Skip to content

mtl5_bench capability card -- ryzen-7-8845hs

The platform benchmark’s capability cards for the ryzen-7-8845hs machine described on the benchmark systems page: every operation family, measured against this machine’s FMA peak and memory bandwidth (#721). This page shows the latest published session; every session’s raw results stay in the data directory.

Repeatability: 590 of 666 entries within +-5% of the previous session, measured after a cold boot on the same commit (median ratio 1.000 to 1.005 per run; benchmarks/compare_sessions.py). The background_activity records show a Defender security-intelligence update at 03:52Z during this session’s native-fast-t1 run; no Defender scan ran in either session. Relates to #721 and #737.

session20261112T035709Z (UTC)
commitcd6762a6e989
classmedium
runsblas at 1 thread, blas at 8 threads, native-fast at 1 thread, native-fast at 8 threads, native at 1 thread, native at 8 threads
failed checks0
CPUs0,2,4,6,8,10,12,14 (pinned)
clock policypower scheme copy of Balanced with processor state 100-100% and EPP 0, boost aggressive; AC (governor mtl5_performance_-_temporary_copy_of_Balanced)
raw databenchmarks/data/ryzen-7-8845hs/mtl5_bench/20261112T035709Z/
earlier sessions20261112T033940Z

Each column is one run of this session: the build variant and its thread count (1T is one thread). The runs follow below, each with its full entry table.

areanative 1Tnative 8Tnative-fast 1Tnative-fast 8Tblas 1Tblas 8T
blas125%97%107%111%25%93%
blas252%68%69%57%77%101%
blas314%2%17%4%68%40%
factor3%1%3%1%13%2%
eigen4%0%4%0%17%2%
elementwise8%8%8%8%7%7%
sparse25%40%27%40%21%34%
direct0%0%0%0%0%0%
orderingn/an/an/an/an/an/a
krylovn/an/an/an/an/an/a
pcn/an/an/an/an/an/a
smoother35%37%32%37%35%37%
mg34%35%34%35%34%35%
sparse_eign/an/an/an/an/an/a
mixed14%13%69%70%11%11%
integern/an/an/an/an/an/a

The median time of each of the 111 entries in each run. The fastest run of each entry is in bold; — is an entry the run did not measure.

areaoptypeshapenative 1Tnative 8Tnative-fast 1Tnative-fast 8Tblas 1Tblas 8T
blas1dotf64n=83886089.796 ms2.290 ms2.382 ms2.266 ms11.491 ms2.309 ms
blas1dotf32n=83886089.748 ms1.347 ms1.182 ms0.991 ms10.308 ms1.359 ms
blas1axpyf64n=838860811.898 ms3.499 ms3.439 ms3.548 ms11.930 ms3.519 ms
blas1axpyf32n=838860811.855 ms1.852 ms1.646 ms1.678 ms10.299 ms1.776 ms
blas1scalef64n=83886082.292 ms2.114 ms1.903 ms2.134 ms2.287 ms2.084 ms
blas1two_normf64n=83886089.326 ms1.298 ms1.165 ms1.046 ms11.013 ms1.518 ms
blas1one_normf64n=83886085.063 ms5.062 ms5.065 ms5.292 ms5.064 ms5.069 ms
blas1infinity_normf64n=83886085.074 ms5.072 ms5.113 ms5.098 ms5.072 ms5.100 ms
blas1sumf64n=83886085.064 ms5.061 ms5.075 ms5.143 ms5.062 ms5.079 ms
blas1dot_k3f64n=838860887.618 ms87.663 ms91.203 ms97.716 ms84.572 ms86.046 ms
blas2gemvf64n=40963.184 ms2.367 ms3.346 ms2.347 ms3.219 ms2.550 ms
blas2gemvf32n=40962.445 ms1.138 ms1.706 ms1.095 ms1.567 ms1.156 ms
blas2gemv_colmajorf64n=40966.646 ms2.312 ms21.609 ms9.513 ms3.374 ms2.695 ms
blas2gemv_colmajorf32n=40962.256 ms1.157 ms7.868 ms4.853 ms1.502 ms1.251 ms
blas2symvf64n=40968.555 ms8.694 ms8.705 ms9.195 ms1.826 ms1.215 ms
blas2trmvf64n=40961.855 ms1.885 ms1.824 ms2.167 ms1.626 ms1.218 ms
blas2trmvf32n=40961.543 ms1.561 ms0.944 ms1.121 ms0.847 ms0.530 ms
blas2trsvf64n=40961.772 ms1.857 ms1.845 ms2.149 ms1.614 ms1.589 ms
blas2trsvf32n=40961.573 ms1.594 ms0.901 ms1.101 ms0.908 ms0.946 ms
blas2gerf64n=409613.543 ms13.492 ms13.538 ms17.212 ms13.588 ms13.670 ms
blas3gemmf64512x512x51216.376 ms2.468 ms10.978 ms3.236 ms3.629 ms0.605 ms
blas3gemmf32512x512x5128.239 ms1.179 ms5.565 ms1.822 ms1.913 ms0.340 ms
blas3gemmtf64512x512x51211.224 ms11.311 ms7.415 ms5.034 ms2.726 ms1.356 ms
blas3symmf64512x512x51217.706 ms17.614 ms11.973 ms3.734 ms3.806 ms0.607 ms
blas3syrkf64512x512x51213.574 ms13.666 ms13.566 ms14.165 ms3.039 ms1.720 ms
blas3syr2kf64512x512x51223.489 ms23.728 ms23.003 ms24.116 ms4.897 ms2.387 ms
blas3trmmf64512x512x51225.087 ms25.188 ms9.725 ms5.917 ms1.888 ms0.398 ms
blas3trsmf64512x512x51226.572 ms26.898 ms9.845 ms6.195 ms2.842 ms0.557 ms
blas3trsmf32512x512x51225.995 ms26.112 ms6.574 ms4.605 ms1.346 ms0.266 ms
blas3kronf6448x48 (x) 48x489.015 ms9.303 ms9.083 ms9.264 ms9.306 ms9.245 ms
factorluf64512x51232.730 ms23.656 ms32.813 ms23.200 ms4.986 ms9.104 ms
factorluf32512x51230.913 ms22.075 ms31.412 ms22.390 ms2.443 ms4.345 ms
factorcholeskyf64512x51211.394 ms11.374 ms11.382 ms11.397 ms4.350 ms5.960 ms
factorcholeskyf32512x51210.910 ms10.898 ms10.904 ms10.906 ms1.676 ms2.308 ms
factorldltf64512x51211.450 ms11.445 ms11.433 ms11.437 ms11.785 ms11.867 ms
factorldlt_bkf64512x512166.463 ms165.070 ms166.832 ms165.245 ms164.303 ms167.173 ms
factorqrf64512x512645.572 ms397.421 ms645.997 ms395.043 ms10.166 ms11.893 ms
factorlqf64512x51264.390 ms44.839 ms68.522 ms44.608 ms64.507 ms45.484 ms
factorhessenbergf64512x512740.793 ms470.812 ms752.540 ms470.760 ms745.299 ms477.079 ms
factorhouseholderf64512x5123.918 ms1.002 ms4.000 ms1.021 ms3.878 ms1.004 ms
factorinvf64512x51288.836 ms79.564 ms93.824 ms79.181 ms60.512 ms65.954 ms
factorlu_iterative_refinef32>f64512x51233.800 ms24.624 ms34.158 ms25.187 ms5.538 ms7.245 ms
eigeneigenvalue_symmetricf64256x25638.616 ms39.148 ms37.521 ms33.577 ms1.678 ms3.588 ms
eigeneigen_symmetricf64256x256126.776 ms128.315 ms127.218 ms123.975 ms129.208 ms133.014 ms
eigeneigen_symmetric_jacobif64256x256520.252 ms512.298 ms512.542 ms496.530 ms499.815 ms519.779 ms
eigeneigenvaluef64256x25657.420 ms60.275 ms54.014 ms60.808 ms16.673 ms21.078 ms
eigenschurf64256x256110.628 ms113.515 ms109.018 ms112.942 ms21.833 ms26.127 ms
eigensvdf64256x256117.948 ms119.042 ms118.388 ms118.691 ms8.649 ms12.133 ms
