.01xJAX 1.18xbatched_eightrace 1.01xJAX 1.47xbatched_svdtrace 1.01xJAX 1.17xbatched_qrtrace 1.01xJAX 2.95xbatched_svdtrace 1.01xJAX 1.59xbatched_eightrace 1.01xJAX 1.21xbatched_eightrace 1.01xJAX 1.29xbatched_eightrace 1.01xJAX 1.11xembed_diagonaltrace 1.01xJAX 1.32xbatched_svdtrace 1.01xJAX 1.54xbatched_eightrace 1.01xJAX 1.12xbatched_eightrace 1.01xJAX 1.40xbatched_svdtrace 1.01xJAX 2.36xminimumtrace 1.01xJAX 1.05xbatched_choleskytrace 1.01xJAX 1.65xbatched_eightrace 1.01xJAX 3.70xbatched_svdtrace 1.01xJAX 1.38xbatched_svdtrace 1.02xJAX 1.17xsigntrace 1.02xJAX 1.16xbatched_choleskytrace 1.02xJAX 1.41xbatched_svdtrace 1.02xJAX 1.19xbatched_eightrace 1.02xJAX 2.59xbatched_eightrace 1.02xJAX 0.47xabstrace 1.02xJAX 1.21xbatched_svdtrace 1.02xJAX 1.56xbatched_choleskytrace 1.02xJAX 1.47xdynamic_update_sliceJAX 1.02xbatched_qrtrace 1.02xJAX 2.00xmaximumtrace 1.02xJAX 1.04xbatched_choleskytrace 1.02xJAX 3.00xeigvalstrace 1.02xJAX 1.05xbatched_qrtrace 1.02xJAX 1.80xbatched_qrtrace 1.03xJAX 3.11xbatched_eightrace 1.03xJAX 1.12xtensordottrace 1.03xJAX 1.22xtensordottrace 1.03xJAX 1.20xbatched_eightrace 1.03xJAX 4.29xbatched_qrtrace 1.03xJAX 1.87xexptrace 1.03xJAX 1.58xbatched_qrtrace 1.03xJAX 1.92xbatched_choleskytrace 1.03xJAX 1.67xconcatenatetrace 1.03xJAX 1.04xbatched_eightrace 1.03xJAX 2.45xdot_generaltrace 1.03xJAX 1.22xbatched_eightrace 1.03xJAX 2.25xbatched_svdvalsJAX 1.03xnegtrace 1.03xJAX 1.04xbatched_qrtrace 1.03xJAX 5.56xbatched_qrtrace 1.03xJAX 0.75xbatched_svdtrace 1.03xJAX 1.01xbatched_qrtrace 1.03xJAX 0.59xbatched_eightrace 1.03xJAX 2.09xtensordottrace 1.04xJAX 0.97xbatched_lutrace 1.04xJAX 1.15xchain_log1p_exp_multrace 1.04xJAX 0.94xbatched_svdtrace 1.04xJAX 1.55xbatched_choleskytrace 1.04xJAX 1.63xbatched_choleskytrace 1.04xJAX 1.50xbatched_qrtrace 1.04xJAX 1.94xdot_generaltrace 1.04xJAX 0.98xbatched_qrtrace 1.04xJAX 1.43xbatched_svdtrace 1.04xJAX 1.24xbatched_qrtrace 1.04xJAX 1.98xbatched_qrtrace 1.04xJAX 5.42xbatched_svdtrace 1.04xJAX 0.98xbatched_choleskytrace 1.04xJAX 1.67xbatched_choleskytrace 1.04xJAX 1.64xbatched_choleskytrace 1.04xJAX 3.38xbatched_qrtrace 1.04xJAX 1.87xbatched_svdtrace 1.05xJAX 1.02xbatched_solvetrace 1.05xJAX 1.20xbatched_svdtrace 1.05xJAX 1.19xbatched_svdtrace 1.05xJAX 1.17xbatched_svdtrace 1.05xJAX 1.03xlog1ptrace 1.05xJAX 3.01xbatched_svdtrace 1.05xJAX 1.13xbatched_svdtrace 1.05xJAX 1.24xbatched_svdtrace 1.05xJAX 0.92xbatched_svdtrace 1.05xJAX 1.02xbatched_svdtrace 1.05xJAX 1.00xbatched_qrtrace 1.06xJAX 1.84xbatched_svdtrace 1.06xJAX 1.14xbatched_svdtrace 1.06xJAX 1.02xbatched_eightrace 1.06xJAX 