)]}' { "commit": "3d57983ce3968415c20ec5b4d66c8b3a1b10d82d", "tree": "fb912ea55350cfc223216ca7882a3c47d09ee191", "parents": [ "a7ffc3343e66f51e3196924f34d658dc025b6de2" ], "author": { "name": "Jeremy Dorfman", "email": "jdorfman@google.com", "time": "Tue May 12 10:41:03 2026 -0400" }, "committer": { "name": "Jeremy Dorfman", "email": "jdorfman@google.com", "time": "Tue May 12 09:41:46 2026 -0700" }, "message": "[convolve] Optimize av1_convolve_2d_sr_avx2\n\nPreviously, horizontal and vertical convolutions were interleaved\nstrip-by-strip. The horizontal convolution\u0027s inner loop traversed\nvertical row indices, leading to non-local accesses.\n\nThis change optimizes memory spatial and temporal locality by switching\nthe loop order:\n- Swaps column-strip and row processing, making the column-strip (j) the\n inner loop and the row (i) the outer loop during horizontal filtering\n- Enables contiguous, cache-line-aligned sequential memory reads across\n rows, which is more efficient on modern CPUs\n- Resizes the intermediate stack buffer from a reusable single-strip\n buffer (~2.2 KB) to a full block-sized buffer im_block_buf (~35.8 KB)\n to support decoupled horizontal filtering across all strips\n\nNo functional changes. This is very slightly slower for 8x32 and ranges\nbetween neutral to much better for most other block sizes:\n\nBlockWidth/BlockHeight/HorizontalFilter/VerticalFilter\n Baseline CPU New CPU Improvement\n4/4/0/0 14.42n ± 1% 14.57n ± 1% +1.06% (p\u003d0.001 n\u003d10)\n4/4/0/1 14.50n ± 1% 14.59n ± 0% ~ (p\u003d0.052 n\u003d10)\n4/4/0/2 14.40n ± 0% 14.58n ± 1% +1.24% (p\u003d0.000 n\u003d10)\n4/4/0/3 14.41n ± 1% 14.57n ± 1% +1.13% (p\u003d0.001 n\u003d10)\n4/4/1/0 14.47n ± 2% 14.60n ± 0% ~ (p\u003d0.063 n\u003d10)\n4/4/1/1 14.39n ± 1% 14.58n ± 0% +1.27% (p\u003d0.015 n\u003d10)\n4/4/1/2 14.50n ± 1% 14.57n ± 0% +0.54% (p\u003d0.011 n\u003d10)\n4/4/1/3 14.44n ± 1% 14.59n ± 0% +1.04% (p\u003d0.023 n\u003d10)\n4/4/2/0 14.40n ± 1% 14.58n ± 0% +1.21% (p\u003d0.015 n\u003d10)\n4/4/2/1 14.46n ± 0% 14.56n ± 0% +0.65% (p\u003d0.004 n\u003d10)\n4/4/2/2 14.45n ± 2% 14.53n ± 0% ~ (p\u003d0.085 n\u003d10)\n4/4/2/3 14.40n ± 1% 14.58n ± 0% +1.21% (p\u003d0.001 n\u003d10)\n4/4/3/0 14.46n ± 1% 14.57n ± 0% ~ (p\u003d0.165 n\u003d10)\n4/4/3/1 14.49n ± 3% 14.58n ± 1% ~ (p\u003d0.353 n\u003d10)\n4/4/3/2 14.39n ± 1% 14.58n ± 1% +1.36% (p\u003d0.001 n\u003d10)\n4/4/3/3 14.41n ± 1% 14.57n ± 1% +1.10% (p\u003d0.009 n\u003d10)\n4/4/4/4 42.91n ± 1% 41.77n ± 1% -2.64% (p\u003d0.000 n\u003d10)\n4/8/0/0 17.15n ± 1% 17.08n ± 4% ~ (p\u003d0.247 n\u003d10)\n4/8/0/1 17.13n ± 1% 17.18n ± 2% ~ (p\u003d0.315 n\u003d10)\n4/8/0/2 17.20n ± 1% 17.02n ± 3% ~ (p\u003d0.240 n\u003d10)\n4/8/0/3 17.27n ± 1% 17.03n ± 3% ~ (p\u003d0.143 n\u003d10)\n4/8/1/0 17.27n ± 1% 17.10n ± 3% ~ (p\u003d0.143 n\u003d10)\n4/8/1/1 17.13n ± 0% 17.11n ± 3% ~ (p\u003d0.796 n\u003d10)\n4/8/1/2 17.13n ± 2% 17.13n ± 2% ~ (p\u003d0.529 n\u003d10)\n4/8/1/3 17.20n ± 1% 17.05n ± 3% ~ (p\u003d0.160 n\u003d10)\n4/8/2/0 17.22n ± 1% 17.12n ± 3% ~ (p\u003d0.436 n\u003d10)\n4/8/2/1 17.18n ± 1% 17.17n ± 2% ~ (p\u003d0.987 n\u003d10)\n4/8/2/2 17.36n ± 1% 17.16n ± 4% ~ (p\u003d0.424 n\u003d10)\n4/8/2/3 17.21n ± 2% 17.15n ± 2% ~ (p\u003d0.353 n\u003d10)\n4/8/3/0 17.18n ± 1% 17.09n ± 3% ~ (p\u003d0.404 n\u003d10)\n4/8/3/1 17.23n ± 2% 17.25n ± 2% ~ (p\u003d0.724 n\u003d10)\n4/8/3/2 17.15n ± 1% 17.18n ± 2% ~ (p\u003d0.631 n\u003d10)\n4/8/3/3 17.21n ± 2% 17.08n ± 3% ~ (p\u003d0.183 n\u003d10)\n4/8/4/4 56.68n ± 1% 55.07n ± 1% -2.83% (p\u003d0.000 n\u003d10)\n8/4/0/0 20.72n ± 2% 13.98n ± 1% -32.53% (p\u003d0.000 n\u003d10)\n8/4/0/1 20.73n ± 1% 13.98n ± 1% -32.57% (p\u003d0.000 n\u003d10)\n8/4/0/2 20.83n ± 1% 13.95n ± 1% -33.05% (p\u003d0.000 n\u003d10)\n8/4/0/3 20.76n ± 3% 13.94n ± 0% -32.82% (p\u003d0.000 n\u003d10)\n8/4/1/0 20.64n ± 4% 13.95n ± 1% -32.40% (p\u003d0.000 n\u003d10)\n8/4/1/1 20.96n ± 2% 13.99n ± 0% -33.28% (p\u003d0.000 n\u003d10)\n8/4/1/2 20.93n ± 2% 13.96n ± 0% -33.28% (p\u003d0.000 n\u003d10)\n8/4/1/3 20.58n ± 2% 13.96n ± 0% -32.17% (p\u003d0.000 n\u003d10)\n8/4/2/0 20.72n ± 2% 13.94n ± 1% -32.72% (p\u003d0.000 n\u003d10)\n8/4/2/1 20.86n ± 3% 13.96n ± 1% -33.07% (p\u003d0.000 n\u003d10)\n8/4/2/2 20.70n ± 1% 13.95n ± 1% -32.58% (p\u003d0.000 n\u003d10)\n8/4/2/3 20.59n ± 2% 13.96n ± 1% -32.19% (p\u003d0.000 n\u003d10)\n8/4/3/0 20.58n ± 2% 13.95n ± 1% -32.18% (p\u003d0.000 n\u003d10)\n8/4/3/1 20.62n ± 1% 13.95n ± 0% -32.33% (p\u003d0.000 n\u003d10)\n8/4/3/2 20.76n ± 1% 14.00n ± 1% -32.57% (p\u003d0.000 n\u003d10)\n8/4/3/3 20.83n ± 4% 13.96n ± 1% -32.97% (p\u003d0.000 n\u003d10)\n8/4/4/4 58.96n ± 1% 59.39n ± 2% +0.73% (p\u003d0.015 n\u003d10)\n8/8/0/0 21.55n ± 1% 18.17n ± 0% -15.68% (p\u003d0.000 n\u003d10)\n8/8/0/1 21.51n ± 0% 18.16n ± 0% -15.58% (p\u003d0.000 n\u003d10)\n8/8/0/2 21.56n ± 1% 18.16n ± 0% -15.79% (p\u003d0.000 n\u003d10)\n8/8/0/3 21.58n ± 1% 18.17n ± 0% -15.81% (p\u003d0.000 n\u003d10)\n8/8/1/0 21.56n ± 1% 18.16n ± 0% -15.79% (p\u003d0.000 n\u003d10)\n8/8/1/1 21.62n ± 1% 18.18n ± 1% -15.88% (p\u003d0.000 n\u003d10)\n8/8/1/2 21.59n ± 1% 18.16n ± 0% -15.89% (p\u003d0.000 n\u003d10)\n8/8/1/3 21.50n ± 1% 18.16n ± 0% -15.54% (p\u003d0.000 n\u003d10)\n8/8/2/0 21.57n ± 0% 18.14n ± 0% -15.89% (p\u003d0.000 n\u003d10)\n8/8/2/1 21.52n ± 1% 18.15n ± 1% -15.64% (p\u003d0.000 n\u003d10)\n8/8/2/2 21.58n ± 1% 18.16n ± 0% -15.84% (p\u003d0.000 n\u003d10)\n8/8/2/3 21.57n ± 0% 18.16n ± 0% -15.82% (p\u003d0.000 n\u003d10)\n8/8/3/0 21.61n ± 1% 18.17n ± 0% -15.90% (p\u003d0.000 n\u003d10)\n8/8/3/1 21.52n ± 0% 18.16n ± 0% -15.63% (p\u003d0.000 n\u003d10)\n8/8/3/2 21.60n ± 1% 18.16n ± 0% -15.95% (p\u003d0.000 n\u003d10)\n8/8/3/3 21.52n ± 1% 18.16n ± 0% -15.63% (p\u003d0.000 n\u003d10)\n8/8/4/4 76.97n ± 0% 78.07n ± 1% +1.42% (p\u003d0.000 n\u003d10)\n8/16/0/0 26.12n ± 0% 25.93n ± 1% -0.72% (p\u003d0.002 n\u003d10)\n8/16/0/1 26.19n ± 1% 25.96n ± 1% -0.89% (p\u003d0.020 n\u003d10)\n8/16/0/2 26.11n ± 0% 25.97n ± 1% -0.53% (p\u003d0.007 n\u003d10)\n8/16/0/3 26.13n ± 0% 25.98n ± 1% -0.56% (p\u003d0.023 n\u003d10)\n8/16/1/0 26.16n ± 1% 26.00n ± 0% -0.63% (p\u003d0.004 n\u003d10)\n8/16/1/1 26.09n ± 0% 25.94n ± 1% -0.57% (p\u003d0.001 n\u003d10)\n8/16/1/2 26.10n ± 0% 25.97n ± 1% -0.50% (p\u003d0.009 n\u003d10)\n8/16/1/3 26.08n ± 1% 25.99n ± 0% -0.34% (p\u003d0.045 n\u003d10)\n8/16/2/0 26.16n ± 1% 25.98n ± 1% -0.68% (p\u003d0.000 n\u003d10)\n8/16/2/1 26.07n ± 1% 25.99n ± 0% -0.33% (p\u003d0.011 n\u003d10)\n8/16/2/2 26.10n ± 1% 25.95n ± 0% -0.59% (p\u003d0.001 n\u003d10)\n8/16/2/3 26.12n ± 0% 25.97n ± 1% -0.57% (p\u003d0.043 n\u003d10)\n8/16/3/0 26.19n ± 1% 25.94n ± 1% -0.94% (p\u003d0.002 n\u003d10)\n8/16/3/1 26.12n ± 1% 25.95n ± 0% -0.65% (p\u003d0.000 n\u003d10)\n8/16/3/2 26.13n ± 1% 25.97n ± 1% -0.59% (p\u003d0.004 n\u003d10)\n8/16/3/3 26.14n ± 0% 25.96n ± 1% -0.66% (p\u003d0.001 n\u003d10)\n8/16/4/4 112.0n ± 0% 113.5n ± 1% +1.39% (p\u003d0.000 n\u003d10)\n16/8/0/0 37.37n ± 1% 20.04n ± 2% -46.39% (p\u003d0.000 n\u003d10)\n16/8/0/1 37.19n ± 1% 20.03n ± 0% -46.16% (p\u003d0.000 n\u003d10)\n16/8/0/2 37.41n ± 1% 20.06n ± 0% -46.38% (p\u003d0.000 n\u003d10)\n16/8/0/3 37.37n ± 1% 20.04n ± 1% -46.38% (p\u003d0.000 n\u003d10)\n16/8/1/0 37.22n ± 1% 20.02n ± 0% -46.22% (p\u003d0.000 n\u003d10)\n16/8/1/1 37.40n ± 2% 20.01n ± 1% -46.49% (p\u003d0.000 n\u003d10)\n16/8/1/2 37.53n ± 1% 20.00n ± 0% -46.71% (p\u003d0.000 n\u003d10)\n16/8/1/3 37.33n ± 1% 20.02n ± 1% -46.37% (p\u003d0.000 