Skip to content

mtl5_bench capability card -- xeon-e5-2420v2

The platform benchmark’s capability cards for the xeon-e5-2420v2 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: 637 of 666 entries within +-5% of the previous session, measured after a cold boot on the same commit (median ratio 0.998 to 1.001 per run; benchmarks/compare_sessions.py). Relates to #721 and #737.

session20261111T193510Z (UTC)
commitbfa1c2822688
classmedium
runsblas at 1 thread, blas at 6 threads, native-fast at 1 thread, native-fast at 6 threads, native at 1 thread, native at 6 threads
failed checks0
CPUs0,1,2,3,4,5 (pinned)
clock policygovernor performance on every CPU, turbo on; performance 1200-2700 MHz on cpus 0,1,2,3,4,5 (governor performance)
raw databenchmarks/data/xeon-e5-2420v2/mtl5_bench/20261111T193510Z/
earlier sessions20261109T035023Z, 20261111T190602Z

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 6Tnative-fast 1Tnative-fast 6Tblas 1Tblas 6T
blas124%55%101%98%24%54%
blas294%88%91%84%94%104%
blas340%8%72%28%162%126%
factor17%3%17%3%54%18%
eigen15%3%15%3%51%9%
elementwise9%8%9%8%9%7%
sparse57%69%56%58%56%68%
direct2%0%1%0%2%0%
orderingn/an/an/an/an/an/a
krylovn/an/an/an/an/an/a
pcn/an/an/an/an/an/a
smoother79%73%79%72%79%72%
mg79%74%78%62%79%74%
sparse_eign/an/an/an/an/an/a
mixed23%22%75%69%23%22%
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 6Tnative-fast 1Tnative-fast 6Tblas 1Tblas 6T
blas1dotf64n=8388608184.501 ms32.893 ms11.950 ms10.905 ms180.998 ms32.728 ms
blas1dotf32n=838860827.429 ms5.533 ms5.963 ms5.494 ms27.095 ms5.537 ms
blas1axpyf64n=8388608218.067 ms38.924 ms18.827 ms17.914 ms213.023 ms38.283 ms
blas1axpyf32n=838860891.685 ms16.377 ms9.019 ms8.469 ms91.531 ms16.522 ms
blas1scalef64n=838860812.459 ms12.079 ms13.211 ms12.674 ms12.619 ms12.140 ms
blas1two_normf64n=8388608185.147 ms32.925 ms6.388 ms5.459 ms183.955 ms32.883 ms
blas1one_normf64n=83886089.757 ms9.769 ms10.333 ms9.865 ms9.705 ms9.749 ms
blas1infinity_normf64n=838860813.076 ms12.997 ms13.777 ms13.055 ms13.055 ms13.000 ms
blas1sumf64n=83886089.721 ms9.761 ms10.209 ms9.830 ms9.760 ms9.777 ms
blas1dot_k3f64n=8388608281.893 ms280.150 ms272.662 ms273.778 ms277.771 ms279.926 ms
blas2gemvf64n=409613.142 ms10.963 ms13.273 ms10.938 ms14.263 ms11.306 ms
blas2gemvf32n=40968.246 ms5.509 ms6.450 ms5.483 ms6.237 ms5.465 ms
blas2gemv_colmajorf64n=409613.398 ms10.956 ms40.732 ms21.167 ms12.939 ms10.927 ms
blas2gemv_colmajorf32n=40966.614 ms5.503 ms17.404 ms11.157 ms6.616 ms5.698 ms
blas2symvf64n=409616.147 ms16.195 ms16.138 ms16.031 ms6.948 ms5.786 ms
blas2trmvf64n=40967.017 ms7.049 ms7.059 ms6.855 ms7.578 ms5.728 ms
blas2trmvf32n=40964.556 ms4.725 ms3.526 ms3.599 ms3.689 ms2.963 ms
blas2trsvf64n=40966.840 ms6.824 ms7.066 ms6.866 ms7.464 ms7.405 ms
blas2trsvf32n=40964.329 ms4.622 ms3.614 ms3.587 ms3.456 ms3.515 ms
blas2gerf64n=409628.119 ms27.852 ms28.123 ms27.944 ms28.043 ms27.907 ms
blas3gemmf64512x512x51264.296 ms11.545 ms27.725 ms5.933 ms14.028 ms2.875 ms
blas3gemmf32512x512x51234.799 ms6.427 ms13.517 ms2.909 ms8.149 ms1.806 ms
blas3gemmtf64512x512x51228.665 ms29.382 ms16.977 ms5.321 ms8.734 ms2.546 ms
blas3symmf64512x512x51259.937 ms60.597 ms28.856 ms6.897 ms14.089 ms2.900 ms
blas3syrkf64512x512x51237.686 ms38.015 ms36.868 ms37.050 ms7.878 ms2.519 ms
blas3syr2kf64512x512x51264.406 ms65.983 ms63.609 ms64.310 ms15.034 ms3.711 ms
blas3trmmf64512x512x51233.164 ms32.931 ms17.787 ms8.166 ms7.459 ms1.608 ms
blas3trsmf64512x512x51232.472 ms32.632 ms18.178 ms8.943 ms8.166 ms1.767 ms
blas3trsmf32512x512x51216.230 ms16.407 ms8.881 ms4.718 ms5.319 ms1.242 ms
blas3kronf6448x48 (x) 48x4838.832 ms38.690 ms38.540 ms38.408 ms38.615 ms38.589 ms
factorluf64512x51230.799 ms23.852 ms29.596 ms23.113 ms9.239 ms5.537 ms
factorluf32512x51216.527 ms13.991 ms16.040 ms13.575 ms6.431 ms3.901 ms
factorcholeskyf64512x51224.842 ms25.283 ms25.624 ms25.494 ms5.891 ms4.081 ms
factorcholeskyf32512x51223.592 ms23.800 ms24.180 ms24.031 ms3.928 ms3.108 ms
factorldltf64512x51224.844 ms25.103 ms25.865 ms25.768 ms25.363 ms25.485 ms
factorldlt_bkf64512x512150.471 ms139.070 ms134.445 ms133.132 ms146.733 ms133.535 ms
factorqrf64512x512524.547 ms323.622 ms497.393 ms439.443 ms19.966 ms12.973 ms
factorlqf64512x51273.535 ms51.363 ms74.031 ms50.588 ms74.673 ms51.643 ms
factorhessenbergf64512x512633.534 ms417.343 ms614.277 ms406.816 ms630.391 ms595.367 ms
factorhouseholderf64512x5123.210 ms0.985 ms3.170 ms0.981 ms3.230 ms0.974 ms
factorinvf64512x512151.234 ms143.911 ms153.982 ms148.554 ms129.311 ms131.199 ms
factorlu_iterative_refinef32>f64512x51221.708 ms19.420 ms21.144 ms19.093 ms12.848 ms8.487 ms
eigeneigenvalue_symmetricf64256x25662.064 ms63.741 ms60.777 ms60.932 ms6.239 ms5.791 ms
eigeneigen_symmetricf64256x256172.766 ms172.253 ms173.587 ms171.823 ms174.177 ms170.646 ms
eigeneigen_symmetric_jacobif64256x2561000.721 ms996.068 ms1002.472 ms988.132 ms999.183 ms999.865 ms
eigeneigenvaluef64256x256108.517 ms107.492 ms104.061 ms105.251 ms53.177 ms56.535 ms
eigenschurf64256x256266.236 ms264.391 ms254.869 ms259.080 ms72.095 ms75.059 ms
eigensvdf64256x256301.331 ms302.720 ms259.287 ms260.619 ms29.500 ms26.503 ms
eigenexpmf64256x25685.762 ms37.541 ms50.632 ms32.297 ms35.728 ms27.353 ms
eigensylvesterf64256x256609.242 ms582.347 ms575.084 ms564.506 ms189.501 ms196.965 ms
elementwiseexpf64n=8388608151.803 ms151.636 ms151.136 ms153.676 ms150.786 ms159.123 ms
elementwiselogf64n=8388608150.937 ms142.557 ms140.081 ms141.178 ms141.186 ms146.509 ms