1.19xbatched_svdtrace 1.06xJAX 1.03xbatched_choleskytrace 1.06xJAX 1.73xbatched_svdtrace 1.06xJAX 1.53xbatched_svdtrace 1.06xJAX 1.01xbatched_eightrace 1.06xJAX 1.14xbatched_choleskytrace 1.06xJAX 1.69xbatched_svdtrace 1.06xJAX 1.21xbatched_choleskytrace 1.06xJAX 1.71xbatched_qrtrace 1.06xJAX 2.99xclamptrace 1.06xJAX 1.52xbatched_svdtrace 1.06xJAX 1.42xbatched_svdtrace 1.06xJAX 1.42xbatched_choleskytrace 1.06xJAX 1.73xbatched_svdtrace 1.07xJAX 1.46xbatched_qrtrace 1.07xJAX 3.01xbatched_svdtrace 1.07xJAX 1.01xbatched_svdtrace 1.07xJAX 1.01xbatched_svdtrace 1.07xJAX 1.17xbatched_svdtrace 1.07xJAX 1.20xlog_softmaxeager 1.07xJAX 2.33xbatched_svdtrace 1.07xJAX 1.18xmultrace 1.07xJAX 1.17xbatched_qrtrace 1.07xJAX 1.94xreduce_sum_alltrace 1.07xJAX 1.81xbatched_svdtrace 1.07xJAX 1.03xbatched_svdtrace 1.07xJAX 1.56xbatched_eigvalshtrace 1.07xJAX 2.14xbatched_eigvalshtrace 1.07xJAX 1.63xbatched_svdtrace 1.07xJAX 1.48xbatched_svdtrace 1.08xJAX 2.67xmultrace 1.08xJAX 1.64xbatched_svdtrace 1.08xJAX 0.99xbatched_svdvalsJAX 1.08xbatched_qrtrace 1.08xJAX 1.82xreduce_min_alltrace 1.08xJAX 3.82xbatched_eigvalshtrace 1.08xJAX 0.88xbatched_qrtrace 1.08xJAX 1.89xbatched_svdtrace 1.08xJAX 0.98xbatched_svdtrace 1.08xJAX 1.05xbatched_svdtrace 1.08xJAX 0.97xbatched_qrtrace 1.08xJAX 2.06xbatched_svdtrace 1.08xJAX 1.19xbatched_qrtrace 1.08xJAX 3.28xbatched_qrtrace 1.08xJAX 2.03xeigtrace 1.08xJAX 1.11xbatched_svdtrace 1.08xJAX 0.23xeigtrace 1.08xJAX 1.28xdot_generaltrace 1.09xJAX 1.28xbatched_svdtrace 1.09xJAX 1.97xcholeskytrace 1.09xJAX 5.06xbatched_svdtrace 1.09xJAX 3.36xbatched_svdtrace 1.09xJAX 0.87xbatched_svdtrace 1.09xJAX 4.42xbatched_eightrace 1.09xJAX 2.03xbatched_svdtrace 1.09xJAX 0.24xbatched_svdtrace 1.09xJAX 1.22xbatched_svdtrace 1.09xJAX 1.15xbatched_eigvalshtrace 1.09xJAX 1.74xbatched_svdtrace 1.09xJAX 0.91xreduce_prod_alltrace 1.09xJAX 1.94xbatched_qrtrace 1.09xJAX 1.88xbatched_qrtrace 1.10xJAX 5.81xreduce_max_alltrace 1.10xJAX 3.86xbatched_svdtrace 1.10xJAX 1.05xbatched_qrtrace 1.10xJAX 2.99xreduce_sum_alltrace 1.10xJAX 1.85xbatched_svdvalsJAX 1.10xbatched_svdtrace 1.10xJAX 1.08xbatched_svdtrace 1.10xJAX 1.04xbatched_svdtrace 1.10xJAX 1.04xbatched_qrtrace 1.10xJAX 2.28xreduce_max_alltrace 1.11xJAX 3.69xbatched_svdtrace 1.11xJAX 1.12xreversetrace 1.11xJAX 1.09xbatched_svdtrace 1.11xJAX 1.12xbatched_eigvalshtrace 1.11xJAX 1.70xqrtrace 1.11xJAX 1.25xeigvalshtrace 1.11xJAX 2.14xbatched_svdtrace 1.11xJAX 1.06xreduce_min_alltrace 1.12xJAX 5.14xbatched_qrtrace 1.12xJAX 1.45xbatched_eightrace 1.12xJAX 0.77xbatched_svdtrace 1.12xJAX 0.43xbatched_svdtrace 