n\u003d10)\n16/8/2/0 37.14n ± 1% 20.02n ± 1% -46.10% (p\u003d0.000 n\u003d10)\n16/8/2/1 37.54n ± 1% 20.04n ± 1% -46.62% (p\u003d0.000 n\u003d10)\n16/8/2/2 37.19n ± 2% 20.03n ± 1% -46.16% (p\u003d0.000 n\u003d10)\n16/8/2/3 37.75n ± 2% 20.05n ± 1% -46.89% (p\u003d0.000 n\u003d10)\n16/8/3/0 37.47n ± 1% 20.04n ± 1% -46.51% (p\u003d0.000 n\u003d10)\n16/8/3/1 37.40n ± 1% 20.03n ± 0% -46.45% (p\u003d0.000 n\u003d10)\n16/8/3/2 37.42n ± 1% 19.96n ± 0% -46.67% (p\u003d0.000 n\u003d10)\n16/8/3/3 37.42n ± 2% 20.03n ± 1% -46.47% (p\u003d0.000 n\u003d10)\n16/8/4/4 142.9n ± 1% 141.9n ± 1% -0.70% (p\u003d0.035 n\u003d10)\n16/16/0/0 45.08n ± 0% 30.92n ± 1% -31.41% (p\u003d0.000 n\u003d10)\n16/16/0/1 45.07n ± 1% 30.99n ± 1% -31.25% (p\u003d0.000 n\u003d10)\n16/16/0/2 45.16n ± 1% 30.80n ± 1% -31.79% (p\u003d0.000 n\u003d10)\n16/16/0/3 45.20n ± 0% 30.80n ± 2% -31.86% (p\u003d0.000 n\u003d10)\n16/16/1/0 44.99n ± 0% 30.75n ± 2% -31.65% (p\u003d0.000 n\u003d10)\n16/16/1/1 45.27n ± 1% 30.86n ± 1% -31.82% (p\u003d0.000 n\u003d10)\n16/16/1/2 45.14n ± 0% 30.89n ± 1% -31.56% (p\u003d0.000 n\u003d10)\n16/16/1/3 45.01n ± 1% 30.91n ± 1% -31.33% (p\u003d0.000 n\u003d10)\n16/16/2/0 45.07n ± 0% 30.83n ± 1% -31.58% (p\u003d0.000 n\u003d10)\n16/16/2/1 45.05n ± 2% 30.83n ± 2% -31.57% (p\u003d0.000 n\u003d10)\n16/16/2/2 45.08n ± 0% 30.78n ± 1% -31.71% (p\u003d0.000 n\u003d10)\n16/16/2/3 45.13n ± 1% 30.83n ± 1% -31.69% (p\u003d0.000 n\u003d10)\n16/16/3/0 45.06n ± 1% 30.92n ± 0% -31.37% (p\u003d0.000 n\u003d10)\n16/16/3/1 45.14n ± 1% 30.88n ± 2% -31.58% (p\u003d0.000 n\u003d10)\n16/16/3/2 45.13n ± 1% 30.80n ± 1% -31.75% (p\u003d0.000 n\u003d10)\n16/16/3/3 45.25n ± 1% 30.96n ± 1% -31.57% (p\u003d0.000 n\u003d10)\n16/16/4/4 212.7n ± 0% 211.9n ± 0% -0.37% (p\u003d0.035 n\u003d10)\n16/32/0/0 66.68n ± 0% 54.86n ± 6% -17.74% (p\u003d0.000 n\u003d10)\n16/32/0/1 66.65n ± 0% 55.89n ± 7% -16.13% (p\u003d0.000 n\u003d10)\n16/32/0/2 66.68n ± 0% 58.62n ± 10% -12.10% (p\u003d0.000 n\u003d10)\n16/32/0/3 66.69n ± 0% 53.86n ± 11% -19.23% (p\u003d0.000 n\u003d10)\n16/32/1/0 66.69n ± 1% 56.56n ± 6% -15.19% (p\u003d0.000 n\u003d10)\n16/32/1/1 66.70n ± 0% 53.82n ± 8% -19.30% (p\u003d0.000 n\u003d10)\n16/32/1/2 66.71n ± 1% 53.06n ± 12% -20.46% (p\u003d0.000 n\u003d10)\n16/32/1/3 66.65n ± 0% 57.09n ± 8% -14.34% (p\u003d0.000 n\u003d10)\n16/32/2/0 66.69n ± 1% 55.87n ± 6% -16.22% (p\u003d0.000 n\u003d10)\n16/32/2/1 66.75n ± 0% 53.30n ± 11% -20.15% (p\u003d0.000 n\u003d10)\n16/32/2/2 