elementwisesinf64n=8388608172.986 ms172.453 ms171.322 ms171.237 ms172.071 ms171.699 ms
elementwisesqrtf64n=838860890.716 ms92.119 ms90.254 ms92.044 ms90.985 ms89.941 ms
elementwiseerff64n=83886082478.330 ms2481.949 ms2470.162 ms2472.458 ms2476.708 ms2476.149 ms
sparsespmvf64512^2 grid, nnz 13086722.463 ms2.303 ms2.464 ms2.312 ms2.458 ms2.320 ms
sparsespmv_transf64512^2 grid, nnz 13086723.079 ms3.112 ms4.124 ms4.104 ms3.097 ms3.087 ms
sparsecrs_densef64512^2 grid, nnz 1308672 x 1615.127 ms11.353 ms15.072 ms11.378 ms15.204 ms11.345 ms
sparsespgemmf64256^2 grid, nnz 32665616.424 ms16.053 ms16.446 ms16.364 ms16.224 ms16.262 ms
sparselower_solve_levelf64192^2 grid, nnz(L) 10035645.777 ms5.770 ms5.822 ms5.800 ms5.776 ms5.831 ms
directsparse_choleskyf6496^2 grid, nnz 45696274.115 ms273.553 ms238.908 ms240.712 ms273.021 ms275.330 ms
directsparse_ldltf6496^2 grid, nnz 4569669.604 ms69.419 ms71.511 ms70.354 ms69.556 ms69.596 ms
directsupernodal_ldltf6496^2 grid, nnz 4569626.544 ms26.388 ms27.030 ms26.307 ms26.648 ms26.254 ms
directsparse_luf6496^2 grid, nnz 45696116.264 ms115.723 ms117.001 ms117.408 ms119.204 ms116.281 ms
directsupernodal_luf6496^2 grid, nnz 45696153.213 ms153.331 ms154.406 ms156.562 ms155.239 ms153.847 ms
directnative_kluf6496^2 grid, nnz 4569647.885 ms48.018 ms48.760 ms48.816 ms48.043 ms48.136 ms
directsparse_qrf6448^2 grid, nnz 11328231.569 ms231.533 ms234.125 ms226.836 ms232.139 ms231.070 ms
orderingamdf6496^2 grid, nnz 45696, scrambled8.966 ms8.909 ms9.063 ms9.045 ms9.135 ms8.903 ms
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled12.318 ms12.274 ms12.162 ms12.353 ms12.372 ms12.201 ms
orderingrcmf6496^2 grid, nnz 45696, scrambled0.631 ms0.632 ms0.620 ms0.631 ms0.625 ms0.615 ms
orderingblock_triangular_formf6432 planted blocks, n 819240.956 ms40.676 ms40.620 ms40.744 ms41.388 ms40.831 ms
krylovcg+ic0f64128^2 grid, n 16384231.752 ms233.678 ms79.629 ms80.159 ms229.514 ms232.458 ms
krylovminres+diagf64128^2 grid, n 16384393.721 ms394.637 ms110.372 ms111.355 ms396.802 ms397.041 ms
krylovbicg+ilu0f64128^2 grid, n 16384126.980 ms126.593 ms64.425 ms64.214 ms126.671 ms239.140 ms
krylovbicgstab+ilu0f64128^2 grid, n 16384101.953 ms101.729 ms33.612 ms31.644 ms101.910 ms102.078 ms
krylovbicgstab_ell+ilu0f64128^2 grid, n 16384121.936 ms121.185 ms37.815 ms38.311 ms121.217 ms121.437 ms
krylovcgs+ilu0f64128^2 grid, n 1638479.133 ms79.198 ms36.870 ms37.171 ms78.757 ms78.925 ms
krylovgmres30+ilu0f64128^2 grid, n 16384519.380 ms520.399 ms68.906 ms66.202 ms516.955 ms514.809 ms
krylovqmr+ilu0f64128^2 grid, n 16384175.151 ms174.551 ms74.107 ms74.367 ms176.476 ms174.287 ms
krylovtfqmr+ilu0f64128^2 grid, n 1638490.825 ms91.725 ms37.183 ms37.722 ms91.619 ms91.537 ms
krylovidr_s1+ilu0f64128^2 grid, n 16384136.209 ms137.763 ms43.287 ms96.788 ms139.590 ms136.001 ms
krylovidr_s2+ilu0f64128^2 grid, n 16384137.658 ms139.507 ms41.879 ms41.768 ms140.118 ms138.021 ms
krylovidr_s4+ilu0f64128^2 grid, n 16384170.841 ms171.042 ms44.653 ms47.106 ms170.930 ms173.402 ms
krylovidr_s8+ilu0f64128^2 grid, n 16384249.427 ms251.140 ms53.810 ms55.228 ms248.486 ms248.610 ms
pcdiagonalf64128^2 grid, rows scaled0.127 ms0.129 ms0.129 ms0.130 ms0.128 ms0.129 ms
pcilu_0f64128^2 grid4.393 ms4.471 ms4.476 ms4.458 ms4.456 ms4.442 ms
pcilutf64128^2 grid34.283 ms33.093 ms33.710 ms33.400 ms33.044 ms33.100 ms
pcic_0f64128^2 grid2.861 ms2.919 ms2.866 ms2.871 ms2.915 ms2.869 ms
pcildlf64128^2 grid2.717 ms2.767 ms2.771 ms2.800 ms2.936 ms2.712 ms
pcssorf64128^2 grid0.075 ms0.079 ms0.075 ms0.074 ms0.119 ms0.077 ms
pcblock_diagonal4f64128^2 grid, rows scaled2.061 ms2.182 ms2.068 ms2.186 ms3.389 ms3.518 ms
smootherjacobif64512^2 grid, nnz 13086723.653 ms3.587 ms3.578 ms3.588 ms3.623 ms3.597 ms
smoothergauss_seidelf64512^2 grid, nnz 13086723.109 ms3.153 ms3.113 ms3.134 ms3.123 ms3.119 ms
smoothersor1.5f64512^2 grid, nnz 13086723.804 ms3.808 ms3.799 ms3.879 ms3.828 ms3.804 ms
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)35.903 ms34.898 ms35.497 ms40.859 ms35.710 ms34.676 ms
sparse_eiglanczosf6464x65 Laplacian, k=4, m=80140.896 ms146.722 ms139.647 ms139.520 ms142.049 ms142.441 ms
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=80189.695 ms188.201 ms181.001 ms180.686 ms148.131 ms147.324 ms
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=40147.248 ms146.308 ms152.067 ms149.311 ms144.134 ms142.578 ms
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps42.413 ms42.820 ms16.329 ms16.176 ms42.513 ms42.322 ms
mixeddotf32>f64n=8388608187.790 ms188.901 ms7.204 ms7.170 ms188.768 ms189.054 ms
mixedgemmf32>f64512x512x512113.327 ms20.564 ms62.449 ms11.774 ms113.542 ms20.551 ms
mixedblock_scaled_doti8+f32n=8388608, block 322.237 ms2.336 ms2.266 ms2.266 ms2.367 ms2.340 ms
integerdoti32n=8388608, |v| <= 156.469 ms6.485 ms6.039 ms6.024 ms6.620 ms6.459 ms
integerdoti16>i32n=8388608, |v| <= 153.159 ms3.108 ms2.985 ms2.998 ms3.112 ms3.148 ms
integerdoti8>i32n=8388608, |v| <= 155.010 ms5.081 ms2.085 ms2.073 ms5.015 ms5.006 ms
integerdotu8i8>i32n=8388608, |v| <= 155.094 ms4.987 ms1.677 ms1.641 ms4.999 ms4.991 ms
integergemmi32512^3, |v| <= 204757.725 ms10.387 ms18.952 ms3.975 ms57.681 ms10.410 ms
integergemmi16>i32512^3, |v| <= 204745.202 ms8.249 ms22.131 ms4.553 ms45.933 ms8.237 ms
integergemmi8>i32512^3, |v| <= 12723.766 ms4.348 ms22.000 ms4.511 ms23.593 ms4.345 ms
integergemm_quadi8>i32512^3, |v| <= 12723.701 ms4.350 ms16.194 ms3.342 ms23.568 ms4.343 ms
integergemm_quadu8i8>i32512^3, |v| <= 12771.276 ms13.148 ms12.365 ms2.590 ms70.878 ms13.104 ms