1.12xJAX 0.46xbatched_qrtrace 1.12xJAX 2.17xbatched_triangular_solvetrace 1.12xJAX 1.31xbatched_eightrace 1.13xJAX 1.49xbatched_eightrace 1.13xJAX 0.78xhouseholder_qreager 1.13xJAX 1.74xeigtrace 1.13xJAX 1.10xbatched_svdtrace 1.13xJAX 1.88xbatched_svdvalsJAX 1.13xreduce_sum_axis1trace 1.13xJAX 1.46xbatched_svdvalsJAX 1.13xbatched_qrtrace 1.14xJAX 1.85xsvdtrace 1.14xJAX 1.08xexpm1trace 1.14xJAX 1.59xbatched_eightrace 1.14xJAX 1.34xbatched_svdtrace 1.14xJAX 4.38xbatched_eightrace 1.14xJAX 2.34xbatched_eightrace 1.14xJAX 0.70xbatched_eightrace 1.14xJAX 0.98xbatched_svdtrace 1.14xJAX 0.33xbatched_qrtrace 1.15xJAX 2.04xbatched_svdvalsJAX 1.15xbatched_qrtrace 1.15xJAX 2.11xextract_diagonaltrace 1.15xJAX 1.41xbatched_qrtrace 1.15xJAX 1.63xeigvalstrace 1.15xJAX 1.19xbatched_eightrace 1.15xJAX 0.57xbatched_svdtrace 1.15xJAX 0.98xbatched_eigvalshtrace 1.16xJAX 1.92xbatched_svdtrace 1.16xJAX 0.41xbatched_svdtrace 1.16xJAX 0.89xbatched_triangular_solvetrace 1.16xJAX 1.43xbatched_eightrace 1.16xJAX 4.82xbatched_svdtrace 1.16xJAX 0.38xbatched_svdvalsJAX 1.16xbatched_qrtrace 1.16xJAX 2.40xbatched_qrtrace 1.17xJAX 2.40xbatched_svdtrace 1.17xJAX 0.35xbatched_solvetrace 1.17xJAX 1.88xbatched_qrtrace 1.17xJAX 2.20xsvdtrace 1.17xJAX 1.28xbatched_qrtrace 1.17xJAX 2.22xremtrace 1.17xJAX 1.94xdot_general_with_conjJAX 1.18xbatched_qrtrace 1.18xJAX 2.36xbatched_svdvalsJAX 1.18xpinv_with_rtoltrace 1.19xJAX 1.25xbatched_svdvalsJAX 1.19xbatched_svdvalsJAX 1.19xbatched_qrtrace 1.19xJAX 1.37xbatched_svdvalsJAX 1.19xbatched_qrtrace 1.19xJAX 2.63xbatched_svdvalsJAX 1.20xbatched_svdvalsJAX 1.20xbatched_svdvalsJAX 1.20xdivtrace 1.20xJAX 1.14xbatched_choleskytrace 1.20xJAX 2.30xeigvalshtrace 1.20xJAX 2.49xbatched_svdtrace 1.20xJAX 1.29xpinvtrace 1.21xJAX 1.43xreduce_prod_axis1trace 1.21xJAX 1.52xdot_general_with_conjJAX 1.21xreduce_max_alltrace 1.21xJAX 5.14xbatched_triangular_solvetrace 1.22xJAX 1.30xbatched_svdtrace 1.22xJAX 1.33xbatched_svdvalsJAX 1.22xbatched_svdvalsJAX 1.23xbatched_svdvalsJAX 1.23xbatched_svdtrace 1.23xJAX 1.37xbatched_svdvalsJAX 1.23xbatched_svdvalsJAX 1.23xbatched_svdvalsJAX 1.23xbatched_svdvalsJAX 1.23xbatched_svdvalsJAX 1.23xbatched_svdvalsJAX 1.24xbatched_svdvalsJAX 1.24xbatched_triangular_solvetrace 1.24xJAX 1.45xbatched_triangular_solvetrace 1.24xJAX 1.32xbatched_svdvalsJAX 1.24xbatched_eigvalshtrace 1.24xJAX 2.66xbatched_choleskytrace 1.25xJAX 1.59xreduce_min_alltrace 1.25xJAX 3.70xbatched_svdvalsJAX 1.25xpowtrace 1.26xJAX 1.58xbatched_svdvalsJAX 1.26xremtrace 1.26xJAX 2.58xbatched_svdvalsJAX 1.26xbatched_svdtrace 1.27xJAX 