eigenexpmf64256x25630.289 ms19.867 ms26.294 ms19.380 ms15.903 ms18.256 ms
eigensylvesterf64256x256261.271 ms259.695 ms259.941 ms259.498 ms72.035 ms80.883 ms
elementwiseexpf64n=838860827.455 ms32.543 ms27.353 ms27.896 ms27.847 ms28.560 ms
elementwiselogf64n=838860841.556 ms42.263 ms42.478 ms42.556 ms43.053 ms43.937 ms
elementwisesinf64n=838860838.614 ms38.782 ms39.362 ms39.823 ms38.927 ms39.470 ms
elementwisesqrtf64n=838860821.302 ms21.317 ms21.114 ms21.289 ms21.365 ms21.387 ms
elementwiseerff64n=838860832.112 ms31.949 ms31.732 ms31.967 ms36.179 ms34.953 ms
sparsespmvf64512^2 grid, nnz 13086720.832 ms0.202 ms0.851 ms0.205 ms2.424 ms0.198 ms
sparsespmv_transf64512^2 grid, nnz 13086721.381 ms1.269 ms1.324 ms1.289 ms1.370 ms1.516 ms
sparsecrs_densef64512^2 grid, nnz 1308672 x 166.517 ms2.260 ms6.177 ms2.288 ms6.594 ms2.320 ms
sparsespgemmf64256^2 grid, nnz 32665618.294 ms18.221 ms18.036 ms18.861 ms17.840 ms18.291 ms
sparselower_solve_levelf64192^2 grid, nnz(L) 10035641.786 ms1.689 ms1.815 ms1.753 ms1.752 ms1.705 ms
directsparse_choleskyf6496^2 grid, nnz 45696104.848 ms110.628 ms99.470 ms98.618 ms103.386 ms101.545 ms
directsparse_ldltf6496^2 grid, nnz 4569631.221 ms31.983 ms29.388 ms29.054 ms30.551 ms31.941 ms
directsupernodal_ldltf6496^2 grid, nnz 4569618.305 ms17.916 ms16.270 ms15.885 ms17.140 ms18.977 ms
directsparse_luf6496^2 grid, nnz 4569654.321 ms54.741 ms54.953 ms55.213 ms54.899 ms54.848 ms
directsupernodal_luf6496^2 grid, nnz 4569673.608 ms73.319 ms74.370 ms72.275 ms73.333 ms74.254 ms
directnative_kluf6496^2 grid, nnz 4569625.380 ms26.601 ms26.666 ms25.865 ms25.679 ms26.133 ms
directsparse_qrf6448^2 grid, nnz 1132875.715 ms76.275 ms84.782 ms83.988 ms79.417 ms80.579 ms
orderingamdf6496^2 grid, nnz 45696, scrambled3.599 ms3.699 ms3.675 ms3.703 ms3.651 ms3.723 ms
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled5.528 ms5.590 ms5.691 ms5.668 ms5.609 ms5.692 ms
orderingrcmf6496^2 grid, nnz 45696, scrambled0.419 ms0.433 ms0.445 ms0.444 ms0.427 ms0.422 ms
orderingblock_triangular_formf6432 planted blocks, n 819214.074 ms14.011 ms15.474 ms14.117 ms14.147 ms14.111 ms
krylovcg+ic0f64128^2 grid, n 1638439.256 ms39.203 ms33.757 ms32.635 ms40.253 ms40.303 ms
krylovminres+diagf64128^2 grid, n 1638443.754 ms43.675 ms31.894 ms32.034 ms46.171 ms46.739 ms
krylovbicg+ilu0f64128^2 grid, n 1638430.510 ms30.456 ms28.291 ms27.434 ms31.072 ms31.310 ms
krylovbicgstab+ilu0f64128^2 grid, n 1638416.668 ms16.644 ms13.784 ms13.466 ms17.261 ms17.299 ms
krylovbicgstab_ell+ilu0f64128^2 grid, n 1638421.684 ms21.608 ms18.055 ms17.877 ms22.413 ms22.467 ms
krylovcgs+ilu0f64128^2 grid, n 1638417.593 ms17.556 ms15.877 ms15.724 ms17.946 ms17.965 ms
krylovgmres30+ilu0f64128^2 grid, n 1638450.784 ms50.802 ms32.118 ms31.046 ms54.792 ms54.976 ms
krylovqmr+ilu0f64128^2 grid, n 1638433.841 ms33.736 ms29.495 ms28.980 ms34.736 ms34.869 ms
krylovtfqmr+ilu0f64128^2 grid, n 1638419.777 ms19.742 ms17.583 ms17.304 ms20.221 ms20.462 ms
krylovidr_s1+ilu0f64128^2 grid, n 1638423.595 ms23.531 ms19.592 ms19.322 ms24.339 ms24.452 ms
krylovidr_s2+ilu0f64128^2 grid, n 1638422.770 ms22.697 ms19.357 ms18.431 ms23.512 ms23.691 ms
krylovidr_s4+ilu0f64128^2 grid, n 1638425.309 ms25.289 ms19.720 ms19.587 ms26.343 ms26.532 ms
krylovidr_s8+ilu0f64128^2 grid, n 1638430.987 ms31.172 ms22.163 ms22.431 ms32.786 ms33.729 ms
pcdiagonalf64128^2 grid, rows scaled0.036 ms0.036 ms0.036 ms0.036 ms0.036 ms0.036 ms
pcilu_0f64128^2 grid2.383 ms2.437 ms2.403 ms2.433 ms2.401 ms2.535 ms
pcilutf64128^2 grid23.169 ms23.199 ms23.132 ms23.180 ms22.279 ms22.782 ms
pcic_0f64128^2 grid1.814 ms1.827 ms1.833 ms1.828 ms1.839 ms1.870 ms
pcildlf64128^2 grid1.734 ms1.752 ms1.761 ms1.758 ms1.764 ms1.773 ms
pcssorf64128^2 grid0.021 ms0.021 ms0.021 ms0.022 ms0.021 ms0.022 ms
pcblock_diagonal4f64128^2 grid, rows scaled0.663 ms0.661 ms0.665 ms0.664 ms1.321 ms1.329 ms
smootherjacobif64512^2 grid, nnz 13086721.585 ms1.456 ms1.736 ms1.467 ms1.588 ms1.461 ms
smoothergauss_seidelf64512^2 grid, nnz 13086721.493 ms1.494 ms1.501 ms1.495 ms1.540 ms1.499 ms
smoothersor1.5f64512^2 grid, nnz 13086721.780 ms1.768 ms1.818 ms1.783 ms1.791 ms1.764 ms
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)15.830 ms15.505 ms16.074 ms15.564 ms15.984 ms15.486 ms
sparse_eiglanczosf6464x65 Laplacian, k=4, m=8032.230 ms32.329 ms32.399 ms32.454 ms32.228 ms32.459 ms
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=8046.726 ms46.791 ms47.014 ms46.887 ms32.894 ms33.106 ms
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=4056.665 ms53.987 ms55.463 ms54.576 ms57.155 ms58.023 ms
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps7.733 ms7.712 ms6.453 ms6.519 ms7.998 ms8.046 ms
mixeddotf32>f64n=83886089.321 ms9.354 ms1.840 ms1.786 ms11.424 ms11.440 ms
mixedgemmf32>f64512x512x51229.049 ms4.032 ms15.320 ms2.635 ms29.514 ms5.023 ms
mixedblock_scaled_doti8+f32n=8388608, block 321.377 ms1.403 ms0.476 ms0.475 ms1.478 ms1.506 ms
integerdoti32n=8388608, |v| <= 153.123 ms3.123 ms1.202 ms1.229 ms3.132 ms3.136 ms
integerdoti16>i32n=8388608, |v| <= 152.879 ms2.877 ms0.463 ms0.473 ms2.878 ms2.893 ms
integerdoti8>i32n=8388608, |v| <= 151.382 ms1.399 ms0.347 ms0.346 ms1.394 ms1.408 ms
integerdotu8i8>i32n=8388608, |v| <= 151.381 ms1.399 ms0.318 ms0.318 ms1.394 ms1.409 ms
integergemmi32512^3, |v| <= 20478.087 ms1.181 ms14.928 ms2.438 ms8.394 ms1.281 ms
integergemmi16>i32512^3, |v| <= 204730.291 ms4.176 ms17.341 ms2.544 ms30.301 ms4.212 ms
integergemmi8>i32512^3, |v| <= 12730.562 ms4.187 ms17.723 ms2.569 ms30.554 ms4.241 ms
integergemm_quadi8>i32512^3, |v| <= 12730.551 ms4.193 ms3.711 ms0.674 ms30.558 ms4.240 ms
integergemm_quadu8i8>i32512^3, |v| <= 12727.274 ms3.800 ms3.688 ms0.672 ms27.237 ms3.859 ms