xeon-e5-2420v2 — native, 1 thread, class medium

Section titled “xeon-e5-2420v2 — native, 1 thread, class medium”
CPUIntel(R) Xeon(R) CPU E5-2420 v2 @ 2.20GHz (x86_64, 12 logical cores)
OSLinux Ubuntu 26.04.1 LTS (7.0.0-43-generic)
compilerGCC 15.2.0
build -O3 -DNDEBUG, SIMD scalar
commitbfa1c2822688
FMA peak (ceiling)f64 10.5 GFLOP/s, f32 20.9 GFLOP/s
… hardware, SSE2f64 10.5, f32 20.9 GFLOP/s
… this build, scalarf64 0.1, f32 0.6 GFLOP/s
bandwidthread 10.2, update 10.2, triad 8.0 GB/s (ceiling 10.2)
last-level cache15 MiB
areaentriesmedian %roofbest %rooffailed checks
blas11024%106%0
blas21094%100%0
blas31040%45%0
factor1217%39%0
eigen815%28%0
elementwise59%14%0
sparse557%108%0
direct72%4%0
ordering4n/an/a0
krylov13n/an/a0
pc7n/an/a0
smoother379%92%0
mg179%79%0
sparse_eig4n/an/a0
mixed323%78%0
integer9n/an/a0
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheck
blas1dotf64n=8388608184.501 ms0.090.737%ok
blas1dotf32n=838860827.429 ms0.612.4524%ok
blas1axpyf64n=8388608218.067 ms0.080.929%ok
blas1axpyf32n=838860891.685 ms0.181.1011%ok
blas1scalef64n=838860812.459 ms0.6710.77106%ok
blas1two_normf64n=8388608185.147 ms0.090.364%ok
blas1one_normf64n=83886089.757 ms0.866.8867%ok
blas1infinity_normf64n=838860813.076 ms0.645.1350%ok
blas1sumf64n=83886089.721 ms0.866.9068%ok
blas1dot_k3f64n=8388608281.893 ms0.650.486%ok
blas2gemvf64n=409613.142 ms2.5510.22100%ok
blas2gemvf32n=40968.246 ms4.078.1480%ok
blas2gemv_colmajorf64n=409613.398 ms2.5010.0298%ok
blas2gemv_colmajorf32n=40966.614 ms5.0710.1599%ok
blas2symvf64n=409616.147 ms2.084.1641%ok
blas2trmvf64n=40967.017 ms2.399.5794%ok
blas2trmvf32n=40964.556 ms3.687.3772%ok
blas2trsvf64n=40966.840 ms2.459.8296%ok
blas2trsvf32n=40964.329 ms3.887.7676%ok
blas2gerf64n=409628.119 ms1.199.5594%ok
blas3gemmf64512x512x51264.296 ms4.170.1040%ok
blas3gemmf32512x512x51234.799 ms7.710.0937%ok
blas3gemmtf64512x512x51228.665 ms4.680.2245%ok
blas3symmf64512x512x51259.937 ms4.480.1043%ok
blas3syrkf64512x512x51237.686 ms3.560.1134%ok
blas3syr2kf64512x512x51264.406 ms4.170.1040%ok
blas3trmmf64512x512x51233.164 ms4.050.1639%ok
blas3trsmf64512x512x51232.472 ms4.130.1639%ok
blas3trsmf32512x512x51216.230 ms8.270.1640%ok
blas3kronf6448x48 (x) 48x4838.832 ms0.141.0911%ok
factorluf64512x51230.799 ms2.910.1428%ok
factorluf32512x51216.527 ms5.410.1326%ok
factorcholeskyf64512x51224.842 ms1.800.1717%ok
factorcholeskyf32512x51223.592 ms1.900.099%ok
factorldltf64512x51224.844 ms1.800.1717%ok
factorldlt_bkf64512x512150.471 ms0.300.033%ok
factorqrf64512x512524.547 ms0.340.013%ok
factorlqf64512x51273.535 ms2.430.0623%ok
factorhessenbergf64512x512633.534 ms0.710.017%ok
factorhouseholderf64512x5123.210 ms0.331.3113%*ok
factorinvf64512x512151.234 ms1.770.0317%ok
factorlu_iterative_refinef32>f64512x51221.708 ms4.120.1939%ok
eigeneigenvalue_symmetricf64256x25662.064 ms0.360.023%ok
eigeneigen_symmetricf64256x256172.766 ms0.870.018%ok
eigeneigen_symmetric_jacobif64256x2561000.721 ms0.150.001%ok
eigeneigenvaluef64256x256108.517 ms1.550.0115%ok
eigenschurf64256x256266.236 ms1.580.0015%ok
eigensvdf64256x256301.331 ms1.220.0012%ok
eigenexpmf64256x25685.762 ms2.930.0128%ok
eigensylvesterf64256x256609.242 ms1.650.0016%ok
elementwiseexpf64n=8388608151.803 ms0.060.889%ok
elementwiselogf64n=8388608150.937 ms0.060.899%ok
elementwisesinf64n=8388608172.986 ms0.050.788%ok
elementwisesqrtf64n=838860890.716 ms0.091.4814%ok
elementwiseerff64n=83886082478.330 ms0.000.051%ok
sparsespmvf64512^2 grid, nnz 13086722.463 ms1.0611.06108%ok
sparsespmv_transf64512^2 grid, nnz 13086723.079 ms0.858.8587%ok
sparsecrs_densef64512^2 grid, nnz 1308672 x 1615.127 ms2.775.8257%ok
sparsespgemmf64256^2 grid, nnz 32665616.424 ms0.200.707%*ok
sparselower_solve_levelf64192^2 grid, nnz(L) 10035645.777 ms0.352.8828%ok
directsparse_choleskyf6496^2 grid, nnz 45696274.115 ms0.040.000%ok
directsparse_ldltf6496^2 grid, nnz 4569669.604 ms0.160.012%ok
directsupernodal_ldltf6496^2 grid, nnz 4569626.544 ms0.420.034%ok
directsparse_luf6496^2 grid, nnz 45696116.264 ms0.190.012%ok
directsupernodal_luf6496^2 grid, nnz 45696153.213 ms0.150.001%ok
directnative_kluf6496^2 grid, nnz 4569647.885 ms0.470.024%ok
directsparse_qrf6448^2 grid, nnz 11328231.569 ms0.010.000%ok
orderingamdf6496^2 grid, nnz 45696, scrambled8.966 ms0.000.00n/aok
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled12.318 ms0.000.00n/aok
orderingrcmf6496^2 grid, nnz 45696, scrambled0.631 ms0.000.00n/aok
orderingblock_triangular_formf6432 planted blocks, n 819240.956 ms0.000.00n/aok
krylovcg+ic0f64128^2 grid, n 16384231.752 ms0.000.00n/aok
krylovminres+diagf64128^2 grid, n 16384393.721 ms0.000.00n/aok
krylovbicg+ilu0f64128^2 grid, n 16384126.980 ms0.000.00n/aok
krylovbicgstab+ilu0f64128^2 grid, n 16384101.953 ms0.000.00n/aok
krylovbicgstab_ell+ilu0f64128^2 grid, n 16384121.936 ms0.000.00n/aok
krylovcgs+ilu0f64128^2 grid, n 1638479.133 ms0.000.00n/aok
krylovgmres30+ilu0f64128^2 grid, n 16384519.380 ms0.000.00n/aok
krylovqmr+ilu0f64128^2 grid, n 16384175.151 ms0.000.00n/aok
krylovtfqmr+ilu0f64128^2 grid, n 1638490.825 ms0.000.00n/aok
krylovidr_s1+ilu0f64128^2 grid, n 16384136.209 ms0.000.00n/aok
krylovidr_s2+ilu0f64128^2 grid, n 16384137.658 ms0.000.00n/aok
krylovidr_s4+ilu0f64128^2 grid, n 16384170.841 ms0.000.00n/aok
krylovidr_s8+ilu0f64128^2 grid, n 16384249.427 ms0.000.00n/aok
pcdiagonalf64128^2 grid, rows scaled0.127 ms0.000.00n/aok
pcilu_0f64128^2 grid4.393 ms0.000.00n/aok
pcilutf64128^2 grid34.283 ms0.000.00n/aok
pcic_0f64128^2 grid2.861 ms0.000.00n/aok
pcildlf64128^2 grid2.717 ms0.000.00n/aok
pcssorf64128^2 grid0.075 ms0.000.00n/aok
pcblock_diagonal4f64128^2 grid, rows scaled2.061 ms0.000.00n/aok
smootherjacobif64512^2 grid, nnz 13086723.653 ms0.728.0379%ok
smoothergauss_seidelf64512^2 grid, nnz 13086723.109 ms0.849.4392%ok
smoothersor1.5f64512^2 grid, nnz 13086723.804 ms0.697.7176%ok
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)35.903 ms1.078.1079%ok
sparse_eiglanczosf6464x65 Laplacian, k=4, m=80140.896 ms0.000.00n/aok
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=80189.695 ms0.000.00n/aok
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=40147.248 ms0.000.00n/aok
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps42.413 ms0.000.00n/aok
mixeddotf32>f64n=8388608187.790 ms0.090.364%ok
mixedgemmf32>f64512x512x512113.327 ms2.370.0423%ok
mixedblock_scaled_doti8+f32n=8388608, block 322.237 ms7.507.9778%ok
integerdoti32n=8388608, |v| <= 156.469 ms2.5910.37n/aok
integerdoti16>i32n=8388608, |v| <= 153.159 ms5.3110.62n/aok
integerdoti8>i32n=8388608, |v| <= 155.010 ms3.353.35n/aok
integerdotu8i8>i32n=8388608, |v| <= 155.094 ms3.293.29n/aok
integergemmi32512^3, |v| <= 204757.725 ms4.650.05n/aok
integergemmi16>i32512^3, |v| <= 204745.202 ms5.940.05n/aok
integergemmi8>i32512^3, |v| <= 12723.766 ms11.290.07n/aok
integergemm_quadi8>i32512^3, |v| <= 12723.701 ms11.330.07n/aok
integergemm_quadu8i8>i32512^3, |v| <= 12771.276 ms3.770.02n/aok