66.76n ± 1% 53.76n ± 11% -19.48% (p\u003d0.000 n\u003d10)\n16/32/2/3 66.76n ± 0% 57.96n ± 9% -13.19% (p\u003d0.000 n\u003d10)\n16/32/3/0 66.68n ± 0% 53.73n ± 11% -19.42% (p\u003d0.000 n\u003d10)\n16/32/3/1 66.72n ± 0% 54.78n ± 9% -17.91% (p\u003d0.000 n\u003d10)\n16/32/3/2 66.68n ± 0% 57.23n ± 7% -14.17% (p\u003d0.000 n\u003d10)\n16/32/3/3 66.77n ± 0% 53.68n ± 10% -19.61% (p\u003d0.000 n\u003d10)\n16/32/4/4 354.3n ± 0% 356.9n ± 1% +0.74% (p\u003d0.000 n\u003d10)\n32/16/0/0 83.68n ± 1% 40.10n ± 1% -52.08% (p\u003d0.000 n\u003d10)\n32/16/0/1 83.56n ± 1% 39.93n ± 1% -52.22% (p\u003d0.000 n\u003d10)\n32/16/0/2 83.68n ± 1% 40.04n ± 1% -52.16% (p\u003d0.000 n\u003d10)\n32/16/0/3 84.08n ± 1% 40.08n ± 0% -52.33% (p\u003d0.000 n\u003d10)\n32/16/1/0 83.58n ± 1% 40.09n ± 1% -52.04% (p\u003d0.000 n\u003d10)\n32/16/1/1 83.84n ± 1% 40.08n ± 1% -52.20% (p\u003d0.000 n\u003d10)\n32/16/1/2 83.56n ± 1% 40.02n ± 1% -52.11% (p\u003d0.000 n\u003d10)\n32/16/1/3 83.55n ± 1% 39.98n ± 1% -52.15% (p\u003d0.000 n\u003d10)\n32/16/2/0 83.55n ± 1% 40.08n ± 1% -52.02% (p\u003d0.000 n\u003d10)\n32/16/2/1 83.84n ± 1% 40.00n ± 0% -52.29% (p\u003d0.000 n\u003d10)\n32/16/2/2 83.75n ± 1% 39.80n ± 4% -52.48% (p\u003d0.000 n\u003d10)\n32/16/2/3 83.71n ± 1% 39.98n ± 1% -52.24% (p\u003d0.000 n\u003d10)\n32/16/3/0 83.63n ± 1% 40.08n ± 1% -52.08% (p\u003d0.000 n\u003d10)\n32/16/3/1 83.83n ± 1% 40.10n ± 1% -52.17% (p\u003d0.000 n\u003d10)\n32/16/3/2 83.64n ± 1% 39.89n ± 1% -52.31% (p\u003d0.000 n\u003d10)\n32/16/3/3 83.77n ± 0% 40.07n ± 1% -52.17% (p\u003d0.000 n\u003d10)\n32/16/4/4 419.2n ± 1% 415.3n ± 1% -0.94% (p\u003d0.000 n\u003d10)\n32/32/0/0 123.33n ± 1% 76.38n ± 1% -38.07% (p\u003d0.000 n\u003d10)\n32/32/0/1 123.21n ± 0% 76.17n ± 2% -38.18% (p\u003d0.000 n\u003d10)\n32/32/0/2 123.32n ± 1% 76.63n ± 2% -37.86% (p\u003d0.000 n\u003d10)\n32/32/0/3 123.31n ± 0% 76.73n ± 1% -37.77% (p\u003d0.000 n\u003d10)\n32/32/1/0 123.14n ± 1% 76.39n ± 2% -37.96% (p\u003d0.000 n\u003d10)\n32/32/1/1 123.21n ± 0% 76.71n ± 1% -37.74% (p\u003d0.000 n\u003d10)\n32/32/1/2 123.00n ± 1% 76.97n ± 1% -37.42% (p\u003d0.000 n\u003d10)\n32/32/1/3 123.26n ± 0% 76.79n ± 1% -37.70% (p\u003d0.000 n\u003d10)\n32/32/2/0 123.32n ± 0% 76.83n ± 3% -37.70% (p\u003d0.000 n\u003d10)\n32/32/2/1 123.15n ± 1% 76.64n ± 3% -37.77% (p\u003d0.000 n\u003d10)\n32/32/2/2 122.94n ± 1% 76.95n ± 3% -37.41% (p\u003d0.000 n\u003d10)\n32/32/2/3 123.11n ± 0% 76.72n ± 1% -37.68% (p\u003d0.000 n\u003d10)\n32/32/3/0 123.27n ± 0% 