ryzen-7-8845hs — native, 1 thread, class medium

Section titled “ryzen-7-8845hs — native, 1 thread, class medium”
CPUAMD Ryzen 7 8845HS w/ Radeon 780M Graphics (x86_64, 16 logical cores)
OSWindows 10.0.26300 (26H2)
compilerMSVC 19.52.37029
build/DWIN32 /D_WINDOWS /EHsc /O2 /Ob2 /DNDEBUG, SIMD scalar
commitcd6762a6e989
FMA peak (ceiling)f64 80.1 GFLOP/s, f32 160.2 GFLOP/s
… hardware, AVX-512Ff64 80.1, f32 160.2 GFLOP/s
… this build, scalarf64 1.9, f32 1.9 GFLOP/s
bandwidthread 43.3, update 53.3, triad 39.0 GB/s (ceiling 53.3)
last-level cache16 MiB
areaentriesmedian %roofbest %rooffailed checks
blas11025%110%0
blas21052%79%0
blas31014%20%0
factor123%5%0
eigen84%10%0
elementwise58%12%0
sparse525%61%0
direct70%1%0
ordering4n/an/a0
krylov13n/an/a0
pc7n/an/a0
smoother335%37%0
mg134%34%0
sparse_eig4n/an/a0
mixed314%24%0
integer9n/an/a0
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheck
blas1dotf64n=83886089.796 ms1.7113.7026%ok
blas1dotf32n=83886089.748 ms1.726.8813%ok
blas1axpyf64n=838860811.898 ms1.4116.9232%ok
blas1axpyf32n=838860811.855 ms1.428.4916%ok
blas1scalef64n=83886082.292 ms3.6658.56110%ok
blas1two_normf64n=83886089.326 ms1.807.2014%ok
blas1one_normf64n=83886085.063 ms1.6613.2525%ok
blas1infinity_normf64n=83886085.074 ms1.6513.2325%ok
blas1sumf64n=83886085.064 ms1.6613.2525%ok
blas1dot_k3f64n=838860887.618 ms2.111.533%ok
blas2gemvf64n=40963.184 ms10.5442.1779%ok
blas2gemvf32n=40962.445 ms13.7327.4752%ok
blas2gemv_colmajorf64n=40966.646 ms5.0520.2138%ok
blas2gemv_colmajorf32n=40962.256 ms14.8729.7656%ok
blas2symvf64n=40968.555 ms3.927.8515%ok
blas2trmvf64n=40961.855 ms9.0436.2168%ok
blas2trmvf32n=40961.543 ms10.8821.7741%ok
blas2trsvf64n=40961.772 ms9.4737.9071%ok
blas2trsvf32n=40961.573 ms10.6721.3640%ok
blas2gerf64n=409613.543 ms2.4819.8337%ok
blas3gemmf64512x512x51216.376 ms16.390.3820%ok
blas3gemmf32512x512x5128.239 ms32.580.3820%ok
blas3gemmtf64512x512x51211.224 ms11.960.5615%ok
blas3symmf64512x512x51217.706 ms15.160.3619%ok
blas3syrkf64512x512x51213.574 ms9.890.3112%ok
blas3syr2kf64512x512x51223.489 ms11.430.2714%ok
blas3trmmf64512x512x51225.087 ms5.350.217%ok
blas3trsmf64512x512x51226.572 ms5.050.206%ok
blas3trsmf32512x512x51225.995 ms5.160.103%ok
blas3kronf6448x48 (x) 48x489.015 ms0.594.719%ok
factorluf64512x51232.730 ms2.730.133%ok
factorluf32512x51230.913 ms2.890.072%ok
factorcholeskyf64512x51211.394 ms3.930.375%ok
factorcholeskyf32512x51210.910 ms4.100.193%ok
factorldltf64512x51211.450 ms3.910.375%ok
factorldlt_bkf64512x512166.463 ms0.270.030%ok
factorqrf64512x512645.572 ms0.280.010%ok
factorlqf64512x51264.390 ms2.780.073%ok
factorhessenbergf64512x512740.793 ms0.600.011%ok
factorhouseholderf64512x5123.918 ms0.271.072%*ok
factorinvf64512x51288.836 ms3.020.054%ok
factorlu_iterative_refinef32>f64512x51233.800 ms2.650.123%ok
eigeneigenvalue_symmetricf64256x25638.616 ms0.580.031%ok
eigeneigen_symmetricf64256x256126.776 ms1.190.011%ok
eigeneigen_symmetric_jacobif64256x256520.252 ms0.290.000%ok
eigeneigenvaluef64256x25657.420 ms2.920.024%ok
eigenschurf64256x256110.628 ms3.790.015%ok
eigensvdf64256x256117.948 ms3.130.014%ok
eigenexpmf64256x25630.289 ms8.310.0310%ok
eigensylvesterf64256x256261.271 ms3.850.015%ok
elementwiseexpf64n=838860827.455 ms0.314.899%ok
elementwiselogf64n=838860841.556 ms0.203.236%ok
elementwisesinf64n=838860838.614 ms0.223.487%ok
elementwisesqrtf64n=838860821.302 ms0.396.3012%ok
elementwiseerff64n=838860832.112 ms0.264.188%ok
sparsespmvf64512^2 grid, nnz 13086720.832 ms3.1532.7461%ok
sparsespmv_transf64512^2 grid, nnz 13086721.381 ms1.9019.7237%ok
sparsecrs_densef64512^2 grid, nnz 1308672 x 166.517 ms6.4313.5125%ok
sparsespgemmf64256^2 grid, nnz 32665618.294 ms0.180.631%*ok
sparselower_solve_levelf64192^2 grid, nnz(L) 10035641.786 ms1.129.3217%*ok
directsparse_choleskyf6496^2 grid, nnz 45696104.848 ms0.110.010%ok
directsparse_ldltf6496^2 grid, nnz 4569631.221 ms0.360.020%ok
directsupernodal_ldltf6496^2 grid, nnz 4569618.305 ms0.610.041%ok
directsparse_luf6496^2 grid, nnz 4569654.321 ms0.410.011%ok
directsupernodal_luf6496^2 grid, nnz 4569673.608 ms0.300.010%ok
directnative_kluf6496^2 grid, nnz 4569625.380 ms0.880.031%ok
directsparse_qrf6448^2 grid, nnz 1132875.715 ms0.030.000%ok
orderingamdf6496^2 grid, nnz 45696, scrambled3.599 ms0.000.00n/aok
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled5.528 ms0.000.00n/aok
orderingrcmf6496^2 grid, nnz 45696, scrambled0.419 ms0.000.00n/aok
orderingblock_triangular_formf6432 planted blocks, n 819214.074 ms0.000.00n/aok
krylovcg+ic0f64128^2 grid, n 1638439.256 ms0.000.00n/aok
krylovminres+diagf64128^2 grid, n 1638443.754 ms0.000.00n/aok
krylovbicg+ilu0f64128^2 grid, n 1638430.510 ms0.000.00n/aok
krylovbicgstab+ilu0f64128^2 grid, n 1638416.668 ms0.000.00n/aok
krylovbicgstab_ell+ilu0f64128^2 grid, n 1638421.684 ms0.000.00n/aok
krylovcgs+ilu0f64128^2 grid, n 1638417.593 ms0.000.00n/aok
krylovgmres30+ilu0f64128^2 grid, n 1638450.784 ms0.000.00n/aok
krylovqmr+ilu0f64128^2 grid, n 1638433.841 ms0.000.00n/aok
krylovtfqmr+ilu0f64128^2 grid, n 1638419.777 ms0.000.00n/aok
krylovidr_s1+ilu0f64128^2 grid, n 1638423.595 ms0.000.00n/aok
krylovidr_s2+ilu0f64128^2 grid, n 1638422.770 ms0.000.00n/aok
krylovidr_s4+ilu0f64128^2 grid, n 1638425.309 ms0.000.00n/aok
krylovidr_s8+ilu0f64128^2 grid, n 1638430.987 ms0.000.00n/aok
pcdiagonalf64128^2 grid, rows scaled0.036 ms0.000.00n/aok
pcilu_0f64128^2 grid2.383 ms0.000.00n/aok
pcilutf64128^2 grid23.169 ms0.000.00n/aok
pcic_0f64128^2 grid1.814 ms0.000.00n/aok
pcildlf64128^2 grid1.734 ms0.000.00n/aok
pcssorf64128^2 grid0.021 ms0.000.00n/aok
pcblock_diagonal4f64128^2 grid, rows scaled0.663 ms0.000.00n/aok
smootherjacobif64512^2 grid, nnz 13086721.585 ms1.6518.5035%ok
smoothergauss_seidelf64512^2 grid, nnz 13086721.493 ms1.7519.6437%ok
smoothersor1.5f64512^2 grid, nnz 13086721.780 ms1.4716.4731%ok
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)15.830 ms2.4318.3634%ok
sparse_eiglanczosf6464x65 Laplacian, k=4, m=8032.230 ms0.000.00n/aok
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=8046.726 ms0.000.00n/aok
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=4056.665 ms0.000.00n/aok
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps7.733 ms0.000.00n/aok
mixeddotf32>f64n=83886089.321 ms1.807.2014%ok
mixedgemmf32>f64512x512x51229.049 ms9.240.1412%ok
mixedblock_scaled_doti8+f32n=8388608, block 321.377 ms12.1912.9524%ok
integerdoti32n=8388608, |v| <= 153.123 ms5.3721.49n/aok
integerdoti16>i32n=8388608, |v| <= 152.879 ms5.8311.66n/aok
integerdoti8>i32n=8388608, |v| <= 151.382 ms12.1412.14n/aok
integerdotu8i8>i32n=8388608, |v| <= 151.381 ms12.1512.15n/aok
integergemmi32512^3, |v| <= 20478.087 ms33.190.39n/aok
integergemmi16>i32512^3, |v| <= 204730.291 ms8.860.07n/aok
integergemmi8>i32512^3, |v| <= 12730.562 ms8.780.05n/aok
integergemm_quadi8>i32512^3, |v| <= 12730.551 ms8.790.05n/aok
integergemm_quadu8i8>i32512^3, |v| <= 12727.274 ms9.840.06n/aok

* working set fits in the last-level cache: a cache rate, shown as a % of the DRAM roofline.