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

xeon-e5-2420v2 — native, 6 threads, class medium

Section titled “xeon-e5-2420v2 — native, 6 threads, class medium”
CPUIntel(R) Xeon(R) CPU E5-2420 v2 @ 2.20GHz (x86_64, 12 logical cores)
OSLinux Ubuntu 26.04.1 LTS (7.0.0-43-generic)
compilerGCC 15.2.0
build -O3 -DNDEBUG, SIMD scalar
commitbfa1c2822688
FMA peak (ceiling)f64 59.0 GFLOP/s, f32 118.1 GFLOP/s
… hardware, SSE2f64 59.0, f32 118.1 GFLOP/s
… this build, scalarf64 0.5, f32 3.5 GFLOP/s
bandwidthread 11.2, update 10.5, triad 8.4 GB/s (ceiling 11.2)
last-level cache15 MiB
areaentriesmedian %roofbest %rooffailed checksmedian parallel efficiency
blas11055%108%083%
blas21088%109%017%
blas3108%39%017%
factor123%38%020%
eigen83%11%017%
elementwise58%13%017%
sparse569%105%017%
direct70%1%017%
ordering4n/an/a017%
krylov13n/an/a017%
pc7n/an/a016%
smoother373%83%017%
mg174%74%017%
sparse_eig4n/an/a017%
mixed322%68%017%
integer9n/an/a090%
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheckspeedup (efficiency)
blas1dotf64n=838860832.893 ms0.514.0836%ok5.6x (93%)
blas1dotf32n=83886085.533 ms3.0312.13108%ok5.0x (83%)
blas1axpyf64n=838860838.924 ms0.435.1746%ok5.6x (93%)
blas1axpyf32n=838860816.377 ms1.026.1555%ok5.6x (93%)
blas1scalef64n=838860812.079 ms0.6911.1199%ok1.0x (17%)
blas1two_normf64n=838860832.925 ms0.512.0418%ok5.6x (94%)
blas1one_normf64n=83886089.769 ms0.866.8761%ok1.0x (17%)
blas1infinity_normf64n=838860812.997 ms0.655.1646%ok1.0x (17%)
blas1sumf64n=83886089.761 ms0.866.8861%ok1.0x (17%)
blas1dot_k3f64n=8388608280.150 ms0.660.484%ok1.0x (17%)
blas2gemvf64n=409610.963 ms3.0612.25109%ok1.2x (20%)
blas2gemvf32n=40965.509 ms6.0912.19109%ok1.5x (25%)
blas2gemv_colmajorf64n=409610.956 ms3.0612.26109%ok1.2x (20%)
blas2gemv_colmajorf32n=40965.503 ms6.1012.20109%ok1.2x (20%)
blas2symvf64n=409616.195 ms2.074.1537%ok1.0x (17%)
blas2trmvf64n=40967.049 ms2.389.5385%ok1.0x (17%)
blas2trmvf32n=40964.725 ms3.557.1163%ok1.0x (16%)
blas2trsvf64n=40966.824 ms2.469.8488%ok1.0x (17%)
blas2trsvf32n=40964.622 ms3.637.2765%ok0.9x (16%)
blas2gerf64n=409627.852 ms1.209.6486%ok1.0x (17%)
blas3gemmf64512x512x51211.545 ms23.250.5439%ok5.6x (93%)
blas3gemmf32512x512x5126.427 ms41.770.4935%ok5.4x (90%)
blas3gemmtf64512x512x51229.382 ms4.570.218%ok1.0x (16%)
blas3symmf64512x512x51260.597 ms4.430.108%ok1.0x (16%)
blas3syrkf64512x512x51238.015 ms3.530.116%ok1.0x (17%)
blas3syr2kf64512x512x51265.983 ms4.070.107%ok1.0x (16%)
blas3trmmf64512x512x51232.931 ms4.080.167%ok1.0x (17%)
blas3trsmf64512x512x51232.632 ms4.110.167%ok1.0x (17%)
blas3trsmf32512x512x51216.407 ms8.180.167%ok1.0x (16%)
blas3kronf6448x48 (x) 48x4838.690 ms0.141.1010%ok1.0x (17%)
factorluf64512x51223.852 ms3.750.186%ok1.3x (22%)
factorluf32512x51213.991 ms6.400.155%ok1.2x (20%)
factorcholeskyf64512x51225.283 ms1.770.173%ok1.0x (16%)
factorcholeskyf32512x51223.800 ms1.880.092%ok1.0x (17%)
factorldltf64512x51225.103 ms1.780.173%ok1.0x (16%)
factorldlt_bkf64512x512139.070 ms0.320.031%ok1.1x (18%)
factorqrf64512x512323.622 ms0.550.011%ok1.6x (27%)
factorlqf64512x51251.363 ms3.480.086%ok1.4x (24%)
factorhessenbergf64512x512417.343 ms1.070.012%ok1.5x (25%)
factorhouseholderf64512x5120.985 ms1.074.2638%*ok3.3x (54%)
factorinvf64512x512143.911 ms1.870.033%ok1.1x (18%)
factorlu_iterative_refinef32>f64512x51219.420 ms4.610.228%ok1.1x (19%)
eigeneigenvalue_symmetricf64256x25663.741 ms0.350.021%ok1.0x (16%)
eigeneigen_symmetricf64256x256172.253 ms0.880.011%ok1.0x (17%)
eigeneigen_symmetric_jacobif64256x256996.068 ms0.150.000%ok1.0x (17%)
eigeneigenvaluef64256x256107.492 ms1.560.013%ok1.0x (17%)
eigenschurf64256x256264.391 ms1.590.003%ok1.0x (17%)
eigensvdf64256x256302.720 ms1.220.002%ok1.0x (17%)
eigenexpmf64256x25637.541 ms6.700.0311%ok2.3x (38%)
eigensylvesterf64256x256582.347 ms1.730.003%ok1.0x (17%)
elementwiseexpf64n=8388608151.636 ms0.060.898%ok1.0x (17%)
elementwiselogf64n=8388608142.557 ms0.060.948%ok1.1x (18%)
elementwisesinf64n=8388608172.453 ms0.050.787%ok1.0x (17%)
elementwisesqrtf64n=838860892.119 ms0.091.4613%ok1.0x (16%)
elementwiseerff64n=83886082481.949 ms0.000.050%ok1.0x (17%)
sparsespmvf64512^2 grid, nnz 13086722.303 ms1.1411.82105%ok1.1x (18%)
sparsespmv_transf64512^2 grid, nnz 13086723.112 ms0.848.7578%ok1.0x (16%)
sparsecrs_densef64512^2 grid, nnz 1308672 x 1611.353 ms3.697.7669%ok1.3x (22%)
sparsespgemmf64256^2 grid, nnz 32665616.053 ms0.200.726%*ok1.0x (17%)
sparselower_solve_levelf64192^2 grid, nnz(L) 10035645.770 ms0.352.8926%ok1.0x (17%)
directsparse_choleskyf6496^2 grid, nnz 45696273.553 ms0.040.000%ok1.0x (17%)
directsparse_ldltf6496^2 grid, nnz 4569669.419 ms0.160.010%ok1.0x (17%)
directsupernodal_ldltf6496^2 grid, nnz 4569626.388 ms0.420.031%ok1.0x (17%)
directsparse_luf6496^2 grid, nnz 45696115.723 ms0.190.010%ok1.0x (17%)
directsupernodal_luf6496^2 grid, nnz 45696153.331 ms0.150.000%ok1.0x (17%)
directnative_kluf6496^2 grid, nnz 4569648.018 ms0.470.021%ok1.0x (17%)
directsparse_qrf6448^2 grid, nnz 11328231.533 ms0.010.000%ok1.0x (17%)
orderingamdf6496^2 grid, nnz 45696, scrambled8.909 ms0.000.00n/aok1.0x (17%)
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled12.274 ms0.000.00n/aok1.0x (17%)
orderingrcmf6496^2 grid, nnz 45696, scrambled0.632 ms0.000.00n/aok1.0x (17%)
orderingblock_triangular_formf6432 planted blocks, n 819240.676 ms0.000.00n/aok1.0x (17%)
krylovcg+ic0f64128^2 grid, n 16384233.678 ms0.000.00n/aok1.0x (17%)
krylovminres+diagf64128^2 grid, n 16384394.637 ms0.000.00n/aok1.0x (17%)
krylovbicg+ilu0f64128^2 grid, n 16384126.593 ms0.000.00n/aok1.0x (17%)
krylovbicgstab+ilu0f64128^2 grid, n 16384101.729 ms0.000.00n/aok1.0x (17%)
krylovbicgstab_ell+ilu0f64128^2 grid, n 16384121.185 ms0.000.00n/aok1.0x (17%)
krylovcgs+ilu0f64128^2 grid, n 1638479.198 ms0.000.00n/aok1.0x (17%)
krylovgmres30+ilu0f64128^2 grid, n 16384520.399 ms0.000.00n/aok1.0x (17%)
krylovqmr+ilu0f64128^2 grid, n 16384174.551 ms0.000.00n/aok1.0x (17%)
krylovtfqmr+ilu0f64128^2 grid, n 1638491.725 ms0.000.00n/aok1.0x (17%)
krylovidr_s1+ilu0f64128^2 grid, n 16384137.763 ms0.000.00n/aok1.0x (16%)
krylovidr_s2+ilu0f64128^2 grid, n 16384139.507 ms0.000.00n/aok1.0x (16%)
krylovidr_s4+ilu0f64128^2 grid, n 16384171.042 ms0.000.00n/aok1.0x (17%)
krylovidr_s8+ilu0f64128^2 grid, n 16384251.140 ms0.000.00n/aok1.0x (17%)
pcdiagonalf64128^2 grid, rows scaled0.129 ms0.000.00n/aok1.0x (16%)
pcilu_0f64128^2 grid4.471 ms0.000.00n/aok1.0x (16%)
pcilutf64128^2 grid33.093 ms0.000.00n/aok1.0x (17%)
pcic_0f64128^2 grid2.919 ms0.000.00n/aok1.0x (16%)
pcildlf64128^2 grid2.767 ms0.000.00n/aok1.0x (16%)
pcssorf64128^2 grid0.079 ms0.000.00n/aok0.9x (16%)
pcblock_diagonal4f64128^2 grid, rows scaled2.182 ms0.000.00n/aok0.9x (16%)
smootherjacobif64512^2 grid, nnz 13086723.587 ms0.738.1873%ok1.0x (17%)
smoothergauss_seidelf64512^2 grid, nnz 13086723.153 ms0.839.3083%ok1.0x (16%)
smoothersor1.5f64512^2 grid, nnz 13086723.808 ms0.697.7069%ok1.0x (17%)
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)34.898 ms1.108.3374%ok1.0x (17%)
sparse_eiglanczosf6464x65 Laplacian, k=4, m=80146.722 ms0.000.00n/aok1.0x (16%)
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=80188.201 ms0.000.00n/aok1.0x (17%)
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=40146.308 ms0.000.00n/aok1.0x (17%)
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps42.820 ms0.000.00n/aok1.0x (17%)
mixeddotf32>f64n=8388608188.901 ms0.090.363%ok1.0x (17%)
mixedgemmf32>f64512x512x51220.564 ms13.050.2022%ok5.5x (92%)
mixedblock_scaled_doti8+f32n=8388608, block 322.336 ms7.187.6368%ok1.0x (16%)
integerdoti32n=8388608, |v| <= 156.485 ms2.5910.35n/aok1.0x (17%)
integerdoti16>i32n=8388608, |v| <= 153.108 ms5.4010.79n/aok1.0x (17%)
integerdoti8>i32n=8388608, |v| <= 155.081 ms3.303.30n/aok1.0x (16%)
integerdotu8i8>i32n=8388608, |v| <= 154.987 ms3.363.36n/aok1.0x (17%)
integergemmi32512^3, |v| <= 204710.387 ms25.840.30n/aok5.6x (93%)
integergemmi16>i32512^3, |v| <= 20478.249 ms32.540.25n/aok5.5x (91%)
integergemmi8>i32512^3, |v| <= 1274.348 ms61.730.36n/aok5.5x (91%)
integergemm_quadi8>i32512^3, |v| <= 1274.350 ms61.710.36n/aok5.4x (91%)
integergemm_quadu8i8>i32512^3, |v| <= 12713.148 ms20.420.12n/aok5.4x (90%)