1.14xpinv_with_rtoltrace 1.27xJAX 1.47xbatched_svdtrace 1.27xJAX 1.31xbatched_svdvalsJAX 1.28xqrtrace 1.28xJAX 1.37xbatched_svdvalsJAX 1.28xexpm1trace 1.28xJAX 2.12xbatched_svdvalsJAX 1.28xchain_log1p_exp_multrace 1.29xJAX 1.18xbatched_svdtrace 1.29xJAX 1.23xbatched_svdvalsJAX 1.31xtriutrace 1.32xJAX 1.91xbatched_eightrace 1.32xJAX 0.79xbatched_svdtrace 1.32xJAX 1.24xbatched_svdvalsJAX 1.32xbatched_eigvalshtrace 1.33xJAX 1.09xbatched_svdvalsJAX 1.34xreduce_meaneager 1.36xJAX 2.72xbatched_svdvalsJAX 1.36xbatched_triangular_solvetrace 1.36xJAX 1.89xbatched_qrtrace 1.37xJAX 1.52xbatched_choleskytrace 1.38xJAX 1.83xbatched_choleskytrace 1.38xJAX 2.53xdynamic_update_sliceJAX 1.38xbatched_svdvalsJAX 1.39xreduce_sum_axis0trace 1.39xJAX 0.53xbatched_svdvalsJAX 1.40xcholeskytrace 1.42xJAX 1.42xsolvetrace 1.44xJAX 1.41xbatched_svdvalsJAX 1.44xexptrace 1.46xJAX 2.20xbatched_qrtrace 1.47xJAX 1.83xbatched_svdvalsJAX 1.47xsoftmaxeager 1.51xJAX 4.18xconcatenatetrace 1.52xJAX 1.37xlog_softmaxeager 1.52xJAX 4.28xbatched_eightrace 1.53xJAX 0.85xbatched_lutrace 1.53xJAX 1.32xbatched_eightrace 1.54xJAX 0.64xhouseholder_qreager 1.54xJAX 2.82xbatched_eightrace 1.57xJAX 1.00xbatched_solvetrace 1.58xJAX 3.23xpinvtrace 1.59xJAX 1.19xreduce_meaneager 1.60xJAX 5.29xpadtrace 1.61xJAX 1.06xbatched_eightrace 1.62xJAX 0.89xinvtrace 1.62xJAX 1.37xdynamic_update_sliceJAX 1.66xextract_diagonaltrace 1.67xJAX 1.45xbatched_eightrace 1.67xJAX 1.81xpadtrace 1.69xJAX 0.97xbatched_svdtrace 1.71xJAX 3.13xbatched_qrtrace 1.71xJAX 1.91xbatched_choleskytrace 1.74xJAX 2.57xslogdettrace 1.74xJAX 2.18xbatched_lu_solveJAX 1.78xsolvetrace 1.78xJAX 1.60xdynamic_slicetrace 1.81xJAX 1.32xreduce_min_axis1trace 1.83xJAX 2.21xreduce_meaneager 1.85xJAX 3.85xbatched_lu_solveJAX 1.86xinvtrace 1.89xJAX 1.23xbatched_eightrace 1.91xJAX 1.70xbatched_lu_factorJAX 1.92xlstsqeager 1.95xtrace 1.92xJAX 5.04xbatched_solvetrace 1.94xJAX 3.26xbatched_eigvalshtrace 1.95xJAX 1.21xbatched_lu_factorJAX 1.96xbatched_lu_solveJAX 1.96xbatched_svdvalsJAX 1.98xbatched_lu_solveJAX 1.98xnorm_frotrace 1.98xJAX 3.70xbatched_lutrace 2.00xJAX 3.75xbatched_svdvalsJAX 2.00xbatched_lu_solveJAX 2.01xbatched_svdvalsJAX 2.03xbatched_eightrace 2.03xJAX 1.74xbatched_eigvalshtrace 2.04xJAX 1.36xbatched_eightrace 2.07xJAX 2.03xbatched_eightrace 2.07xJAX 1.83xbatched_solvetrace 2.08xJAX 2.72xlstsqeager 2.08xtrace 2.09xJAX 8.01xbatched_svdvalsJAX 2.08xbatched_svdvalsJAX 2.09xbatched_svdvalsJAX 2.11xbatched_svdvalsJAX 2.15xbatched_lutrace 2.16xJAX 1.53xdynamic_slicetrace 2.18xJAX 1.26xbatched_solvetrace 2.20xJAX 5.00xbatched_svdvalsJAX 