ryzen-7-8845hs — native, 8 threads, class medium

Section titled “ryzen-7-8845hs — native, 8 threads, class medium”
CPUAMD Ryzen 7 8845HS w/ Radeon 780M Graphics (x86_64, 16 logical cores)
OSWindows 10.0.26300 (26H2)
compilerMSVC 19.52.37029
build/DWIN32 /D_WINDOWS /EHsc /O2 /Ob2 /DNDEBUG, SIMD scalar
commitcd6762a6e989
FMA peak (ceiling)f64 629.0 GFLOP/s, f32 1257.0 GFLOP/s
… hardware, AVX-512Ff64 629.0, f32 1257.0 GFLOP/s
… this build, scalarf64 14.4, f32 14.4 GFLOP/s
bandwidthread 53.0, update 53.5, triad 41.3 GB/s (ceiling 53.5)
last-level cache16 MiB
areaentriesmedian %roofbest %rooffailed checksmedian parallel efficiency
blas11097%119%043%
blas21068%110%013%
blas3102%18%012%
factor121%8%017%
eigen80%2%012%
elementwise58%12%012%
sparse540%252%014%
direct70%0%012%
ordering4n/an/a012%
krylov13n/an/a013%
pc7n/an/a012%
smoother337%38%013%
mg135%35%013%
sparse_eig4n/an/a013%
mixed313%24%012%
integer9n/an/a086%
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheckspeedup (efficiency)
blas1dotf64n=83886082.290 ms7.3358.61110%ok4.3x (53%)
blas1dotf32n=83886081.347 ms12.4549.8193%ok7.2x (90%)
blas1axpyf64n=83886083.499 ms4.8057.54108%ok3.4x (43%)
blas1axpyf32n=83886081.852 ms9.0654.35102%ok6.4x (80%)
blas1scalef64n=83886082.114 ms3.9763.48119%ok1.1x (14%)
blas1two_normf64n=83886081.298 ms12.9251.7097%ok7.2x (90%)
blas1one_normf64n=83886085.062 ms1.6613.2625%ok1.0x (13%)
blas1infinity_normf64n=83886085.072 ms1.6513.2325%ok1.0x (13%)
blas1sumf64n=83886085.061 ms1.6613.2625%ok1.0x (13%)
blas1dot_k3f64n=838860887.663 ms2.111.533%ok1.0x (12%)
blas2gemvf64n=40962.367 ms14.1856.73106%ok1.3x (17%)
blas2gemvf32n=40961.138 ms29.4858.99110%ok2.1x (27%)
blas2gemv_colmajorf64n=40962.312 ms14.5158.07109%ok2.9x (36%)
blas2gemv_colmajorf32n=40961.157 ms29.0058.02108%ok1.9x (24%)
blas2symvf64n=40968.694 ms3.867.7314%ok1.0x (12%)
blas2trmvf64n=40961.885 ms8.9035.6367%ok1.0x (12%)
blas2trmvf32n=40961.561 ms10.7521.5240%ok1.0x (12%)
blas2trsvf64n=40961.857 ms9.0436.1868%ok1.0x (12%)
blas2trsvf32n=40961.594 ms10.5221.0739%ok1.0x (12%)
blas2gerf64n=409613.492 ms2.4919.9037%ok1.0x (13%)
blas3gemmf64512x512x5122.468 ms108.782.5517%ok6.6x (83%)
blas3gemmf32512x512x5121.179 ms227.602.6718%ok7.0x (87%)
blas3gemmtf64512x512x51211.311 ms11.870.562%ok1.0x (12%)
blas3symmf64512x512x51217.614 ms15.240.362%ok1.0x (13%)
blas3syrkf64512x512x51213.666 ms9.820.312%ok1.0x (12%)
blas3syr2kf64512x512x51223.728 ms11.310.272%ok1.0x (12%)
blas3trmmf64512x512x51225.188 ms5.330.211%ok1.0x (12%)
blas3trsmf64512x512x51226.898 ms4.990.191%ok1.0x (12%)
blas3trsmf32512x512x51226.112 ms5.140.100%ok1.0x (12%)
blas3kronf6448x48 (x) 48x489.303 ms0.574.579%ok1.0x (12%)
factorluf64512x51223.656 ms3.780.181%ok1.4x (17%)
factorluf32512x51222.075 ms4.050.100%ok1.4x (18%)
factorcholeskyf64512x51211.374 ms3.930.371%*ok1.0x (13%)
factorcholeskyf32512x51210.898 ms4.110.190%*ok1.0x (13%)
factorldltf64512x51211.445 ms3.910.371%*ok1.0x (13%)
factorldlt_bkf64512x512165.070 ms0.270.030%*ok1.0x (13%)
factorqrf64512x512397.421 ms0.450.010%ok1.6x (20%)
factorlqf64512x51244.839 ms3.990.091%ok1.4x (18%)
factorhessenbergf64512x512470.812 ms0.950.010%ok1.6x (20%)
factorhouseholderf64512x5121.002 ms1.054.198%*ok3.9x (49%)
factorinvf64512x51279.564 ms3.370.051%ok1.1x (14%)
factorlu_iterative_refinef32>f64512x51224.624 ms3.630.171%ok1.4x (17%)
eigeneigenvalue_symmetricf64256x25639.148 ms0.570.030%ok1.0x (12%)
eigeneigen_symmetricf64256x256128.315 ms1.180.010%ok1.0x (12%)
eigeneigen_symmetric_jacobif64256x256512.298 ms0.290.000%ok1.0x (13%)
eigeneigenvaluef64256x25660.275 ms2.780.020%ok1.0x (12%)
eigenschurf64256x256113.515 ms3.690.011%ok1.0x (12%)
eigensvdf64256x256119.042 ms3.100.010%ok1.0x (12%)
eigenexpmf64256x25619.867 ms12.670.052%ok1.5x (19%)
eigensylvesterf64256x256259.695 ms3.880.011%ok1.0x (13%)
elementwiseexpf64n=838860832.543 ms0.264.128%ok0.8x (11%)
elementwiselogf64n=838860842.263 ms0.203.186%ok1.0x (12%)
elementwisesinf64n=838860838.782 ms0.223.466%ok1.0x (12%)
elementwisesqrtf64n=838860821.317 ms0.396.3012%ok1.0x (12%)
elementwiseerff64n=838860831.949 ms0.264.208%ok1.0x (13%)
sparsespmvf64512^2 grid, nnz 13086720.202 ms12.93134.54252% !ok4.1x (51%)
sparsespmv_transf64512^2 grid, nnz 13086721.269 ms2.0621.4740%ok1.1x (14%)
sparsecrs_densef64512^2 grid, nnz 1308672 x 162.260 ms18.5338.9673%ok2.9x (36%)
sparsespgemmf64256^2 grid, nnz 32665618.221 ms0.180.631%*ok1.0x (13%)
sparselower_solve_levelf64192^2 grid, nnz(L) 10035641.689 ms1.199.8618%*ok1.1x (13%)
directsparse_choleskyf6496^2 grid, nnz 45696110.628 ms0.100.010%ok0.9x (12%)
directsparse_ldltf6496^2 grid, nnz 4569631.983 ms0.350.020%ok1.0x (12%)
directsupernodal_ldltf6496^2 grid, nnz 4569617.916 ms0.620.040%ok1.0x (13%)
directsparse_luf6496^2 grid, nnz 4569654.741 ms0.410.010%ok1.0x (12%)
directsupernodal_luf6496^2 grid, nnz 4569673.319 ms0.310.010%ok1.0x (13%)
directnative_kluf6496^2 grid, nnz 4569626.601 ms0.840.030%ok1.0x (12%)
directsparse_qrf6448^2 grid, nnz 1132876.275 ms0.020.000%*ok1.0x (12%)
orderingamdf6496^2 grid, nnz 45696, scrambled3.699 ms0.000.00n/aok1.0x (12%)
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled5.590 ms0.000.00n/aok1.0x (12%)
orderingrcmf6496^2 grid, nnz 45696, scrambled0.433 ms0.000.00n/aok1.0x (12%)
orderingblock_triangular_formf6432 planted blocks, n 819214.011 ms0.000.00n/aok1.0x (13%)
krylovcg+ic0f64128^2 grid, n 1638439.203 ms0.000.00n/aok1.0x (13%)
krylovminres+diagf64128^2 grid, n 1638443.675 ms0.000.00n/aok1.0x (13%)
krylovbicg+ilu0f64128^2 grid, n 1638430.456 ms0.000.00n/aok1.0x (13%)
krylovbicgstab+ilu0f64128^2 grid, n 1638416.644 ms0.000.00n/aok1.0x (13%)
krylovbicgstab_ell+ilu0f64128^2 grid, n 1638421.608 ms0.000.00n/aok1.0x (13%)
krylovcgs+ilu0f64128^2 grid, n 1638417.556 ms0.000.00n/aok1.0x (13%)
krylovgmres30+ilu0f64128^2 grid, n 1638450.802 ms0.000.00n/aok1.0x (12%)
krylovqmr+ilu0f64128^2 grid, n 1638433.736 ms0.000.00n/aok1.0x (13%)
krylovtfqmr+ilu0f64128^2 grid, n 1638419.742 ms0.000.00n/aok1.0x (13%)
krylovidr_s1+ilu0f64128^2 grid, n 1638423.531 ms0.000.00n/aok1.0x (13%)
krylovidr_s2+ilu0f64128^2 grid, n 1638422.697 ms0.000.00n/aok1.0x (13%)
krylovidr_s4+ilu0f64128^2 grid, n 1638425.289 ms0.000.00n/aok1.0x (13%)
krylovidr_s8+ilu0f64128^2 grid, n 1638431.172 ms0.000.00n/aok1.0x (12%)
pcdiagonalf64128^2 grid, rows scaled0.036 ms0.000.00n/aok1.0x (12%)
pcilu_0f64128^2 grid2.437 ms0.000.00n/aok1.0x (12%)
pcilutf64128^2 grid23.199 ms0.000.00n/aok1.0x (12%)
pcic_0f64128^2 grid1.827 ms0.000.00n/aok1.0x (12%)
pcildlf64128^2 grid1.752 ms0.000.00n/aok1.0x (12%)
pcssorf64128^2 grid0.021 ms0.000.00n/aok1.0x (12%)
pcblock_diagonal4f64128^2 grid, rows scaled0.661 ms0.000.00n/aok1.0x (13%)
smootherjacobif64512^2 grid, nnz 13086721.456 ms1.8020.1438%ok1.1x (14%)
smoothergauss_seidelf64512^2 grid, nnz 13086721.494 ms1.7519.6337%ok1.0x (12%)
smoothersor1.5f64512^2 grid, nnz 13086721.768 ms1.4816.5931%ok1.0x (13%)
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)15.505 ms2.4818.7535%ok1.0x (13%)
sparse_eiglanczosf6464x65 Laplacian, k=4, m=8032.329 ms0.000.00n/aok1.0x (12%)
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=8046.791 ms0.000.00n/aok1.0x (12%)
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=4053.987 ms0.000.00n/aok1.0x (13%)
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps7.712 ms0.000.00n/aok1.0x (13%)
mixeddotf32>f64n=83886089.354 ms1.797.1713%ok1.0x (12%)
mixedgemmf32>f64512x512x5124.032 ms66.571.0411%ok7.2x (90%)
mixedblock_scaled_doti8+f32n=8388608, block 321.403 ms11.9612.7124%ok1.0x (12%)
integerdoti32n=8388608, |v| <= 153.123 ms5.3721.49n/aok1.0x (12%)
integerdoti16>i32n=8388608, |v| <= 152.877 ms5.8311.67n/aok1.0x (13%)
integerdoti8>i32n=8388608, |v| <= 151.399 ms12.0012.00n/aok1.0x (12%)
integerdotu8i8>i32n=8388608, |v| <= 151.399 ms11.9911.99n/aok1.0x (12%)
integergemmi32512^3, |v| <= 20471.181 ms227.222.66n/aok6.8x (86%)
integergemmi16>i32512^3, |v| <= 20474.176 ms64.280.50n/aok7.3x (91%)
integergemmi8>i32512^3, |v| <= 1274.187 ms64.110.38n/aok7.3x (91%)
integergemm_quadi8>i32512^3, |v| <= 1274.193 ms64.020.38n/aok7.3x (91%)
integergemm_quadu8i8>i32512^3, |v| <= 1273.800 ms70.630.41n/aok7.2x (90%)

* working set fits in the last-level cache: a cache rate, shown as a % of the DRAM roofline.

! more than 150% of the DRAM roofline for a bandwidth-bound result: the entry’s byte model is probably wrong.