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

xeon-e5-2420v2 — native-fast, 1 thread, class medium

Section titled “xeon-e5-2420v2 — native-fast, 1 thread, class medium”
CPUIntel(R) Xeon(R) CPU E5-2420 v2 @ 2.20GHz (x86_64, 12 logical cores)
OSLinux Ubuntu 26.04.1 LTS (7.0.0-43-generic)
compilerGCC 15.2.0
build -O3 -DNDEBUG [MTL5_NATIVE_ARCH=ON], SIMD SSE4
commitbfa1c2822688
FMA peak (ceiling)f64 10.5 GFLOP/s, f32 21.1 GFLOP/s
… hardware, SSE2f64 10.5, f32 21.1 GFLOP/s
… this build, SSE4f64 10.5, f32 21.0 GFLOP/s
bandwidthread 10.4, update 10.4, triad 8.1 GB/s (ceiling 10.4)
last-level cache15 MiB
areaentriesmedian %roofbest %rooffailed checks
blas110101%108%0
blas21091%100%0
blas31072%94%0
factor1217%40%0
eigen815%47%0
elementwise59%14%0
sparse556%106%0
direct71%4%0
ordering4n/an/a0
krylov13n/an/a0
pc7n/an/a0
smoother379%90%0
mg178%78%0
sparse_eig4n/an/a0
mixed375%89%0
integer9n/an/a0
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheck
blas1dotf64n=838860811.950 ms1.4011.23108%ok
blas1dotf32n=83886085.963 ms2.8111.25108%ok
blas1axpyf64n=838860818.827 ms0.8910.69102%ok
blas1axpyf32n=83886089.019 ms1.8611.16107%ok
blas1scalef64n=838860813.211 ms0.6310.1697%ok
blas1two_normf64n=83886086.388 ms2.6310.51101%ok
blas1one_normf64n=838860810.333 ms0.816.4962%ok
blas1infinity_normf64n=838860813.777 ms0.614.8747%ok
blas1sumf64n=838860810.209 ms0.826.5763%ok
blas1dot_k3f64n=8388608272.662 ms0.680.496%ok
blas2gemvf64n=409613.273 ms2.5310.1297%ok
blas2gemvf32n=40966.450 ms5.2010.41100%ok
blas2gemv_colmajorf64n=409640.732 ms0.823.3032%ok
blas2gemv_colmajorf32n=409617.404 ms1.933.8637%ok
blas2symvf64n=409616.138 ms2.084.1640%ok
blas2trmvf64n=40967.059 ms2.389.5291%ok
blas2trmvf32n=40963.526 ms4.769.5391%ok
blas2trsvf64n=40967.066 ms2.379.5191%ok
blas2trsvf32n=40963.614 ms4.649.2989%ok
blas2gerf64n=409628.123 ms1.199.5591%ok
blas3gemmf64512x512x51227.725 ms9.680.2392%ok
blas3gemmf32512x512x51213.517 ms19.860.2394%ok
blas3gemmtf64512x512x51216.977 ms7.910.3775%ok
blas3symmf64512x512x51228.856 ms9.300.2288%ok
blas3syrkf64512x512x51236.868 ms3.640.1135%ok
blas3syr2kf64512x512x51263.609 ms4.220.1040%ok
blas3trmmf64512x512x51217.787 ms7.550.2972%ok
blas3trsmf64512x512x51218.178 ms7.380.2970%ok
blas3trsmf32512x512x5128.881 ms15.110.3072%ok
blas3kronf6448x48 (x) 48x4838.540 ms0.141.1011%ok
factorluf64512x51229.596 ms3.020.1429%ok
factorluf32512x51216.040 ms5.580.1327%ok
factorcholeskyf64512x51225.624 ms1.750.1617%ok
factorcholeskyf32512x51224.180 ms1.850.099%ok
factorldltf64512x51225.865 ms1.730.1616%ok
factorldlt_bkf64512x512134.445 ms0.330.033%ok
factorqrf64512x512497.393 ms0.360.013%ok
factorlqf64512x51274.031 ms2.420.0623%ok
factorhessenbergf64512x512614.277 ms0.730.017%ok
factorhouseholderf64512x5123.170 ms0.331.3213%*ok
factorinvf64512x512153.982 ms1.740.0317%ok
factorlu_iterative_refinef32>f64512x51221.144 ms4.230.2040%ok
eigeneigenvalue_symmetricf64256x25660.777 ms0.370.023%ok
eigeneigen_symmetricf64256x256173.587 ms0.870.018%ok
eigeneigen_symmetric_jacobif64256x2561002.472 ms0.150.001%ok
eigeneigenvaluef64256x256104.061 ms1.610.0115%ok
eigenschurf64256x256254.869 ms1.650.0016%ok
eigensvdf64256x256259.287 ms1.420.0014%ok
eigenexpmf64256x25650.632 ms4.970.0247%ok
eigensylvesterf64256x256575.084 ms1.750.0017%ok
elementwiseexpf64n=8388608151.136 ms0.060.899%ok
elementwiselogf64n=8388608140.081 ms0.060.969%ok
elementwisesinf64n=8388608171.322 ms0.050.788%ok
elementwisesqrtf64n=838860890.254 ms0.091.4914%ok
elementwiseerff64n=83886082470.162 ms0.000.051%ok
sparsespmvf64512^2 grid, nnz 13086722.464 ms1.0611.05106%ok
sparsespmv_transf64512^2 grid, nnz 13086724.124 ms0.636.6063%ok
sparsecrs_densef64512^2 grid, nnz 1308672 x 1615.072 ms2.785.8456%ok
sparsespgemmf64256^2 grid, nnz 32665616.446 ms0.200.707%*ok
sparselower_solve_levelf64192^2 grid, nnz(L) 10035645.822 ms0.342.8627%ok
directsparse_choleskyf6496^2 grid, nnz 45696238.908 ms0.050.000%ok
directsparse_ldltf6496^2 grid, nnz 4569671.511 ms0.160.011%ok
directsupernodal_ldltf6496^2 grid, nnz 4569627.030 ms0.410.034%ok
directsparse_luf6496^2 grid, nnz 45696117.001 ms0.190.012%ok
directsupernodal_luf6496^2 grid, nnz 45696154.406 ms0.140.001%ok
directnative_kluf6496^2 grid, nnz 4569648.760 ms0.460.014%ok
directsparse_qrf6448^2 grid, nnz 11328234.125 ms0.010.000%ok
orderingamdf6496^2 grid, nnz 45696, scrambled9.063 ms0.000.00n/aok
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled12.162 ms0.000.00n/aok
orderingrcmf6496^2 grid, nnz 45696, scrambled0.620 ms0.000.00n/aok
orderingblock_triangular_formf6432 planted blocks, n 819240.620 ms0.000.00n/aok
krylovcg+ic0f64128^2 grid, n 1638479.629 ms0.000.00n/aok
krylovminres+diagf64128^2 grid, n 16384110.372 ms0.000.00n/aok
krylovbicg+ilu0f64128^2 grid, n 1638464.425 ms0.000.00n/aok
krylovbicgstab+ilu0f64128^2 grid, n 1638433.612 ms0.000.00n/aok
krylovbicgstab_ell+ilu0f64128^2 grid, n 1638437.815 ms0.000.00n/aok
krylovcgs+ilu0f64128^2 grid, n 1638436.870 ms0.000.00n/aok
krylovgmres30+ilu0f64128^2 grid, n 1638468.906 ms0.000.00n/aok
krylovqmr+ilu0f64128^2 grid, n 1638474.107 ms0.000.00n/aok
krylovtfqmr+ilu0f64128^2 grid, n 1638437.183 ms0.000.00n/aok
krylovidr_s1+ilu0f64128^2 grid, n 1638443.287 ms0.000.00n/aok
krylovidr_s2+ilu0f64128^2 grid, n 1638441.879 ms0.000.00n/aok
krylovidr_s4+ilu0f64128^2 grid, n 1638444.653 ms0.000.00n/aok
krylovidr_s8+ilu0f64128^2 grid, n 1638453.810 ms0.000.00n/aok
pcdiagonalf64128^2 grid, rows scaled0.129 ms0.000.00n/aok
pcilu_0f64128^2 grid4.476 ms0.000.00n/aok
pcilutf64128^2 grid33.710 ms0.000.00n/aok
pcic_0f64128^2 grid2.866 ms0.000.00n/aok
pcildlf64128^2 grid2.771 ms0.000.00n/aok
pcssorf64128^2 grid0.075 ms0.000.00n/aok
pcblock_diagonal4f64128^2 grid, rows scaled2.068 ms0.000.00n/aok
smootherjacobif64512^2 grid, nnz 13086723.578 ms0.738.2079%ok
smoothergauss_seidelf64512^2 grid, nnz 13086723.113 ms0.849.4290%ok
smoothersor1.5f64512^2 grid, nnz 13086723.799 ms0.697.7274%ok
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)35.497 ms1.088.1978%ok
sparse_eiglanczosf6464x65 Laplacian, k=4, m=80139.647 ms0.000.00n/aok
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=80181.001 ms0.000.00n/aok
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=40152.067 ms0.000.00n/aok
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps16.329 ms0.000.00n/aok
mixeddotf32>f64n=83886087.204 ms2.339.3289%ok
mixedgemmf32>f64512x512x51262.449 ms4.300.0741%ok
mixedblock_scaled_doti8+f32n=8388608, block 322.266 ms7.407.8775%ok
integerdoti32n=8388608, |v| <= 156.039 ms2.7811.11n/aok
integerdoti16>i32n=8388608, |v| <= 152.985 ms5.6211.24n/aok
integerdoti8>i32n=8388608, |v| <= 152.085 ms8.058.05n/aok
integerdotu8i8>i32n=8388608, |v| <= 151.677 ms10.0110.01n/aok
integergemmi32512^3, |v| <= 204718.952 ms14.160.17n/aok
integergemmi16>i32512^3, |v| <= 204722.131 ms12.130.09n/aok
integergemmi8>i32512^3, |v| <= 12722.000 ms12.200.07n/aok
integergemm_quadi8>i32512^3, |v| <= 12716.194 ms16.580.10n/aok
integergemm_quadu8i8>i32512^3, |v| <= 12712.365 ms21.710.13n/aok