2.23xconcatenatetrace 2.23xJAX 1.21xbatched_lu_factorJAX 2.25xreversetrace 2.25xJAX 1.50xreversetrace 2.26xJAX 1.28xlayer_normeager 2.33xJAX 3.86xbatched_lu_solveJAX 2.34xbatched_lu_factorJAX 2.38xlayer_normeager 2.38xJAX 4.03xpowtrace 2.40xJAX 3.66xgelueager 2.41xJAX 1.36xsoftpluseager 2.41xJAX 3.68xbatched_solvetrace 2.42xJAX 4.51xscattertrace 2.43xJAX 0.87xsoftpluseager 2.45xJAX 2.74xbatched_eigvalshtrace 2.47xJAX 2.44xbatched_solvetrace 2.48xJAX 4.00xbatched_lutrace 2.49xJAX 4.29xpowtrace 2.49xJAX 3.27xdettrace 2.51xJAX 2.19xbatched_qrtrace 2.51xJAX 2.52xeigvalstrace 2.52xJAX 1.21xbatched_eigvalshtrace 2.54xJAX 3.54xnorm_frotrace 2.56xJAX 7.00xbatched_solvetrace 2.59xJAX 5.13xreduce_max_axis1trace 2.61xJAX 2.77xerfeager 2.62xJAX 1.08xsigmoideager 2.70xJAX 1.65xdettrace 2.71xJAX 1.88xsilueager 2.72xJAX 1.74xreduce_sum_axis1trace 2.73xJAX 1.48xsoftmaxeager 2.76xJAX 5.75xlog_softmaxeager 2.79xJAX 7.24xtriangular_solvetrace 2.79xJAX 4.70xreduce_min_axis1trace 2.94xJAX 2.01xsoftmaxeager 3.01xJAX 7.44xcast_f64_f32trace 3.10xJAX 1.27xgelu_tanheager 3.22xJAX 1.28xgelu_tanheager 3.27xJAX 1.31xbatched_svdvalsJAX 3.33xlayer_normeager 3.37xJAX 5.58xbatched_qrtrace 3.40xJAX 5.22xbatched_lu_factorJAX 3.53xsoftpluseager 3.54xJAX 1.70xcast_f64_f32trace 3.57xJAX 1.72xnorm_frotrace 3.58xJAX 1.02xcholeskytrace 3.59xJAX 2.70xbatched_svdvalsJAX 3.60xbatched_svdvalsJAX 3.61xbatched_svdvalsJAX 3.63xbatched_lu_factorJAX 3.67xsoftmaxeager 3.70xJAX 9.39xbatched_lu_factorJAX 3.72xbatched_svdvalsJAX 3.72xerfeager 3.72xJAX 0.65xerfeager 3.78xJAX 1.41xreduce_prod_axis1trace 3.79xJAX 1.63xsoftpluseager 3.83xJAX 1.71xcast_f64_f32trace 3.84xJAX 1.70xsilueager 3.90xJAX 2.01xreduce_sum_axis1trace 3.91xJAX 1.78xbatched_qrtrace 3.91xJAX 2.05xtake_along_axiseager 4.00xJAX 0.38xnorm_frotrace 4.02xJAX 1.39xsigmoideager 4.03xJAX 1.91xtake_along_axiseager 4.05xJAX 0.48xsoftmaxeager 4.13xJAX 7.15xlog_softmaxeager 4.14xJAX 10.92xreduce_max_axis1trace 4.15xJAX 2.79xhouseholder_qreager 4.18xJAX 3.20xbatched_svdtrace 4.28xJAX 5.78xbatched_solvetrace 4.33xJAX 2.16xtake_along_axiseager 4.34xJAX 0.47xbatched_lu_factorJAX 4.35xtake_along_axiseager 4.36xJAX 0.50xbatched_lutrace 4.46xJAX 2.20xbatched_lu_factorJAX 4.48xbatched_lu_factorJAX 4.59xbatched_lu_solveJAX 4.60xbatched_lu_solveJAX 4.62xtriangular_solvetrace 4.63xJAX 10.54xlog_softmaxeager 4.67xJAX 8.66xtriangular_solvetrace 4.74xJAX 8.17xlog_softmaxeager 4.79xJAX 6.61xbatched_lu_solveJAX 4.86xreduce_prod_axis1trace 5.06xJAX 1.40xbatched_eightrace 5.19xJAX 1.84xhouseholder_qreager 5.27xJAX 2.74xbatched_solvetrace 5.43xJAX 