ryzen-7-8845hs — native-fast, 1 thread, class medium

Section titled “ryzen-7-8845hs — native-fast, 1 thread, class medium”
CPUAMD Ryzen 7 8845HS w/ Radeon 780M Graphics (x86_64, 16 logical cores)
OSWindows 10.0.26300 (26H2)
compilerMSVC 19.52.37029
build/DWIN32 /D_WINDOWS /EHsc /O2 /Ob2 /DNDEBUG, SIMD SSE2
commitcd6762a6e989
FMA peak (ceiling)f64 80.1 GFLOP/s, f32 160.1 GFLOP/s
… hardware, AVX-512Ff64 80.1, f32 160.1 GFLOP/s
… this build, SSE2f64 40.0, f32 80.1 GFLOP/s
bandwidthread 47.2, update 52.8, triad 39.0 GB/s (ceiling 52.8)
last-level cache16 MiB
areaentriesmedian %roofbest %rooffailed checks
blas110107%133%0
blas21069%76%0
blas31017%31%0
factor123%5%0
eigen84%12%0
elementwise58%12%0
sparse527%61%0
direct70%1%0
ordering4n/an/a0
krylov13n/an/a0
pc7n/an/a0
smoother332%37%0
mg134%34%0
sparse_eig4n/an/a0
mixed369%71%0
integer9n/an/a0
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheck
blas1dotf64n=83886082.382 ms7.0456.34107%ok
blas1dotf32n=83886081.182 ms14.2056.79107%ok
blas1axpyf64n=83886083.439 ms4.8858.55111%ok
blas1axpyf32n=83886081.646 ms10.1961.17116%ok
blas1scalef64n=83886081.903 ms4.4170.53133%ok
blas1two_normf64n=83886081.165 ms14.4057.59109%ok
blas1one_normf64n=83886085.065 ms1.6613.2525%ok
blas1infinity_normf64n=83886085.113 ms1.6413.1225%ok
blas1sumf64n=83886085.075 ms1.6513.2225%ok
blas1dot_k3f64n=838860891.203 ms2.021.473%ok
blas2gemvf64n=40963.346 ms10.0340.1376%ok
blas2gemvf32n=40961.706 ms19.6739.3774%ok
blas2gemv_colmajorf64n=409621.609 ms1.556.2112%ok
blas2gemv_colmajorf32n=40967.868 ms4.268.5316%ok
blas2symvf64n=40968.705 ms3.857.7215%ok
blas2trmvf64n=40961.824 ms9.2036.8270%ok
blas2trmvf32n=40960.944 ms17.7635.5667%ok
blas2trsvf64n=40961.845 ms9.0936.4069%ok
blas2trsvf32n=40960.901 ms18.6237.2771%ok
blas2gerf64n=409613.538 ms2.4819.8338%ok
blas3gemmf64512x512x51210.978 ms24.450.5731%ok
blas3gemmf32512x512x5125.565 ms48.240.5730%ok
blas3gemmtf64512x512x5127.415 ms18.100.8523%ok
blas3symmf64512x512x51211.973 ms22.420.5328%ok
blas3syrkf64512x512x51213.566 ms9.890.3112%ok
blas3syr2kf64512x512x51223.003 ms11.670.2715%ok
blas3trmmf64512x512x5129.725 ms13.800.5417%ok
blas3trsmf64512x512x5129.845 ms13.630.5317%ok
blas3trsmf32512x512x5126.574 ms20.420.4013%ok
blas3kronf6448x48 (x) 48x489.083 ms0.584.689%ok
factorluf64512x51232.813 ms2.730.133%ok
factorluf32512x51231.412 ms2.850.072%ok
factorcholeskyf64512x51211.382 ms3.930.375%ok
factorcholeskyf32512x51210.904 ms4.100.193%ok
factorldltf64512x51211.433 ms3.910.375%ok
factorldlt_bkf64512x512166.832 ms0.270.030%ok
factorqrf64512x512645.997 ms0.280.010%ok
factorlqf64512x51268.522 ms2.610.063%ok
factorhessenbergf64512x512752.540 ms0.590.011%ok
factorhouseholderf64512x5124.000 ms0.261.052%*ok
factorinvf64512x51293.824 ms2.860.044%ok
factorlu_iterative_refinef32>f64512x51234.158 ms2.620.123%ok
eigeneigenvalue_symmetricf64256x25637.521 ms0.600.031%ok
eigeneigen_symmetricf64256x256127.218 ms1.190.011%ok
eigeneigen_symmetric_jacobif64256x256512.542 ms0.290.000%ok
eigeneigenvaluef64256x25654.014 ms3.110.024%ok
eigenschurf64256x256109.018 ms3.850.015%ok
eigensvdf64256x256118.388 ms3.120.014%ok
eigenexpmf64256x25626.294 ms9.570.0412%ok
eigensylvesterf64256x256259.941 ms3.870.015%ok
elementwiseexpf64n=838860827.353 ms0.314.919%ok
elementwiselogf64n=838860842.478 ms0.203.166%ok
elementwisesinf64n=838860839.362 ms0.213.416%ok
elementwisesqrtf64n=838860821.114 ms0.406.3612%ok
elementwiseerff64n=838860831.732 ms0.264.238%ok
sparsespmvf64512^2 grid, nnz 13086720.851 ms3.0832.0061%ok
sparsespmv_transf64512^2 grid, nnz 13086721.324 ms1.9820.5739%ok
sparsecrs_densef64512^2 grid, nnz 1308672 x 166.177 ms6.7814.2527%ok
sparsespgemmf64256^2 grid, nnz 32665618.036 ms0.180.641%*ok
sparselower_solve_levelf64192^2 grid, nnz(L) 10035641.815 ms1.119.1717%*ok
directsparse_choleskyf6496^2 grid, nnz 4569699.470 ms0.110.010%ok
directsparse_ldltf6496^2 grid, nnz 4569629.388 ms0.380.020%ok
directsupernodal_ldltf6496^2 grid, nnz 4569616.270 ms0.690.041%ok
directsparse_luf6496^2 grid, nnz 4569654.953 ms0.410.011%ok
directsupernodal_luf6496^2 grid, nnz 4569674.370 ms0.300.010%ok
directnative_kluf6496^2 grid, nnz 4569626.666 ms0.840.031%ok
directsparse_qrf6448^2 grid, nnz 1132884.782 ms0.020.000%ok
orderingamdf6496^2 grid, nnz 45696, scrambled3.675 ms0.000.00n/aok
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled5.691 ms0.000.00n/aok
orderingrcmf6496^2 grid, nnz 45696, scrambled0.445 ms0.000.00n/aok
orderingblock_triangular_formf6432 planted blocks, n 819215.474 ms0.000.00n/aok
krylovcg+ic0f64128^2 grid, n 1638433.757 ms0.000.00n/aok
krylovminres+diagf64128^2 grid, n 1638431.894 ms0.000.00n/aok
krylovbicg+ilu0f64128^2 grid, n 1638428.291 ms0.000.00n/aok
krylovbicgstab+ilu0f64128^2 grid, n 1638413.784 ms0.000.00n/aok
krylovbicgstab_ell+ilu0f64128^2 grid, n 1638418.055 ms0.000.00n/aok
krylovcgs+ilu0f64128^2 grid, n 1638415.877 ms0.000.00n/aok
krylovgmres30+ilu0f64128^2 grid, n 1638432.118 ms0.000.00n/aok
krylovqmr+ilu0f64128^2 grid, n 1638429.495 ms0.000.00n/aok
krylovtfqmr+ilu0f64128^2 grid, n 1638417.583 ms0.000.00n/aok
krylovidr_s1+ilu0f64128^2 grid, n 1638419.592 ms0.000.00n/aok
krylovidr_s2+ilu0f64128^2 grid, n 1638419.357 ms0.000.00n/aok
krylovidr_s4+ilu0f64128^2 grid, n 1638419.720 ms0.000.00n/aok
krylovidr_s8+ilu0f64128^2 grid, n 1638422.163 ms0.000.00n/aok
pcdiagonalf64128^2 grid, rows scaled0.036 ms0.000.00n/aok
pcilu_0f64128^2 grid2.403 ms0.000.00n/aok
pcilutf64128^2 grid23.132 ms0.000.00n/aok
pcic_0f64128^2 grid1.833 ms0.000.00n/aok
pcildlf64128^2 grid1.761 ms0.000.00n/aok
pcssorf64128^2 grid0.021 ms0.000.00n/aok
pcblock_diagonal4f64128^2 grid, rows scaled0.665 ms0.000.00n/aok
smootherjacobif64512^2 grid, nnz 13086721.736 ms1.5116.8932%ok
smoothergauss_seidelf64512^2 grid, nnz 13086721.501 ms1.7419.5437%ok
smoothersor1.5f64512^2 grid, nnz 13086721.818 ms1.4416.1331%ok
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)16.074 ms2.3918.0834%ok
sparse_eiglanczosf6464x65 Laplacian, k=4, m=8032.399 ms0.000.00n/aok
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=8047.014 ms0.000.00n/aok
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=4055.463 ms0.000.00n/aok
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps6.453 ms0.000.00n/aok
mixeddotf32>f64n=83886081.840 ms9.1236.4669%ok
mixedgemmf32>f64512x512x51215.320 ms17.520.2722%ok
mixedblock_scaled_doti8+f32n=8388608, block 320.476 ms35.2237.4271%ok
integerdoti32n=8388608, |v| <= 151.202 ms13.9655.84n/aok
integerdoti16>i32n=8388608, |v| <= 150.463 ms36.2472.47n/aok
integerdoti8>i32n=8388608, |v| <= 150.347 ms48.3548.35n/aok
integerdotu8i8>i32n=8388608, |v| <= 150.318 ms52.7952.79n/aok
integergemmi32512^3, |v| <= 204714.928 ms17.980.21n/aok
integergemmi16>i32512^3, |v| <= 204717.341 ms15.480.12n/aok
integergemmi8>i32512^3, |v| <= 12717.723 ms15.150.09n/aok
integergemm_quadi8>i32512^3, |v| <= 1273.711 ms72.340.42n/aok
integergemm_quadu8i8>i32512^3, |v| <= 1273.688 ms72.780.43n/aok

* working set fits in the last-level cache: a cache rate, shown as a % of the DRAM roofline.