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

xeon-e5-2420v2 — native-fast, 6 threads, class medium

Section titled “xeon-e5-2420v2 — native-fast, 6 threads, class medium”
CPUIntel(R) Xeon(R) CPU E5-2420 v2 @ 2.20GHz (x86_64, 12 logical cores)
OSLinux Ubuntu 26.04.1 LTS (7.0.0-43-generic)
compilerGCC 15.2.0
build -O3 -DNDEBUG [MTL5_NATIVE_ARCH=ON], SIMD SSE4
commitbfa1c2822688
FMA peak (ceiling)f64 59.1 GFLOP/s, f32 118.2 GFLOP/s
… hardware, SSE2f64 59.1, f32 118.2 GFLOP/s
… this build, SSE4f64 59.0, f32 118.1 GFLOP/s
bandwidthread 11.4, update 10.7, triad 8.5 GB/s (ceiling 11.4)
last-level cache15 MiB
areaentriesmedian %roofbest %rooffailed checksmedian parallel efficiency
blas11098%108%018%
blas21084%107%017%
blas31028%78%036%
factor123%37%019%
eigen83%13%017%
elementwise58%13%017%
sparse558%103%017%
direct70%1%017%
ordering4n/an/a017%
krylov13n/an/a017%
pc7n/an/a017%
smoother372%82%017%
mg162%62%014%
sparse_eig4n/an/a017%
mixed369%82%017%
integer9n/an/a079%
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheckspeedup (efficiency)
blas1dotf64n=838860810.905 ms1.5412.31108%ok1.1x (18%)
blas1dotf32n=83886085.494 ms3.0512.21107%ok1.1x (18%)
blas1axpyf64n=838860817.914 ms0.9411.2498%ok1.1x (18%)
blas1axpyf32n=83886088.469 ms1.9811.89104%ok1.1x (18%)
blas1scalef64n=838860812.674 ms0.6610.5993%ok1.0x (17%)
blas1two_normf64n=83886085.459 ms3.0712.29108%ok1.2x (20%)
blas1one_normf64n=83886089.865 ms0.856.8060%ok1.0x (17%)
blas1infinity_normf64n=838860813.055 ms0.645.1445%ok1.1x (18%)
blas1sumf64n=83886089.830 ms0.856.8360%ok1.0x (17%)
blas1dot_k3f64n=8388608273.778 ms0.670.494%ok1.0x (17%)
blas2gemvf64n=409610.938 ms3.0712.28107%ok1.2x (20%)
blas2gemvf32n=40965.483 ms6.1212.25107%ok1.2x (20%)
blas2gemv_colmajorf64n=409621.167 ms1.596.3456%ok1.9x (32%)
blas2gemv_colmajorf32n=409611.157 ms3.016.0253%ok1.6x (26%)
blas2symvf64n=409616.031 ms2.094.1937%ok1.0x (17%)
blas2trmvf64n=40966.855 ms2.459.8086%ok1.0x (17%)
blas2trmvf32n=40963.599 ms4.669.3382%ok1.0x (16%)
blas2trsvf64n=40966.866 ms2.449.7886%ok1.0x (17%)
blas2trsvf32n=40963.587 ms4.689.3682%ok1.0x (17%)
blas2gerf64n=409627.944 ms1.209.6184%ok1.0x (17%)
blas3gemmf64512x512x5125.933 ms45.251.0677%ok4.7x (78%)
blas3gemmf32512x512x5122.909 ms92.271.0878%ok4.6x (77%)
blas3gemmtf64512x512x5125.321 ms25.221.1843%ok3.2x (53%)
blas3symmf64512x512x5126.897 ms38.920.9166%ok4.2x (70%)
blas3syrkf64512x512x51237.050 ms3.620.116%ok1.0x (17%)
blas3syr2kf64512x512x51264.310 ms4.170.107%ok1.0x (16%)
blas3trmmf64512x512x5128.166 ms16.440.6428%ok2.2x (36%)
blas3trsmf64512x512x5128.943 ms15.010.5925%ok2.0x (34%)
blas3trsmf32512x512x5124.718 ms28.450.5624%ok1.9x (31%)
blas3kronf6448x48 (x) 48x4838.408 ms0.141.1110%ok1.0x (17%)
factorluf64512x51223.113 ms3.870.187%ok1.3x (21%)
factorluf32512x51213.575 ms6.590.156%ok1.2x (20%)
factorcholeskyf64512x51225.494 ms1.750.163%ok1.0x (17%)
factorcholeskyf32512x51224.031 ms1.860.092%ok1.0x (17%)
factorldltf64512x51225.768 ms1.740.163%ok1.0x (17%)
factorldlt_bkf64512x512133.132 ms0.340.031%ok1.0x (17%)
factorqrf64512x512439.443 ms0.410.011%ok1.1x (19%)
factorlqf64512x51250.588 ms3.540.086%ok1.5x (24%)
factorhessenbergf64512x512406.816 ms1.100.012%ok1.5x (25%)
factorhouseholderf64512x5120.981 ms1.074.2737%*ok3.2x (54%)
factorinvf64512x512148.554 ms1.810.033%ok1.0x (17%)
factorlu_iterative_refinef32>f64512x51219.093 ms4.690.228%ok1.1x (18%)
eigeneigenvalue_symmetricf64256x25660.932 ms0.370.021%ok1.0x (17%)
eigeneigen_symmetricf64256x256171.823 ms0.880.011%ok1.0x (17%)
eigeneigen_symmetric_jacobif64256x256988.132 ms0.150.000%ok1.0x (17%)
eigeneigenvaluef64256x256105.251 ms1.590.013%ok1.0x (16%)
eigenschurf64256x256259.080 ms1.620.003%ok1.0x (16%)
eigensvdf64256x256260.619 ms1.420.002%ok1.0x (17%)
eigenexpmf64256x25632.297 ms7.790.0313%ok1.6x (26%)
eigensylvesterf64256x256564.506 ms1.780.003%ok1.0x (17%)
elementwiseexpf64n=8388608153.676 ms0.050.878%ok1.0x (16%)
elementwiselogf64n=8388608141.178 ms0.060.958%ok1.0x (17%)
elementwisesinf64n=8388608171.237 ms0.050.787%ok1.0x (17%)
elementwisesqrtf64n=838860892.044 ms0.091.4613%ok1.0x (16%)
elementwiseerff64n=83886082472.458 ms0.000.050%ok1.0x (17%)
sparsespmvf64512^2 grid, nnz 13086722.312 ms1.1311.78103%ok1.1x (18%)
sparsespmv_transf64512^2 grid, nnz 13086724.104 ms0.646.6458%ok1.0x (17%)
sparsecrs_densef64512^2 grid, nnz 1308672 x 1611.378 ms3.687.7468%ok1.3x (22%)
sparsespgemmf64256^2 grid, nnz 32665616.364 ms0.200.706%*ok1.0x (17%)
sparselower_solve_levelf64192^2 grid, nnz(L) 10035645.800 ms0.352.8725%ok1.0x (17%)
directsparse_choleskyf6496^2 grid, nnz 45696240.712 ms0.050.000%ok1.0x (17%)
directsparse_ldltf6496^2 grid, nnz 4569670.354 ms0.160.010%ok1.0x (17%)
directsupernodal_ldltf6496^2 grid, nnz 4569626.307 ms0.430.031%ok1.0x (17%)
directsparse_luf6496^2 grid, nnz 45696117.408 ms0.190.010%ok1.0x (17%)
directsupernodal_luf6496^2 grid, nnz 45696156.562 ms0.140.000%ok1.0x (16%)
directnative_kluf6496^2 grid, nnz 4569648.816 ms0.460.011%ok1.0x (17%)
directsparse_qrf6448^2 grid, nnz 11328226.836 ms0.010.000%ok1.0x (17%)
orderingamdf6496^2 grid, nnz 45696, scrambled9.045 ms0.000.00n/aok1.0x (17%)
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled12.353 ms0.000.00n/aok1.0x (16%)
orderingrcmf6496^2 grid, nnz 45696, scrambled0.631 ms0.000.00n/aok1.0x (16%)
orderingblock_triangular_formf6432 planted blocks, n 819240.744 ms0.000.00n/aok1.0x (17%)
krylovcg+ic0f64128^2 grid, n 1638480.159 ms0.000.00n/aok1.0x (17%)
krylovminres+diagf64128^2 grid, n 16384111.355 ms0.000.00n/aok1.0x (17%)
krylovbicg+ilu0f64128^2 grid, n 1638464.214 ms0.000.00n/aok1.0x (17%)
krylovbicgstab+ilu0f64128^2 grid, n 1638431.644 ms0.000.00n/aok1.1x (18%)
krylovbicgstab_ell+ilu0f64128^2 grid, n 1638438.311 ms0.000.00n/aok1.0x (16%)
krylovcgs+ilu0f64128^2 grid, n 1638437.171 ms0.000.00n/aok1.0x (17%)
krylovgmres30+ilu0f64128^2 grid, n 1638466.202 ms0.000.00n/aok1.0x (17%)
krylovqmr+ilu0f64128^2 grid, n 1638474.367 ms0.000.00n/aok1.0x (17%)
krylovtfqmr+ilu0f64128^2 grid, n 1638437.722 ms0.000.00n/aok1.0x (16%)
krylovidr_s1+ilu0f64128^2 grid, n 1638496.788 ms0.000.00n/aok0.4x (7%)
krylovidr_s2+ilu0f64128^2 grid, n 1638441.768 ms0.000.00n/aok1.0x (17%)
krylovidr_s4+ilu0f64128^2 grid, n 1638447.106 ms0.000.00n/aok0.9x (16%)
krylovidr_s8+ilu0f64128^2 grid, n 1638455.228 ms0.000.00n/aok1.0x (16%)
pcdiagonalf64128^2 grid, rows scaled0.130 ms0.000.00n/aok1.0x (17%)
pcilu_0f64128^2 grid4.458 ms0.000.00n/aok1.0x (17%)
pcilutf64128^2 grid33.400 ms0.000.00n/aok1.0x (17%)
pcic_0f64128^2 grid2.871 ms0.000.00n/aok1.0x (17%)
pcildlf64128^2 grid2.800 ms0.000.00n/aok1.0x (16%)
pcssorf64128^2 grid0.074 ms0.000.00n/aok1.0x (17%)
pcblock_diagonal4f64128^2 grid, rows scaled2.186 ms0.000.00n/aok0.9x (16%)
smootherjacobif64512^2 grid, nnz 13086723.588 ms0.738.1772%ok1.0x (17%)
smoothergauss_seidelf64512^2 grid, nnz 13086723.134 ms0.849.3682%ok1.0x (17%)
smoothersor1.5f64512^2 grid, nnz 13086723.879 ms0.677.5666%ok1.0x (16%)
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)40.859 ms0.947.1162%ok0.9x (14%)
sparse_eiglanczosf6464x65 Laplacian, k=4, m=80139.520 ms0.000.00n/aok1.0x (17%)
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=80180.686 ms0.000.00n/aok1.0x (17%)
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=40149.311 ms0.000.00n/aok1.0x (17%)
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps16.176 ms0.000.00n/aok1.0x (17%)
mixeddotf32>f64n=83886087.170 ms2.349.3682%ok1.0x (17%)
mixedgemmf32>f64512x512x51211.774 ms22.800.3639%ok5.3x (88%)
mixedblock_scaled_doti8+f32n=8388608, block 322.266 ms7.407.8769%ok1.0x (17%)
integerdoti32n=8388608, |v| <= 156.024 ms2.7911.14n/aok1.0x (17%)
integerdoti16>i32n=8388608, |v| <= 152.998 ms5.6011.19n/aok1.0x (17%)
integerdoti8>i32n=8388608, |v| <= 152.073 ms8.098.09n/aok1.0x (17%)
integerdotu8i8>i32n=8388608, |v| <= 151.641 ms10.2210.22n/aok1.0x (17%)
integergemmi32512^3, |v| <= 20473.975 ms67.520.79n/aok4.8x (79%)
integergemmi16>i32512^3, |v| <= 20474.553 ms58.950.46n/aok4.9x (81%)
integergemmi8>i32512^3, |v| <= 1274.511 ms59.500.35n/aok4.9x (81%)
integergemm_quadi8>i32512^3, |v| <= 1273.342 ms80.320.47n/aok4.8x (81%)
integergemm_quadu8i8>i32512^3, |v| <= 1272.590 ms103.630.61n/aok4.8x (80%)