2.69xbatched_lu_solveJAX 5.45xbatched_lu_solveJAX 5.48xbatched_lutrace 5.61xJAX 2.22xbatched_lu_solveJAX 5.67xbatched_lu_factorJAX 5.79xrms_normeager 5.94xJAX 2.68xbatched_solvetrace 6.06xJAX 2.04xbatched_lutrace 6.26xJAX 1.17xlstsqJAX 6.54xbatched_lutrace 6.58xJAX 1.45xeigvalshtrace 6.69xJAX 2.76xgelu_tanheager 7.14xJAX 2.10xbatched_solvetrace 7.38xJAX 1.12xsilueager 7.53xJAX 1.73xbatched_solvetrace 7.53xJAX 1.06xbatched_lu_factorJAX 7.58xbatched_lutrace 8.24xJAX 1.66xbatched_lutrace 8.79xJAX 2.15xsigmoideager 8.98xJAX 1.51xbatched_triangular_solvetrace 9.01xJAX 1.58xrms_normeager 9.43xJAX 3.03xbatched_eightrace 9.47xJAX 2.12xbatched_lutrace 9.74xJAX 2.16xbatched_eigvalshtrace 10.58xJAX 6.68xgelueager 10.76xJAX 1.31xbatched_lutrace 10.93xJAX 2.78xscattertrace 11.18xJAX 1.47xscattertrace 11.59xJAX 0.95xslogdettrace 11.80xJAX 1.94xbatched_solvetrace 12.20xJAX 4.10xbatched_triangular_solvetrace 12.45xJAX 2.04xbatched_triangular_solvetrace 12.46xJAX 1.43xbatched_triangular_solvetrace 12.59xJAX 1.32xbatched_triangular_solvetrace 12.86xJAX 1.49xtake_along_axiseager 13.16xJAX 0.62xtake_along_axiseager 14.30xJAX 0.66xbatched_choleskytrace 14.48xJAX 1.90xbatched_solvetrace 15.48xJAX 3.32xlayer_normeager 15.95xJAX 3.71xrms_normeager 16.09xJAX 3.18xgathertrace 16.81xJAX 1.17xbatched_triangular_solvetrace 18.98xJAX 2.62xreduce_meaneager 24.65xJAX 7.78xgathertrace 44.45xJAX 2.40xreduce_meaneager 63.10xJAX 3.71xreduce_meaneager 67.01xJAX 9.24x
Cases without a comparable baseline (60)

add_into, add_into, add_into, broadcast_in_dim_view, broadcast_in_dim_view, broadcast_in_dim_view, conj_into, conj_into, conj_into, copy_read_into, copy_read_into, copy_read_into, div_into, div_into, div_into, dot_general_read_into, dot_general_read_into, dot_general_read_into, dot_general_read_into_accum, dot_general_read_into_accum, dot_general_read_into_accum, mul_into, mul_into, mul_into, neg_into, neg_into, neg_into, rank_revealing_qr, rank_revealing_qr, rank_revealing_qr, rank_revealing_qr, rank_revealing_qr, rank_revealing_qr, rank_revealing_qr_deficient, rank_revealing_qr_deficient, rank_revealing_qr_deficient, rank_revealing_qr_deficient, rank_revealing_qr_deficient, rank_revealing_qr_deficient, reduce_sum_squares_axis0, reduce_sum_squares_axis0, reduce_sum_squares_axis0, reshape_view, reshape_view, reshape_view, slice_view, slice_view, slice_view, solve_read_into, solve_read_into, solve_read_into, solve_read_into, solve_read_into, solve_read_into, sub_into, sub_into, sub_into, transpose_view, transpose_view, transpose_view

Diagnostic reports (4)

zen5-cpu