ryzen-7-8845hs — native-fast, 8 threads, class medium

Section titled “ryzen-7-8845hs — native-fast, 8 threads, class medium”
CPUAMD Ryzen 7 8845HS w/ Radeon 780M Graphics (x86_64, 16 logical cores)
OSWindows 10.0.26300 (26H2)
compilerMSVC 19.52.37029
build/DWIN32 /D_WINDOWS /EHsc /O2 /Ob2 /DNDEBUG, SIMD SSE2
commitcd6762a6e989
FMA peak (ceiling)f64 625.1 GFLOP/s, f32 1250.1 GFLOP/s
… hardware, AVX-512Ff64 625.1, f32 1250.1 GFLOP/s
… this build, SSE2f64 312.8, f32 624.6 GFLOP/s
bandwidthread 53.2, update 53.3, triad 41.3 GB/s (ceiling 53.3)
last-level cache16 MiB
areaentriesmedian %roofbest %rooffailed checksmedian parallel efficiency
blas110111%127%012%
blas21057%115%012%
blas3104%13%020%
factor121%8%018%
eigen80%2%013%
elementwise58%12%012%
sparse540%249%013%
direct70%0%013%
ordering4n/an/a013%
krylov13n/an/a013%
pc7n/an/a012%
smoother337%37%013%
mg135%35%013%
sparse_eig4n/an/a013%
mixed370%70%013%
integer9n/an/a069%
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheckspeedup (efficiency)
blas1dotf64n=83886082.266 ms7.4059.24111%ok1.1x (13%)
blas1dotf32n=83886080.991 ms16.9467.75127%ok1.2x (15%)
blas1axpyf64n=83886083.548 ms4.7356.75106%ok1.0x (12%)
blas1axpyf32n=83886081.678 ms10.0059.98112%ok1.0x (12%)
blas1scalef64n=83886082.134 ms3.9362.89118%ok0.9x (11%)
blas1two_normf64n=83886081.046 ms16.0464.18120%ok1.1x (14%)
blas1one_normf64n=83886085.292 ms1.5912.6824%ok1.0x (12%)
blas1infinity_normf64n=83886085.098 ms1.6513.1625%ok1.0x (13%)
blas1sumf64n=83886085.143 ms1.6313.0524%ok1.0x (12%)
blas1dot_k3f64n=838860897.716 ms1.891.373%ok0.9x (12%)
blas2gemvf64n=40962.347 ms14.2957.21107%ok1.4x (18%)
blas2gemvf32n=40961.095 ms30.6661.34115%ok1.6x (19%)
blas2gemv_colmajorf64n=40969.513 ms3.5314.1226%ok2.3x (28%)
blas2gemv_colmajorf32n=40964.853 ms6.9113.8326%ok1.6x (20%)
blas2symvf64n=40969.195 ms3.657.3114%ok0.9x (12%)
blas2trmvf64n=40962.167 ms7.7431.0058%ok0.8x (11%)
blas2trmvf32n=40961.121 ms14.9629.9656%ok0.8x (11%)
blas2trsvf64n=40962.149 ms7.8131.2559%ok0.9x (11%)
blas2trsvf32n=40961.101 ms15.2430.5057%ok0.8x (10%)
blas2gerf64n=409617.212 ms1.9515.6029%ok0.8x (10%)
blas3gemmf64512x512x5123.236 ms82.951.9413%ok3.4x (42%)
blas3gemmf32512x512x5121.822 ms147.311.7312%ok3.1x (38%)
blas3gemmtf64512x512x5125.034 ms26.661.254%ok1.5x (18%)
blas3symmf64512x512x5123.734 ms71.891.6812%ok3.2x (40%)
blas3syrkf64512x512x51214.165 ms9.480.302%ok1.0x (12%)
blas3syr2kf64512x512x51224.116 ms11.130.262%ok1.0x (12%)
blas3trmmf64512x512x5125.917 ms22.680.894%ok1.6x (21%)
blas3trsmf64512x512x5126.195 ms21.660.853%ok1.6x (20%)
blas3trsmf32512x512x5124.605 ms29.150.572%ok1.4x (18%)
blas3kronf6448x48 (x) 48x489.264 ms0.574.589%ok1.0x (12%)
factorluf64512x51223.200 ms3.860.181%ok1.4x (18%)
factorluf32512x51222.390 ms4.000.090%ok1.4x (18%)
factorcholeskyf64512x51211.397 ms3.930.371%*ok1.0x (12%)
factorcholeskyf32512x51210.906 ms4.100.190%*ok1.0x (12%)
factorldltf64512x51211.437 ms3.910.371%*ok1.0x (12%)
factorldlt_bkf64512x512165.245 ms0.270.030%*ok1.0x (13%)
factorqrf64512x512395.043 ms0.450.010%ok1.6x (20%)
factorlqf64512x51244.608 ms4.010.091%ok1.5x (19%)
factorhessenbergf64512x512470.760 ms0.950.010%ok1.6x (20%)
factorhouseholderf64512x5121.021 ms1.034.118%*ok3.9x (49%)
factorinvf64512x51279.181 ms3.390.051%ok1.2x (15%)
factorlu_iterative_refinef32>f64512x51225.187 ms3.550.171%ok1.4x (17%)
eigeneigenvalue_symmetricf64256x25633.577 ms0.670.030%ok1.1x (14%)
eigeneigen_symmetricf64256x256123.975 ms1.220.010%ok1.0x (13%)
eigeneigen_symmetric_jacobif64256x256496.530 ms0.300.000%ok1.0x (13%)
eigeneigenvaluef64256x25660.808 ms2.760.020%ok0.9x (11%)
eigenschurf64256x256112.942 ms3.710.011%ok1.0x (12%)
eigensvdf64256x256118.691 ms3.110.010%ok1.0x (12%)
eigenexpmf64256x25619.380 ms12.990.052%ok1.4x (17%)
eigensylvesterf64256x256259.498 ms3.880.011%ok1.0x (13%)
elementwiseexpf64n=838860827.896 ms0.304.819%ok1.0x (12%)
elementwiselogf64n=838860842.556 ms0.203.156%ok1.0x (12%)
elementwisesinf64n=838860839.823 ms0.213.376%ok1.0x (12%)
elementwisesqrtf64n=838860821.289 ms0.396.3012%ok1.0x (12%)
elementwiseerff64n=838860831.967 ms0.264.208%ok1.0x (12%)
sparsespmvf64512^2 grid, nnz 13086720.205 ms12.78132.96249% !ok4.2x (52%)
sparsespmv_transf64512^2 grid, nnz 13086721.289 ms2.0321.1240%ok1.0x (13%)
sparsecrs_densef64512^2 grid, nnz 1308672 x 162.288 ms18.3038.4872%ok2.7x (34%)
sparsespgemmf64256^2 grid, nnz 32665618.861 ms0.170.611%*ok1.0x (12%)
sparselower_solve_levelf64192^2 grid, nnz(L) 10035641.753 ms1.159.5018%*ok1.0x (13%)
directsparse_choleskyf6496^2 grid, nnz 4569698.618 ms0.110.010%ok1.0x (13%)
directsparse_ldltf6496^2 grid, nnz 4569629.054 ms0.390.030%ok1.0x (13%)
directsupernodal_ldltf6496^2 grid, nnz 4569615.885 ms0.700.050%ok1.0x (13%)
directsparse_luf6496^2 grid, nnz 4569655.213 ms0.410.010%ok1.0x (12%)
directsupernodal_luf6496^2 grid, nnz 4569672.275 ms0.310.010%ok1.0x (13%)
directnative_kluf6496^2 grid, nnz 4569625.865 ms0.860.030%ok1.0x (13%)
directsparse_qrf6448^2 grid, nnz 1132883.988 ms0.020.000%*ok1.0x (13%)
orderingamdf6496^2 grid, nnz 45696, scrambled3.703 ms0.000.00n/aok1.0x (12%)
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled5.668 ms0.000.00n/aok1.0x (13%)
orderingrcmf6496^2 grid, nnz 45696, scrambled0.444 ms0.000.00n/aok1.0x (13%)
orderingblock_triangular_formf6432 planted blocks, n 819214.117 ms0.000.00n/aok1.1x (14%)
krylovcg+ic0f64128^2 grid, n 1638432.635 ms0.000.00n/aok1.0x (13%)
krylovminres+diagf64128^2 grid, n 1638432.034 ms0.000.00n/aok1.0x (12%)
krylovbicg+ilu0f64128^2 grid, n 1638427.434 ms0.000.00n/aok1.0x (13%)
krylovbicgstab+ilu0f64128^2 grid, n 1638413.466 ms0.000.00n/aok1.0x (13%)
krylovbicgstab_ell+ilu0f64128^2 grid, n 1638417.877 ms0.000.00n/aok1.0x (13%)
krylovcgs+ilu0f64128^2 grid, n 1638415.724 ms0.000.00n/aok1.0x (13%)
krylovgmres30+ilu0f64128^2 grid, n 1638431.046 ms0.000.00n/aok1.0x (13%)
krylovqmr+ilu0f64128^2 grid, n 1638428.980 ms0.000.00n/aok1.0x (13%)
krylovtfqmr+ilu0f64128^2 grid, n 1638417.304 ms0.000.00n/aok1.0x (13%)
krylovidr_s1+ilu0f64128^2 grid, n 1638419.322 ms0.000.00n/aok1.0x (13%)
krylovidr_s2+ilu0f64128^2 grid, n 1638418.431 ms0.000.00n/aok1.1x (13%)
krylovidr_s4+ilu0f64128^2 grid, n 1638419.587 ms0.000.00n/aok1.0x (13%)
krylovidr_s8+ilu0f64128^2 grid, n 1638422.431 ms0.000.00n/aok1.0x (12%)
pcdiagonalf64128^2 grid, rows scaled0.036 ms0.000.00n/aok1.0x (12%)
pcilu_0f64128^2 grid2.433 ms0.000.00n/aok1.0x (12%)
pcilutf64128^2 grid23.180 ms0.000.00n/aok1.0x (12%)
pcic_0f64128^2 grid1.828 ms0.000.00n/aok1.0x (13%)
pcildlf64128^2 grid1.758 ms0.000.00n/aok1.0x (13%)
pcssorf64128^2 grid0.022 ms0.000.00n/aok1.0x (12%)
pcblock_diagonal4f64128^2 grid, rows scaled0.664 ms0.000.00n/aok1.0x (13%)
smootherjacobif64512^2 grid, nnz 13086721.467 ms1.7819.9937%ok1.2x (15%)
smoothergauss_seidelf64512^2 grid, nnz 13086721.495 ms1.7519.6237%ok1.0x (13%)
smoothersor1.5f64512^2 grid, nnz 13086721.783 ms1.4716.4531%ok1.0x (13%)
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)15.564 ms2.4718.6735%ok1.0x (13%)
sparse_eiglanczosf6464x65 Laplacian, k=4, m=8032.454 ms0.000.00n/aok1.0x (12%)
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=8046.887 ms0.000.00n/aok1.0x (13%)
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=4054.576 ms0.000.00n/aok1.0x (13%)
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps6.519 ms0.000.00n/aok1.0x (12%)
mixeddotf32>f64n=83886081.786 ms9.3937.5670%ok1.0x (13%)
mixedgemmf32>f64512x512x5122.635 ms101.861.5916%ok5.8x (73%)
mixedblock_scaled_doti8+f32n=8388608, block 320.475 ms35.3537.5670%ok1.0x (13%)
integerdoti32n=8388608, |v| <= 151.229 ms13.6554.59n/aok1.0x (12%)
integerdoti16>i32n=8388608, |v| <= 150.473 ms35.4470.88n/aok1.0x (12%)
integerdoti8>i32n=8388608, |v| <= 150.346 ms48.4948.49n/aok1.0x (13%)
integerdotu8i8>i32n=8388608, |v| <= 150.318 ms52.8452.84n/aok1.0x (13%)
integergemmi32512^3, |v| <= 20472.438 ms110.091.29n/aok6.1x (77%)
integergemmi16>i32512^3, |v| <= 20472.544 ms105.510.82n/aok6.8x (85%)
integergemmi8>i32512^3, |v| <= 1272.569 ms104.510.61n/aok6.9x (86%)
integergemm_quadi8>i32512^3, |v| <= 1270.674 ms398.452.33n/aok5.5x (69%)
integergemm_quadu8i8>i32512^3, |v| <= 1270.672 ms399.162.34n/aok5.5x (69%)

* working set fits in the last-level cache: a cache rate, shown as a % of the DRAM roofline.

! more than 150% of the DRAM roofline for a bandwidth-bound result: the entry’s byte model is probably wrong.