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

xeon-e5-2420v2 — blas, 1 thread, class medium

Section titled “xeon-e5-2420v2 — blas, 1 thread, class medium”
CPUIntel(R) Xeon(R) CPU E5-2420 v2 @ 2.20GHz (x86_64, 12 logical cores)
OSLinux Ubuntu 26.04.1 LTS (7.0.0-43-generic)
compilerGCC 15.2.0
build -O3 -DNDEBUG, SIMD scalar
commitbfa1c2822688
FMA peak (ceiling)f64 10.5 GFLOP/s, f32 21.0 GFLOP/s
… hardware, SSE2f64 10.5, f32 21.0 GFLOP/s
… this build, scalarf64 0.1, f32 0.7 GFLOP/s
bandwidthread 10.3, update 10.2, triad 8.1 GB/s (ceiling 10.3)
last-level cache15 MiB
areaentriesmedian %roofbest %rooffailed checks
blas11024%104%0
blas21094%105%0
blas310162%183%0
factor1254%92%0
eigen851%119%0
elementwise59%14%0
sparse556%108%0
direct72%4%0
ordering4n/an/a0
krylov13n/an/a0
pc7n/an/a0
smoother379%92%0
mg179%79%0
sparse_eig4n/an/a0
mixed323%73%0
integer9n/an/a0
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheck
blas1dotf64n=8388608180.998 ms0.090.747%ok
blas1dotf32n=838860827.095 ms0.622.4824%ok
blas1axpyf64n=8388608213.023 ms0.080.959%ok
blas1axpyf32n=838860891.531 ms0.181.1011%ok
blas1scalef64n=838860812.619 ms0.6610.64104%ok
blas1two_normf64n=8388608183.955 ms0.090.364%ok
blas1one_normf64n=83886089.705 ms0.866.9167%ok
blas1infinity_normf64n=838860813.055 ms0.645.1450%ok
blas1sumf64n=83886089.760 ms0.866.8867%ok
blas1dot_k3f64n=8388608277.771 ms0.660.486%ok
blas2gemvf64n=409614.263 ms2.359.4192%ok
blas2gemvf32n=40966.237 ms5.3810.77105%ok
blas2gemv_colmajorf64n=409612.939 ms2.5910.38101%ok
blas2gemv_colmajorf32n=40966.616 ms5.0710.1599%ok
blas2symvf64n=40966.948 ms4.839.6794%ok
blas2trmvf64n=40967.578 ms2.218.8686%ok
blas2trmvf32n=40963.689 ms4.559.1089%ok
blas2trsvf64n=40967.464 ms2.259.0088%ok
blas2trsvf32n=40963.456 ms4.859.7295%ok
blas2gerf64n=409628.043 ms1.209.5793%ok
blas3gemmf64512x512x51214.028 ms19.140.45183%ok
blas3gemmf32512x512x5128.149 ms32.940.39157%ok
blas3gemmtf64512x512x5128.734 ms15.370.72147%ok
blas3symmf64512x512x51214.089 ms19.050.45182%ok
blas3syrkf64512x512x5127.878 ms17.040.53162%ok
blas3syr2kf64512x512x51215.034 ms17.850.42170%ok
blas3trmmf64512x512x5127.459 ms17.990.70172%ok
blas3trsmf64512x512x5128.166 ms16.440.64157%ok
blas3trsmf32512x512x5125.319 ms25.230.49120%ok
blas3kronf6448x48 (x) 48x4838.615 ms0.141.1011%ok
factorluf64512x5129.239 ms9.680.4592%ok
factorluf32512x5126.431 ms13.910.3366%ok
factorcholeskyf64512x5125.891 ms7.590.7172%ok
factorcholeskyf32512x5123.928 ms11.390.5354%ok
factorldltf64512x51225.363 ms1.760.1717%ok
factorldlt_bkf64512x512146.733 ms0.300.033%ok
factorqrf64512x51219.966 ms8.960.2185%ok
factorlqf64512x51274.673 ms2.400.0623%ok
factorhessenbergf64512x512630.391 ms0.710.017%ok
factorhouseholderf64512x5123.230 ms0.331.3013%*ok
factorinvf64512x512129.311 ms2.080.0320%ok
factorlu_iterative_refinef32>f64512x51212.848 ms6.960.3366%ok
eigeneigenvalue_symmetricf64256x2566.239 ms3.590.1734%ok
eigeneigen_symmetricf64256x256174.177 ms0.870.018%ok
eigeneigen_symmetric_jacobif64256x256999.183 ms0.150.001%ok
eigeneigenvaluef64256x25653.177 ms3.150.0230%ok
eigenschurf64256x25672.095 ms5.820.0155%ok
eigensvdf64256x25629.500 ms12.510.04119%ok
eigenexpmf64256x25635.728 ms7.040.0367%ok
eigensylvesterf64256x256189.501 ms5.310.0151%ok
elementwiseexpf64n=8388608150.786 ms0.060.899%ok
elementwiselogf64n=8388608141.186 ms0.060.959%ok
elementwisesinf64n=8388608172.071 ms0.050.788%ok
elementwisesqrtf64n=838860890.985 ms0.091.4814%ok
elementwiseerff64n=83886082476.708 ms0.000.051%ok
sparsespmvf64512^2 grid, nnz 13086722.458 ms1.0611.08108%ok
sparsespmv_transf64512^2 grid, nnz 13086723.097 ms0.858.7986%ok
sparsecrs_densef64512^2 grid, nnz 1308672 x 1615.204 ms2.755.7956%ok
sparsespgemmf64256^2 grid, nnz 32665616.224 ms0.200.717%*ok
sparselower_solve_levelf64192^2 grid, nnz(L) 10035645.776 ms0.352.8828%ok
directsparse_choleskyf6496^2 grid, nnz 45696273.021 ms0.040.000%ok
directsparse_ldltf6496^2 grid, nnz 4569669.556 ms0.160.012%ok
directsupernodal_ldltf6496^2 grid, nnz 4569626.648 ms0.420.034%ok
directsparse_luf6496^2 grid, nnz 45696119.204 ms0.190.012%ok
directsupernodal_luf6496^2 grid, nnz 45696155.239 ms0.140.001%ok
directnative_kluf6496^2 grid, nnz 4569648.043 ms0.470.024%ok
directsparse_qrf6448^2 grid, nnz 11328232.139 ms0.010.000%ok
orderingamdf6496^2 grid, nnz 45696, scrambled9.135 ms0.000.00n/aok
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled12.372 ms0.000.00n/aok
orderingrcmf6496^2 grid, nnz 45696, scrambled0.625 ms0.000.00n/aok
orderingblock_triangular_formf6432 planted blocks, n 819241.388 ms0.000.00n/aok
krylovcg+ic0f64128^2 grid, n 16384229.514 ms0.000.00n/aok
krylovminres+diagf64128^2 grid, n 16384396.802 ms0.000.00n/aok
krylovbicg+ilu0f64128^2 grid, n 16384126.671 ms0.000.00n/aok
krylovbicgstab+ilu0f64128^2 grid, n 16384101.910 ms0.000.00n/aok
krylovbicgstab_ell+ilu0f64128^2 grid, n 16384121.217 ms0.000.00n/aok
krylovcgs+ilu0f64128^2 grid, n 1638478.757 ms0.000.00n/aok
krylovgmres30+ilu0f64128^2 grid, n 16384516.955 ms0.000.00n/aok
krylovqmr+ilu0f64128^2 grid, n 16384176.476 ms0.000.00n/aok
krylovtfqmr+ilu0f64128^2 grid, n 1638491.619 ms0.000.00n/aok
krylovidr_s1+ilu0f64128^2 grid, n 16384139.590 ms0.000.00n/aok
krylovidr_s2+ilu0f64128^2 grid, n 16384140.118 ms0.000.00n/aok
krylovidr_s4+ilu0f64128^2 grid, n 16384170.930 ms0.000.00n/aok
krylovidr_s8+ilu0f64128^2 grid, n 16384248.486 ms0.000.00n/aok
pcdiagonalf64128^2 grid, rows scaled0.128 ms0.000.00n/aok
pcilu_0f64128^2 grid4.456 ms0.000.00n/aok
pcilutf64128^2 grid33.044 ms0.000.00n/aok
pcic_0f64128^2 grid2.915 ms0.000.00n/aok
pcildlf64128^2 grid2.936 ms0.000.00n/aok
pcssorf64128^2 grid0.119 ms0.000.00n/aok
pcblock_diagonal4f64128^2 grid, rows scaled3.389 ms0.000.00n/aok
smootherjacobif64512^2 grid, nnz 13086723.623 ms0.728.0979%ok
smoothergauss_seidelf64512^2 grid, nnz 13086723.123 ms0.849.3992%ok
smoothersor1.5f64512^2 grid, nnz 13086723.828 ms0.687.6675%ok
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)35.710 ms1.088.1479%ok
sparse_eiglanczosf6464x65 Laplacian, k=4, m=80142.049 ms0.000.00n/aok
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=80148.131 ms0.000.00n/aok
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=40144.134 ms0.000.00n/aok
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps42.513 ms0.000.00n/aok
mixeddotf32>f64n=8388608188.768 ms0.090.363%ok
mixedgemmf32>f64512x512x512113.542 ms2.360.0423%ok
mixedblock_scaled_doti8+f32n=8388608, block 322.367 ms7.097.5373%ok
integerdoti32n=8388608, |v| <= 156.620 ms2.5310.14n/aok
integerdoti16>i32n=8388608, |v| <= 153.112 ms5.3910.78n/aok
integerdoti8>i32n=8388608, |v| <= 155.015 ms3.353.35n/aok
integerdotu8i8>i32n=8388608, |v| <= 154.999 ms3.363.36n/aok
integergemmi32512^3, |v| <= 204757.681 ms4.650.05n/aok
integergemmi16>i32512^3, |v| <= 204745.933 ms5.840.05n/aok
integergemmi8>i32512^3, |v| <= 12723.593 ms11.380.07n/aok
integergemm_quadi8>i32512^3, |v| <= 12723.568 ms11.390.07n/aok
integergemm_quadu8i8>i32512^3, |v| <= 12770.878 ms3.790.02n/aok