ryzen-7-8845hs — blas, 1 thread, class medium

Section titled “ryzen-7-8845hs — blas, 1 thread, class medium”
CPUAMD Ryzen 7 8845HS w/ Radeon 780M Graphics (x86_64, 16 logical cores)
OSWindows 10.0.26300 (26H2)
compilerMSVC 19.52.37029
build/DWIN32 /D_WINDOWS /EHsc /O2 /Ob2 /DNDEBUG, SIMD scalar
commitcd6762a6e989
FMA peak (ceiling)f64 80.1 GFLOP/s, f32 160.1 GFLOP/s
… hardware, AVX-512Ff64 80.1, f32 160.1 GFLOP/s
… this build, scalarf64 1.6, f32 1.9 GFLOP/s
bandwidthread 42.8, update 53.5, triad 38.9 GB/s (ceiling 53.5)
last-level cache16 MiB
areaentriesmedian %roofbest %rooffailed checks
blas11025%110%0
blas21077%84%0
blas31068%92%0
factor1213%23%0
eigen817%53%0
elementwise57%12%0
sparse521%37%0
direct70%1%0
ordering4n/an/a0
krylov13n/an/a0
pc7n/an/a0
smoother335%36%0
mg134%34%0
sparse_eig4n/an/a0
mixed311%23%0
integer9n/an/a0
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheck
blas1dotf64n=838860811.491 ms1.4611.6822%ok
blas1dotf32n=838860810.308 ms1.636.5112%ok
blas1axpyf64n=838860811.930 ms1.4116.8832%ok
blas1axpyf32n=838860810.299 ms1.639.7718%ok
blas1scalef64n=83886082.287 ms3.6758.69110%ok
blas1two_normf64n=838860811.013 ms1.526.0911%ok
blas1one_normf64n=83886085.064 ms1.6613.2525%ok
blas1infinity_normf64n=83886085.072 ms1.6513.2325%ok
blas1sumf64n=83886085.062 ms1.6613.2625%ok
blas1dot_k3f64n=838860884.572 ms2.181.593%ok
blas2gemvf64n=40963.219 ms10.4241.7178%ok
blas2gemvf32n=40961.567 ms21.4142.8580%ok
blas2gemv_colmajorf64n=40963.374 ms9.9439.8074%ok
blas2gemv_colmajorf32n=40961.502 ms22.3444.7184%ok
blas2symvf64n=40961.826 ms18.3836.7969%ok
blas2trmvf64n=40961.626 ms10.3241.3277%ok
blas2trmvf32n=40960.847 ms19.8139.6774%ok
blas2trsvf64n=40961.614 ms10.4041.6378%ok
blas2trsvf32n=40960.908 ms18.4937.0169%ok
blas2gerf64n=409613.588 ms2.4719.7637%ok
blas3gemmf64512x512x5123.629 ms73.971.7392%ok
blas3gemmf32512x512x5121.913 ms140.301.6488%ok
blas3gemmtf64512x512x5122.726 ms49.242.3161%ok
blas3symmf64512x512x5123.806 ms70.541.6588%ok
blas3syrkf64512x512x5123.039 ms44.161.3855%ok
blas3syr2kf64512x512x5124.897 ms54.821.2868%ok
blas3trmmf64512x512x5121.888 ms71.102.7889%ok
blas3trsmf64512x512x5122.842 ms47.221.8459%ok
blas3trsmf32512x512x5121.346 ms99.691.9562%ok
blas3kronf6448x48 (x) 48x489.306 ms0.574.569%ok
factorluf64512x5124.986 ms17.940.8422%ok
factorluf32512x5122.443 ms36.630.8623%ok
factorcholeskyf64512x5124.350 ms10.280.9613%ok
factorcholeskyf32512x5121.676 ms26.691.2517%ok
factorldltf64512x51211.785 ms3.800.365%ok
factorldlt_bkf64512x512164.303 ms0.270.030%ok
factorqrf64512x51210.166 ms17.600.4122%ok
factorlqf64512x51264.507 ms2.770.073%ok
factorhessenbergf64512x512745.299 ms0.600.011%ok
factorhouseholderf64512x5123.878 ms0.271.082%*ok
factorinvf64512x51260.512 ms4.440.076%ok
factorlu_iterative_refinef32>f64512x5125.538 ms16.160.7620%ok
eigeneigenvalue_symmetricf64256x2561.678 ms13.330.6317%ok
eigeneigen_symmetricf64256x256129.208 ms1.170.011%ok
eigeneigen_symmetric_jacobif64256x256499.815 ms0.300.000%ok
eigeneigenvaluef64256x25616.673 ms10.060.0613%ok
eigenschurf64256x25621.833 ms19.210.0524%ok
eigensvdf64256x2568.649 ms42.680.1253%ok
eigenexpmf64256x25615.903 ms15.820.0720%ok
eigensylvesterf64256x25672.035 ms13.970.0217%ok
elementwiseexpf64n=838860827.847 ms0.304.829%ok
elementwiselogf64n=838860843.053 ms0.193.126%ok
elementwisesinf64n=838860838.927 ms0.223.456%ok
elementwisesqrtf64n=838860821.365 ms0.396.2812%ok
elementwiseerff64n=838860836.179 ms0.233.717%ok
sparsespmvf64512^2 grid, nnz 13086722.424 ms1.0811.2321%ok
sparsespmv_transf64512^2 grid, nnz 13086721.370 ms1.9119.8837%ok
sparsecrs_densef64512^2 grid, nnz 1308672 x 166.594 ms6.3513.3525%ok
sparsespgemmf64256^2 grid, nnz 32665617.840 ms0.180.641%*ok
sparselower_solve_levelf64192^2 grid, nnz(L) 10035641.752 ms1.159.5018%*ok
directsparse_choleskyf6496^2 grid, nnz 45696103.386 ms0.110.010%ok
directsparse_ldltf6496^2 grid, nnz 4569630.551 ms0.370.020%ok
directsupernodal_ldltf6496^2 grid, nnz 4569617.140 ms0.650.041%ok
directsparse_luf6496^2 grid, nnz 4569654.899 ms0.410.011%ok
directsupernodal_luf6496^2 grid, nnz 4569673.333 ms0.310.010%ok
directnative_kluf6496^2 grid, nnz 4569625.679 ms0.870.031%ok
directsparse_qrf6448^2 grid, nnz 1132879.417 ms0.020.000%ok
orderingamdf6496^2 grid, nnz 45696, scrambled3.651 ms0.000.00n/aok
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled5.609 ms0.000.00n/aok
orderingrcmf6496^2 grid, nnz 45696, scrambled0.427 ms0.000.00n/aok
orderingblock_triangular_formf6432 planted blocks, n 819214.147 ms0.000.00n/aok
krylovcg+ic0f64128^2 grid, n 1638440.253 ms0.000.00n/aok
krylovminres+diagf64128^2 grid, n 1638446.171 ms0.000.00n/aok
krylovbicg+ilu0f64128^2 grid, n 1638431.072 ms0.000.00n/aok
krylovbicgstab+ilu0f64128^2 grid, n 1638417.261 ms0.000.00n/aok
krylovbicgstab_ell+ilu0f64128^2 grid, n 1638422.413 ms0.000.00n/aok
krylovcgs+ilu0f64128^2 grid, n 1638417.946 ms0.000.00n/aok
krylovgmres30+ilu0f64128^2 grid, n 1638454.792 ms0.000.00n/aok
krylovqmr+ilu0f64128^2 grid, n 1638434.736 ms0.000.00n/aok
krylovtfqmr+ilu0f64128^2 grid, n 1638420.221 ms0.000.00n/aok
krylovidr_s1+ilu0f64128^2 grid, n 1638424.339 ms0.000.00n/aok
krylovidr_s2+ilu0f64128^2 grid, n 1638423.512 ms0.000.00n/aok
krylovidr_s4+ilu0f64128^2 grid, n 1638426.343 ms0.000.00n/aok
krylovidr_s8+ilu0f64128^2 grid, n 1638432.786 ms0.000.00n/aok
pcdiagonalf64128^2 grid, rows scaled0.036 ms0.000.00n/aok
pcilu_0f64128^2 grid2.401 ms0.000.00n/aok
pcilutf64128^2 grid22.279 ms0.000.00n/aok
pcic_0f64128^2 grid1.839 ms0.000.00n/aok
pcildlf64128^2 grid1.764 ms0.000.00n/aok
pcssorf64128^2 grid0.021 ms0.000.00n/aok
pcblock_diagonal4f64128^2 grid, rows scaled1.321 ms0.000.00n/aok
smootherjacobif64512^2 grid, nnz 13086721.588 ms1.6518.4735%ok
smoothergauss_seidelf64512^2 grid, nnz 13086721.540 ms1.7019.0436%ok
smoothersor1.5f64512^2 grid, nnz 13086721.791 ms1.4616.3831%ok
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)15.984 ms2.4118.1834%ok
sparse_eiglanczosf6464x65 Laplacian, k=4, m=8032.228 ms0.000.00n/aok
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=8032.894 ms0.000.00n/aok
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=4057.155 ms0.000.00n/aok
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps7.998 ms0.000.00n/aok
mixeddotf32>f64n=838860811.424 ms1.475.8711%ok
mixedgemmf32>f64512x512x51229.514 ms9.100.1411%ok
mixedblock_scaled_doti8+f32n=8388608, block 321.478 ms11.3512.0623%ok
integerdoti32n=8388608, |v| <= 153.132 ms5.3621.42n/aok
integerdoti16>i32n=8388608, |v| <= 152.878 ms5.8311.66n/aok
integerdoti8>i32n=8388608, |v| <= 151.394 ms12.0412.04n/aok
integerdotu8i8>i32n=8388608, |v| <= 151.394 ms12.0412.04n/aok
integergemmi32512^3, |v| <= 20478.394 ms31.980.37n/aok
integergemmi16>i32512^3, |v| <= 204730.301 ms8.860.07n/aok
integergemmi8>i32512^3, |v| <= 12730.554 ms8.790.05n/aok
integergemm_quadi8>i32512^3, |v| <= 12730.558 ms8.780.05n/aok
integergemm_quadu8i8>i32512^3, |v| <= 12727.237 ms9.860.06n/aok

* working set fits in the last-level cache: a cache rate, shown as a % of the DRAM roofline.

ryzen-7-8845hs — blas, 8 threads, class medium

Section titled “ryzen-7-8845hs — blas, 8 threads, class medium”
CPUAMD Ryzen 7 8845HS w/ Radeon 780M Graphics (x86_64, 16 logical cores)
OSWindows 10.0.26300 (26H2)
compilerMSVC 19.52.37029
build/DWIN32 /D_WINDOWS /EHsc /O2 /Ob2 /DNDEBUG, SIMD scalar
commitcd6762a6e989
FMA peak (ceiling)f64 625.3 GFLOP/s, f32 1251.4 GFLOP/s
… hardware, AVX-512Ff64 625.3, f32 1251.4 GFLOP/s
… this build, scalarf64 12.2, f32 14.4 GFLOP/s
bandwidthread 53.1, update 53.3, triad 41.2 GB/s (ceiling 53.3)
last-level cache16 MiB
areaentriesmedian %roofbest %rooffailed checksmedian parallel efficiency
blas11093%121%042%
blas210101%119%016%
blas31040%71%063%
factor122%8%011%
eigen82%5%011%
elementwise57%12%012%
sparse534%257%013%
direct70%0%012%
ordering4n/an/a013%
krylov13n/an/a012%
pc7n/an/a012%
smoother337%38%013%
mg135%35%013%
sparse_eig4n/an/a012%
mixed311%22%012%
integer9n/an/a082%
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheckspeedup (efficiency)
blas1dotf64n=83886082.309 ms7.2758.12109%ok5.0x (62%)
blas1dotf32n=83886081.359 ms12.3549.3893%ok7.6x (95%)
blas1axpyf64n=83886083.519 ms4.7757.20107%ok3.4x (42%)
blas1axpyf32n=83886081.776 ms9.4456.67106%ok5.8x (72%)
blas1scalef64n=83886082.084 ms4.0264.40121%ok1.1x (14%)
blas1two_normf64n=83886081.518 ms11.0544.2083%ok7.3x (91%)
blas1one_normf64n=83886085.069 ms1.6513.2425%ok1.0x (12%)
blas1infinity_normf64n=83886085.100 ms1.6413.1625%ok1.0x (12%)
blas1sumf64n=83886085.079 ms1.6513.2125%ok1.0x (12%)
blas1dot_k3f64n=838860886.046 ms2.141.563%ok1.0x (12%)
blas2gemvf64n=40962.550 ms13.1652.6799%ok1.3x (16%)
blas2gemvf32n=40961.156 ms29.0258.07109%ok1.4x (17%)
blas2gemv_colmajorf64n=40962.695 ms12.4549.8293%ok1.3x (16%)
blas2gemv_colmajorf32n=40961.251 ms26.8253.67101%ok1.2x (15%)
blas2symvf64n=40961.215 ms27.6155.28104%ok1.5x (19%)
blas2trmvf64n=40961.218 ms13.7755.15103%ok1.3x (17%)
blas2trmvf32n=40960.530 ms31.6763.41119%ok1.6x (20%)
blas2trsvf64n=40961.589 ms10.5642.2879%ok1.0x (13%)
blas2trsvf32n=40960.946 ms17.7435.5267%ok1.0x (12%)
blas2gerf64n=409613.670 ms2.4519.6437%ok1.0x (12%)
blas3gemmf64512x512x5120.605 ms443.8410.4071%ok6.0x (75%)
blas3gemmf32512x512x5120.340 ms790.689.2763%ok5.6x (70%)
blas3gemmtf64512x512x5121.356 ms98.974.6416%ok2.0x (25%)
blas3symmf64512x512x5120.607 ms442.3810.3771%ok6.3x (78%)
blas3syrkf64512x512x5121.720 ms78.032.4412%ok1.8x (22%)
blas3syr2kf64512x512x5122.387 ms112.482.6418%ok2.1x (26%)
blas3trmmf64512x512x5120.398 ms337.2313.1754%ok4.7x (59%)
blas3trsmf64512x512x5120.557 ms241.059.4239%ok5.1x (64%)
blas3trsmf32512x512x5120.266 ms503.829.8440%ok5.1x (63%)
blas3kronf6448x48 (x) 48x489.245 ms0.574.599%ok1.0x (13%)
factorluf64512x5129.104 ms9.830.462%ok0.5x (7%)
factorluf32512x5124.345 ms20.590.482%ok0.6x (7%)
factorcholeskyf64512x5125.960 ms7.510.701%*ok0.7x (9%)
factorcholeskyf32512x5122.308 ms19.390.912%*ok0.7x (9%)
factorldltf64512x51211.867 ms3.770.351%*ok1.0x (12%)
factorldlt_bkf64512x512167.173 ms0.270.030%*ok1.0x (12%)
factorqrf64512x51211.893 ms15.050.352%ok0.9x (11%)
factorlqf64512x51245.484 ms3.930.091%ok1.4x (18%)
factorhessenbergf64512x512477.079 ms0.940.010%ok1.6x (20%)
factorhouseholderf64512x5121.004 ms1.054.188%*ok3.9x (48%)
factorinvf64512x51265.954 ms4.070.061%ok0.9x (11%)
factorlu_iterative_refinef32>f64512x5127.245 ms12.350.582%ok0.8x (10%)
eigeneigenvalue_symmetricf64256x2563.588 ms6.230.291%ok0.5x (6%)
eigeneigen_symmetricf64256x256133.014 ms1.140.010%ok1.0x (12%)
eigeneigen_symmetric_jacobif64256x256519.779 ms0.290.000%ok1.0x (12%)
eigeneigenvaluef64256x25621.078 ms7.960.051%ok0.8x (10%)
eigenschurf64256x25626.127 ms16.050.043%ok0.8x (10%)
eigensvdf64256x25612.133 ms30.420.095%ok0.7x (9%)
eigenexpmf64256x25618.256 ms13.780.062%ok0.9x (11%)
eigensylvesterf64256x25680.883 ms12.450.022%ok0.9x (11%)
elementwiseexpf64n=838860828.560 ms0.294.709%ok1.0x (12%)
elementwiselogf64n=838860843.937 ms0.193.056%ok1.0x (12%)
elementwisesinf64n=838860839.470 ms0.213.406%ok1.0x (12%)
elementwisesqrtf64n=838860821.387 ms0.396.2812%ok1.0x (12%)
elementwiseerff64n=838860834.953 ms0.243.847%ok1.0x (13%)
sparsespmvf64512^2 grid, nnz 13086720.198 ms13.19137.25257% !ok12.2x (153%)
sparsespmv_transf64512^2 grid, nnz 13086721.516 ms1.7317.9634%ok0.9x (11%)
sparsecrs_densef64512^2 grid, nnz 1308672 x 162.320 ms18.0537.9671%ok2.8x (36%)
sparsespgemmf64256^2 grid, nnz 32665618.291 ms0.180.631%*ok1.0x (12%)
sparselower_solve_levelf64192^2 grid, nnz(L) 10035641.705 ms1.189.7718%*ok1.0x (13%)
directsparse_choleskyf6496^2 grid, nnz 45696101.545 ms0.110.010%ok1.0x (13%)
directsparse_ldltf6496^2 grid, nnz 4569631.941 ms0.350.020%ok1.0x (12%)
directsupernodal_ldltf6496^2 grid, nnz 4569618.977 ms0.590.040%ok0.9x (11%)
directsparse_luf6496^2 grid, nnz 4569654.848 ms0.410.010%ok1.0x (13%)
directsupernodal_luf6496^2 grid, nnz 4569674.254 ms0.300.010%ok1.0x (12%)
directnative_kluf6496^2 grid, nnz 4569626.133 ms0.860.030%ok1.0x (12%)
directsparse_qrf6448^2 grid, nnz 1132880.579 ms0.020.000%*ok1.0x (12%)
orderingamdf6496^2 grid, nnz 45696, scrambled3.723 ms0.000.00n/aok1.0x (12%)
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled5.692 ms0.000.00n/aok1.0x (12%)
orderingrcmf6496^2 grid, nnz 45696, scrambled0.422 ms0.000.00n/aok1.0x (13%)
orderingblock_triangular_formf6432 planted blocks, n 819214.111 ms0.000.00n/aok1.0x (13%)
krylovcg+ic0f64128^2 grid, n 1638440.303 ms0.000.00n/aok1.0x (12%)
krylovminres+diagf64128^2 grid, n 1638446.739 ms0.000.00n/aok1.0x (12%)
krylovbicg+ilu0f64128^2 grid, n 1638431.310 ms0.000.00n/aok1.0x (12%)
krylovbicgstab+ilu0f64128^2 grid, n 1638417.299 ms0.000.00n/aok1.0x (12%)
krylovbicgstab_ell+ilu0f64128^2 grid, n 1638422.467 ms0.000.00n/aok1.0x (12%)
krylovcgs+ilu0f64128^2 grid, n 1638417.965 ms0.000.00n/aok1.0x (12%)
krylovgmres30+ilu0f64128^2 grid, n 1638454.976 ms0.000.00n/aok1.0x (12%)
krylovqmr+ilu0f64128^2 grid, n 1638434.869 ms0.000.00n/aok1.0x (12%)
krylovtfqmr+ilu0f64128^2 grid, n 1638420.462 ms0.000.00n/aok1.0x (12%)
krylovidr_s1+ilu0f64128^2 grid, n 1638424.452 ms0.000.00n/aok1.0x (12%)
krylovidr_s2+ilu0f64128^2 grid, n 1638423.691 ms0.000.00n/aok1.0x (12%)
krylovidr_s4+ilu0f64128^2 grid, n 1638426.532 ms0.000.00n/aok1.0x (12%)
krylovidr_s8+ilu0f64128^2 grid, n 1638433.729 ms0.000.00n/aok1.0x (12%)
pcdiagonalf64128^2 grid, rows scaled0.036 ms0.000.00n/aok1.0x (12%)
pcilu_0f64128^2 grid2.535 ms0.000.00n/aok0.9x (12%)
pcilutf64128^2 grid22.782 ms0.000.00n/aok1.0x (12%)
pcic_0f64128^2 grid1.870 ms0.000.00n/aok1.0x (12%)
pcildlf64128^2 grid1.773 ms0.000.00n/aok1.0x (12%)
pcssorf64128^2 grid0.022 ms0.000.00n/aok1.0x (12%)
pcblock_diagonal4f64128^2 grid, rows scaled1.329 ms0.000.00n/aok1.0x (12%)
smootherjacobif64512^2 grid, nnz 13086721.461 ms1.7920.0738%ok1.1x (14%)
smoothergauss_seidelf64512^2 grid, nnz 13086721.499 ms1.7519.5737%ok1.0x (13%)
smoothersor1.5f64512^2 grid, nnz 13086721.764 ms1.4816.6331%ok1.0x (13%)
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)15.486 ms2.4818.7735%ok1.0x (13%)
sparse_eiglanczosf6464x65 Laplacian, k=4, m=8032.459 ms0.000.00n/aok1.0x (12%)
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=8033.106 ms0.000.00n/aok1.0x (12%)
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=4058.023 ms0.000.00n/aok1.0x (12%)
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps8.046 ms0.000.00n/aok1.0x (12%)
mixeddotf32>f64n=838860811.440 ms1.475.8711%ok1.0x (12%)
mixedgemmf32>f64512x512x5125.023 ms53.440.849%ok5.9x (73%)
mixedblock_scaled_doti8+f32n=8388608, block 321.506 ms11.1411.8322%ok1.0x (12%)
integerdoti32n=8388608, |v| <= 153.136 ms5.3521.40n/aok1.0x (12%)
integerdoti16>i32n=8388608, |v| <= 152.893 ms5.8011.60n/aok1.0x (12%)
integerdoti8>i32n=8388608, |v| <= 151.408 ms11.9111.91n/aok1.0x (12%)
integerdotu8i8>i32n=8388608, |v| <= 151.409 ms11.9011.90n/aok1.0x (12%)
integergemmi32512^3, |v| <= 20471.281 ms209.502.46n/aok6.6x (82%)
integergemmi16>i32512^3, |v| <= 20474.212 ms63.730.50n/aok7.2x (90%)
integergemmi8>i32512^3, |v| <= 1274.241 ms63.290.37n/aok7.2x (90%)
integergemm_quadi8>i32512^3, |v| <= 1274.240 ms63.320.37n/aok7.2x (90%)
integergemm_quadu8i8>i32512^3, |v| <= 1273.859 ms69.550.41n/aok7.1x (88%)

* working set fits in the last-level cache: a cache rate, shown as a % of the DRAM roofline.

! more than 150% of the DRAM roofline for a bandwidth-bound result: the entry’s byte model is probably wrong.