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

xeon-e5-2420v2 — blas, 6 threads, class medium

Section titled “xeon-e5-2420v2 — blas, 6 threads, class medium”
CPUIntel(R) Xeon(R) CPU E5-2420 v2 @ 2.20GHz (x86_64, 12 logical cores)
OSLinux Ubuntu 26.04.1 LTS (7.0.0-43-generic)
compilerGCC 15.2.0
build -O3 -DNDEBUG, SIMD scalar
commitbfa1c2822688
FMA peak (ceiling)f64 58.9 GFLOP/s, f32 117.8 GFLOP/s
… hardware, SSE2f64 58.9, f32 117.8 GFLOP/s
… this build, scalarf64 0.5, f32 3.6 GFLOP/s
bandwidthread 11.3, update 10.6, triad 8.4 GB/s (ceiling 11.3)
last-level cache15 MiB
areaentriesmedian %roofbest %rooffailed checksmedian parallel efficiency
blas11054%107%082%
blas210104%108%020%
blas310126%158%075%
factor1218%38%024%
eigen89%24%017%
elementwise57%13%017%
sparse568%104%017%
direct70%1%017%
ordering4n/an/a017%
krylov13n/an/a017%
pc7n/an/a017%
smoother372%83%017%
mg174%74%017%
sparse_eig4n/an/a017%
mixed322%67%017%
integer9n/an/a090%
Every entry of this run
areaoptypeshapemedianGFLOP/sGB/s%roofcheckspeedup (efficiency)
blas1dotf64n=838860832.728 ms0.514.1036%ok5.5x (92%)
blas1dotf32n=83886085.537 ms3.0312.12107%ok4.9x (82%)
blas1axpyf64n=838860838.283 ms0.445.2646%ok5.6x (93%)
blas1axpyf32n=838860816.522 ms1.026.0954%ok5.5x (92%)
blas1scalef64n=838860812.140 ms0.6911.0698%ok1.0x (17%)
blas1two_normf64n=838860832.883 ms0.512.0418%ok5.6x (93%)
blas1one_normf64n=83886089.749 ms0.866.8861%ok1.0x (17%)
blas1infinity_normf64n=838860813.000 ms0.655.1646%ok1.0x (17%)
blas1sumf64n=83886089.777 ms0.866.8661%ok1.0x (17%)
blas1dot_k3f64n=8388608279.926 ms0.660.484%ok1.0x (17%)
blas2gemvf64n=409611.306 ms2.9711.88105%ok1.3x (21%)
blas2gemvf32n=40965.465 ms6.1412.29108%ok1.1x (19%)
blas2gemv_colmajorf64n=409610.927 ms3.0712.29108%ok1.2x (20%)
blas2gemv_colmajorf32n=40965.698 ms5.8911.78104%ok1.2x (19%)
blas2symvf64n=40965.786 ms5.8011.61102%ok1.2x (20%)
blas2trmvf64n=40965.728 ms2.9311.73104%ok1.3x (22%)
blas2trmvf32n=40962.963 ms5.6611.34100%ok1.2x (21%)
blas2trsvf64n=40967.405 ms2.279.0780%ok1.0x (17%)
blas2trsvf32n=40963.515 ms4.779.5584%ok1.0x (16%)
blas2gerf64n=409627.907 ms1.209.6285%ok1.0x (17%)
blas3gemmf64512x512x5122.875 ms93.372.19158%ok4.9x (81%)
blas3gemmf32512x512x5121.806 ms148.611.74126%ok4.5x (75%)
blas3gemmtf64512x512x5122.546 ms52.722.4789%ok3.4x (57%)
blas3symmf64512x512x5122.900 ms92.562.17157%ok4.9x (81%)
blas3syrkf64512x512x5122.519 ms53.271.6690%ok3.1x (52%)
blas3syr2kf64512x512x5123.711 ms72.341.70123%ok4.1x (68%)
blas3trmmf64512x512x5121.608 ms83.493.26142%ok4.6x (77%)
blas3trsmf64512x512x5121.767 ms75.982.97129%ok4.6x (77%)
blas3trsmf32512x512x5121.242 ms108.032.1192%ok4.3x (71%)
blas3kronf6448x48 (x) 48x4838.589 ms0.141.1010%ok1.0x (17%)
factorluf64512x5125.537 ms16.160.7627%ok1.7x (28%)
factorluf32512x5123.901 ms22.940.5419%ok1.6x (27%)
factorcholeskyf64512x5124.081 ms10.961.0319%ok1.4x (24%)
factorcholeskyf32512x5123.108 ms14.390.6712%ok1.3x (21%)
factorldltf64512x51225.485 ms1.760.163%ok1.0x (17%)
factorldlt_bkf64512x512133.535 ms0.340.031%ok1.1x (18%)
factorqrf64512x51212.973 ms13.790.3223%ok1.5x (26%)
factorlqf64512x51251.643 ms3.470.086%ok1.4x (24%)
factorhessenbergf64512x512595.367 ms0.750.011%ok1.1x (18%)
factorhouseholderf64512x5120.974 ms1.084.3038%*ok3.3x (55%)
factorinvf64512x512131.199 ms2.050.033%ok1.0x (16%)
factorlu_iterative_refinef32>f64512x5128.487 ms10.540.4918%ok1.5x (25%)
eigeneigenvalue_symmetricf64256x2565.791 ms3.860.187%ok1.1x (18%)
eigeneigen_symmetricf64256x256170.646 ms0.880.012%ok1.0x (17%)
eigeneigen_symmetric_jacobif64256x256999.865 ms0.150.000%ok1.0x (17%)
eigeneigenvaluef64256x25656.535 ms2.970.025%ok0.9x (16%)
eigenschurf64256x25675.059 ms5.590.019%ok1.0x (16%)
eigensvdf64256x25626.503 ms13.930.0424%ok1.1x (19%)
eigenexpmf64256x25627.353 ms9.200.0416%ok1.3x (22%)
eigensylvesterf64256x256196.965 ms5.110.019%ok1.0x (16%)
elementwiseexpf64n=8388608159.123 ms0.050.847%ok0.9x (16%)
elementwiselogf64n=8388608146.509 ms0.060.928%ok1.0x (16%)
elementwisesinf64n=8388608171.699 ms0.050.787%ok1.0x (17%)
elementwisesqrtf64n=838860889.941 ms0.091.4913%ok1.0x (17%)
elementwiseerff64n=83886082476.149 ms0.000.050%ok1.0x (17%)
sparsespmvf64512^2 grid, nnz 13086722.320 ms1.1311.74104%ok1.1x (18%)
sparsespmv_transf64512^2 grid, nnz 13086723.087 ms0.858.8278%ok1.0x (17%)
sparsecrs_densef64512^2 grid, nnz 1308672 x 1611.345 ms3.697.7668%ok1.3x (22%)
sparsespgemmf64256^2 grid, nnz 32665616.262 ms0.200.716%*ok1.0x (17%)
sparselower_solve_levelf64192^2 grid, nnz(L) 10035645.831 ms0.342.8525%ok1.0x (17%)
directsparse_choleskyf6496^2 grid, nnz 45696275.330 ms0.040.000%ok1.0x (17%)
directsparse_ldltf6496^2 grid, nnz 4569669.596 ms0.160.010%ok1.0x (17%)
directsupernodal_ldltf6496^2 grid, nnz 4569626.254 ms0.430.031%ok1.0x (17%)
directsparse_luf6496^2 grid, nnz 45696116.281 ms0.190.010%ok1.0x (17%)
directsupernodal_luf6496^2 grid, nnz 45696153.847 ms0.150.000%ok1.0x (17%)
directnative_kluf6496^2 grid, nnz 4569648.136 ms0.460.021%ok1.0x (17%)
directsparse_qrf6448^2 grid, nnz 11328231.070 ms0.010.000%ok1.0x (17%)
orderingamdf6496^2 grid, nnz 45696, scrambled8.903 ms0.000.00n/aok1.0x (17%)
orderingcolamdf6496^2 grid, nnz 45696, columns scrambled12.201 ms0.000.00n/aok1.0x (17%)
orderingrcmf6496^2 grid, nnz 45696, scrambled0.615 ms0.000.00n/aok1.0x (17%)
orderingblock_triangular_formf6432 planted blocks, n 819240.831 ms0.000.00n/aok1.0x (17%)
krylovcg+ic0f64128^2 grid, n 16384232.458 ms0.000.00n/aok1.0x (16%)
krylovminres+diagf64128^2 grid, n 16384397.041 ms0.000.00n/aok1.0x (17%)
krylovbicg+ilu0f64128^2 grid, n 16384239.140 ms0.000.00n/aok0.5x (9%)
krylovbicgstab+ilu0f64128^2 grid, n 16384102.078 ms0.000.00n/aok1.0x (17%)
krylovbicgstab_ell+ilu0f64128^2 grid, n 16384121.437 ms0.000.00n/aok1.0x (17%)
krylovcgs+ilu0f64128^2 grid, n 1638478.925 ms0.000.00n/aok1.0x (17%)
krylovgmres30+ilu0f64128^2 grid, n 16384514.809 ms0.000.00n/aok1.0x (17%)
krylovqmr+ilu0f64128^2 grid, n 16384174.287 ms0.000.00n/aok1.0x (17%)
krylovtfqmr+ilu0f64128^2 grid, n 1638491.537 ms0.000.00n/aok1.0x (17%)
krylovidr_s1+ilu0f64128^2 grid, n 16384136.001 ms0.000.00n/aok1.0x (17%)
krylovidr_s2+ilu0f64128^2 grid, n 16384138.021 ms0.000.00n/aok1.0x (17%)
krylovidr_s4+ilu0f64128^2 grid, n 16384173.402 ms0.000.00n/aok1.0x (16%)
krylovidr_s8+ilu0f64128^2 grid, n 16384248.610 ms0.000.00n/aok1.0x (17%)
pcdiagonalf64128^2 grid, rows scaled0.129 ms0.000.00n/aok1.0x (17%)
pcilu_0f64128^2 grid4.442 ms0.000.00n/aok1.0x (17%)
pcilutf64128^2 grid33.100 ms0.000.00n/aok1.0x (17%)
pcic_0f64128^2 grid2.869 ms0.000.00n/aok1.0x (17%)
pcildlf64128^2 grid2.712 ms0.000.00n/aok1.1x (18%)
pcssorf64128^2 grid0.077 ms0.000.00n/aok1.5x (26%)
pcblock_diagonal4f64128^2 grid, rows scaled3.518 ms0.000.00n/aok1.0x (16%)
smootherjacobif64512^2 grid, nnz 13086723.597 ms0.738.1572%ok1.0x (17%)
smoothergauss_seidelf64512^2 grid, nnz 13086723.119 ms0.849.4083%ok1.0x (17%)
smoothersor1.5f64512^2 grid, nnz 13086723.804 ms0.697.7168%ok1.0x (17%)
mgpoisson2d_vcyclef64255^2 grid, 6 levels, 10 V(1,1)34.676 ms1.118.3874%ok1.0x (17%)
sparse_eiglanczosf6464x65 Laplacian, k=4, m=80142.441 ms0.000.00n/aok1.0x (17%)
sparse_eigarnoldif6464x65 convection-diffusion, k=4, m=80147.324 ms0.000.00n/aok1.0x (17%)
sparse_eigshift_invertf6464x65 Laplacian, k=4 nearest 0, m=40142.578 ms0.000.00n/aok1.0x (17%)
sparse_eigpower_iterationf6464x65 Laplacian, 300 steps42.322 ms0.000.00n/aok1.0x (17%)
mixeddotf32>f64n=8388608189.054 ms0.090.353%ok1.0x (17%)
mixedgemmf32>f64512x512x51220.551 ms13.060.2022%ok5.5x (92%)
mixedblock_scaled_doti8+f32n=8388608, block 322.340 ms7.177.6267%ok1.0x (17%)
integerdoti32n=8388608, |v| <= 156.459 ms2.6010.39n/aok1.0x (17%)
integerdoti16>i32n=8388608, |v| <= 153.148 ms5.3310.66n/aok1.0x (16%)
integerdoti8>i32n=8388608, |v| <= 155.006 ms3.353.35n/aok1.0x (17%)
integerdotu8i8>i32n=8388608, |v| <= 154.991 ms3.363.36n/aok1.0x (17%)
integergemmi32512^3, |v| <= 204710.410 ms25.790.30n/aok5.5x (92%)
integergemmi16>i32512^3, |v| <= 20478.237 ms32.590.25n/aok5.6x (93%)
integergemmi8>i32512^3, |v| <= 1274.345 ms61.790.36n/aok5.4x (91%)
integergemm_quadi8>i32512^3, |v| <= 1274.343 ms61.810.36n/aok5.4x (90%)
integergemm_quadu8i8>i32512^3, |v| <= 12713.104 ms20.480.12n/aok5.4x (90%)

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