76.78n ± 1% -37.72% (p\u003d0.000 n\u003d10)\n32/32/3/1 123.11n ± 1% 76.50n ± 1% -37.86% (p\u003d0.000 n\u003d10)\n32/32/3/2 123.29n ± 1% 76.99n ± 1% -37.55% (p\u003d0.000 n\u003d10)\n32/32/3/3 123.13n ± 1% 76.99n ± 1% -37.47% (p\u003d0.000 n\u003d10)\n32/32/4/4 701.4n ± 0% 694.4n ± 0% -1.00% (p\u003d0.000 n\u003d10)\n32/64/0/0 231.5n ± 1% 140.4n ± 1% -39.35% (p\u003d0.000 n\u003d10)\n32/64/0/1 231.5n ± 1% 140.6n ± 1% -39.28% (p\u003d0.000 n\u003d10)\n32/64/0/2 231.8n ± 1% 140.0n ± 1% -39.58% (p\u003d0.000 n\u003d10)\n32/64/0/3 231.2n ± 1% 139.3n ± 1% -39.76% (p\u003d0.000 n\u003d10)\n32/64/1/0 231.8n ± 0% 140.0n ± 2% -39.61% (p\u003d0.000 n\u003d10)\n32/64/1/1 232.3n ± 1% 140.4n ± 2% -39.57% (p\u003d0.000 n\u003d10)\n32/64/1/2 231.6n ± 0% 139.8n ± 1% -39.64% (p\u003d0.000 n\u003d10)\n32/64/1/3 231.6n ± 1% 140.5n ± 2% -39.33% (p\u003d0.000 n\u003d10)\n32/64/2/0 231.8n ± 1% 140.3n ± 2% -39.50% (p\u003d0.000 n\u003d10)\n32/64/2/1 231.4n ± 1% 138.7n ± 1% -40.06% (p\u003d0.000 n\u003d10)\n32/64/2/2 231.8n ± 1% 140.4n ± 1% -39.44% (p\u003d0.000 n\u003d10)\n32/64/2/3 231.9n ± 1% 138.7n ± 2% -40.17% (p\u003d0.000 n\u003d10)\n32/64/3/0 231.5n ± 0% 140.6n ± 1% -39.28% (p\u003d0.000 n\u003d10)\n32/64/3/1 231.9n ± 1% 139.7n ± 1% -39.75% (p\u003d0.000 n\u003d10)\n32/64/3/2 232.0n ± 0% 139.9n ± 2% -39.70% (p\u003d0.000 n\u003d10)\n32/64/3/3 231.6n ± 0% 139.9n ± 2% -39.62% (p\u003d0.000 n\u003d10)\n32/64/4/4 1.265µ ± 0% 1.261µ ± 0% ~ (p\u003d0.089 n\u003d10)\n64/32/0/0 242.16n ± 1% 61.13n ± 10% -74.76% (p\u003d0.000 n\u003d10)\n64/32/0/1 242.21n ± 1% 55.68n ± 10% -77.01% (p\u003d0.000 n\u003d10)\n64/32/0/2 241.93n ± 1% 58.46n ± 6% -75.84% (p\u003d0.000 n\u003d10)\n64/32/0/3 242.18n ± 1% 61.14n ± 10% -74.75% (p\u003d0.000 n\u003d10)\n64/32/1/0 241.88n ± 0% 61.12n ± 10% -74.73% (p\u003d0.000 n\u003d10)\n64/32/1/1 242.44n ± 1% 55.30n ± 11% -77.19% (p\u003d0.000 n\u003d10)\n64/32/1/2 241.08n ± 1% 58.99n ± 6% -75.53% (p\u003d0.000 n\u003d10)\n64/32/1/3 241.39n ± 0% 61.12n ± 10% -74.68% (p\u003d0.000 n\u003d10)\n64/32/2/0 241.47n ± 1% 55.66n ± 10% -76.95% (p\u003d0.000 n\u003d10)\n64/32/2/1 241.82n ± 1% 55.69n ± 10% -76.97% (p\u003d0.000 n\u003d10)\n64/32/2/2 241.91n ± 1% 61.12n ± 10% -74.74% (p\u003d0.000 n\u003d10)\n64/32/2/3 241.16n ± 1% 61.13n ± 9% -74.65% (p\u003d0.000 n\u003d10)\n64/32/3/0 241.75n ± 1% 61.12n ± 10% -74.72% (p\u003d0.000 n\u003d10)\n64/32/3/1 242.04n ± 1% 55.24n ± 11% -77.18% (p\u003d0.000 n\u003d10)\n64/32/3/2 241.78n ± 1% 61.13n ± 10% -74.72% (p\u003d0.000 n\u003d10)\n64/32/3/3 241.61n ± 0% 61.13n ± 9% -74.70% (p\u003d0.000 n\u003d10)\n64/32/4/4 1.386µ ± 1% 1.373µ ± 1% -0.91% (p\u003d0.004 n\u003d10)\n64/64/0/0 444.8n ± 0% 106.1n ± 0% -76.15% (p\u003d0.000 n\u003d10)\n64/64/0/1 444.0n ± 1% 106.1n ± 0% -76.10% (p\u003d0.000 n\u003d10)\n64/64/0/2 444.8n ± 2% 106.1n ± 0% -76.16% (p\u003d0.000 n\u003d10)\n64/64/0/3 443.4n ± 0% 106.1n ± 0% -76.07% (p\u003d0.000 n\u003d10)\n64/64/1/0 443.2n ± 0% 106.2n ± 0% -76.04% (p\u003d0.000 n\u003d10)\n64/64/1/1 443.6n ± 0% 106.1n ± 0% -76.09% (p\u003d0.000 n\u003d10)\n64/64/1/2 444.3n ± 1% 106.1n ± 0% -76.13% (p\u003d0.000 n\u003d10)\n64/64/1/3 444.4n ± 0% 106.1n ± 1% -76.13% (p\u003d0.000 n\u003d10)\n64/64/2/0 443.6n ± 0% 106.1n ± 0% -76.07% (p\u003d0.000 n\u003d10)\n64/64/2/1 444.0n ± 0% 106.1n ± 0% -76.11% (p\u003d0.000 n\u003d10)\n64/64/2/2 444.1n ± 0% 106.1n ± 0% -76.12% (p\u003d0.000 n\u003d10)\n64/64/2/3 443.7n ± 1% 106.1n ± 0% -76.08% (p\u003d0.000 n\u003d10)\n64/64/3/0 443.8n ± 0% 106.1n ± 0% -76.08% (p\u003d0.000 n\u003d10)\n64/64/3/1 444.3n ± 0% 106.1n ± 0% -76.12% (p\u003d0.000 n\u003d10)\n64/64/3/2 444.0n ± 1% 106.1n ± 0% -76.11% (p\u003d0.000 n\u003d10)\n64/64/3/3 444.3n ± 1% 106.1n ± 0% -76.11% (p\u003d0.000 n\u003d10)\n64/64/4/4 2.506µ ± 0% 2.495µ ± 1% -0.43% (p\u003d0.015 n\u003d10)\n64/128/0/0 880.2n ± 1% 197.5n ± 1% -77.56% (p\u003d0.000 n\u003d10)\n64/128/0/1 882.8n ± 1% 197.5n ± 1% -77.63% (p\u003d0.000 n\u003d10)\n64/128/0/2 881.9n ± 1% 197.4n ± 1% -77.62% (p\u003d0.000 n\u003d10)\n64/128/0/3 882.6n ± 1% 197.5n ± 6% -77.62% (p\u003d0.000 n\u003d10)\n64/128/1/0 881.2n ± 1% 197.4n ± 1% -77.60% (p\u003d0.000 n\u003d10)\n64/128/1/1 880.2n ± 1% 197.3n ± 1% -77.58% (p\u003d0.000 n\u003d10)\n64/128/1/2 879.7n ± 1% 197.6n ± 1% -77.54% (p\u003d0.000 n\u003d10)\n64/128/1/3 880.1n ± 1% 197.4n ± 1% -77.57% (p\u003d0.000 n\u003d10)\n64/128/2/0 882.2n ± 1% 197.5n ± 1% -77.61% (p\u003d0.000 n\u003d10)\n64/128/2/1 879.3n ± 1% 197.5n ± 1% -77.54% (p\u003d0.000 n\u003d10)\n64/128/2/2 880.6n ± 1% 197.3n ± 2% -77.59% (p\u003d0.000 n\u003d10)\n64/128/2/3 878.8n ± 1% 197.5n ± 1% -77.53% (p\u003d0.000 n\u003d10)\n64/128/3/0 881.5n ± 1% 197.3n ± 2% -77.61% (p\u003d0.000 n\u003d10)\n64/128/3/1 880.1n ± 1% 197.5n ± 1% -77.56% (p\u003d0.000 n\u003d10)\n64/128/3/2 881.2n ± 1% 197.4n ± 3% -77.60% (p\u003d0.000 n\u003d10)\n64/128/3/3 880.6n ± 1% 197.4n ± 1% -77.59% (p\u003d0.000 n\u003d10)\n64/128/4/4 4.833µ ± 0% 4.823µ ± 0% ~ (p\u003d0.247 n\u003d10)\n128/64/0/0 869.4n ± 0% 160.4n ± 1% -81.55% (p\u003d0.000 n\u003d10)\n128/64/0/1 868.3n ± 1% 160.4n ± 0% -81.53% (p\u003d0.000 n\u003d10)\n128/64/0/2 867.7n ± 0% 160.4n ± 0% -81.52% (p\u003d0.000 n\u003d10)\n128/64/0/3 868.6n ± 1% 160.4n ± 0% -81.54% (p\u003d0.000 n\u003d10)\n128/64/1/0 868.6n ± 0% 160.4n ± 1% -81.53% (p\u003d0.000 n\u003d10)\n128/64/1/1 867.8n ± 1% 160.5n ± 0% -81.51% (p\u003d0.000 n\u003d10)\n128/64/1/2 867.6n ± 1% 160.4n ± 0% -81.51% (p\u003d0.000 n\u003d10)\n128/64/1/3 867.4n ± 0% 160.4n ± 0% -81.51% (p\u003d0.000 n\u003d10)\n128/64/2/0 867.0n ± 1% 160.5n ± 0% -81.49% (p\u003d0.000 n\u003d10)\n128/64/2/1 867.1n ± 1% 160.4n ± 0% -81.50% (p\u003d0.000 n\u003d10)\n128/64/2/2 867.9n ± 0% 160.5n ± 0% -81.51% (p\u003d0.000 n\u003d10)\n128/64/2/3 867.9n ± 0% 160.4n ± 0% -81.52% (p\u003d0.000 n\u003d10)\n128/64/3/0 868.1n ± 0% 160.3n ± 0% -81.53% (p\u003d0.000 n\u003d10)\n128/64/3/1 867.6n ± 0% 160.4n ± 0% -81.52% (p\u003d0.000 n\u003d10)\n128/64/3/2 867.3n ± 0% 160.4n ± 0% -81.51% (p\u003d0.000 n\u003d10)\n128/64/3/3 869.0n ± 0% 160.4n ± 0% -81.54% (p\u003d0.000 n\u003d10)\n128/64/4/4 5.000µ ± 0% 5.015µ ± 1% +0.30% (p\u003d0.001 n\u003d10)\n128/128/0/0 1880.9n ± 1% 307.6n ± 1% -83.65% (p\u003d0.000 n\u003d10)\n128/128/0/1 1889.2n ± 1% 307.7n ± 0% -83.71% (p\u003d0.000 n\u003d10)\n128/128/0/2 1878.7n ± 1% 307.7n ± 0% -83.62% (p\u003d0.000 n\u003d10)\n128/128/0/3 1882.9n ± 1% 308.1n ± 0% -83.64% (p\u003d0.000 n\u003d10)\n128/128/1/0 1882.6n ± 0% 307.7n ± 0% -83.65% (p\u003d0.000 n\u003d10)\n128/128/1/1 1883.8n ± 1% 307.7n ± 0% -83.67% (p\u003d0.000 n\u003d10)\n128/128/1/2 1879.8n ± 1% 307.8n ± 0% -83.63% (p\u003d0.000 n\u003d10)\n128/128/1/3 1883.5n ± 1% 308.0n ± 0% -83.65% (p\u003d0.000 n\u003d10)\n128/128/2/0 1883.1n ± 1% 307.8n ± 0% -83.65% (p\u003d0.000 n\u003d10)\n128/128/2/1 1880.2n ± 0% 307.8n ± 0% -83.63% (p\u003d0.000 n\u003d10)\n128/128/2/2 1882.4n ± 1% 307.7n ± 0% -83.65% (p\u003d0.000 n\u003d10)\n128/128/2/3 1883.3n ± 1% 307.8n ± 0% -83.65% (p\u003d0.000 n\u003d10)\n128/128/3/0 1882.5n ± 1% 307.7n ± 0% -83.65% (p\u003d0.000 n\u003d10)\n128/128/3/1 1882.0n ± 1% 307.9n ± 0% -83.64% (p\u003d0.000 n\u003d10)\n128/128/3/2 1879.9n ± 1% 307.8n ± 1% -83.63% (p\u003d0.000 n\u003d10)\n128/128/3/3 1883.9n ± 1% 307.8n ± 0% -83.66% (p\u003d0.000 n\u003d10)\n128/128/4/4 9.923µ ± 0% 9.616µ ± 1% -3.09% (p\u003d0.000 n\u003d10)\n4/16/0/0 23.58n ± 1% 23.05n ± 1% -2.26% (p\u003d0.000 n\u003d10)\n4/16/0/1 23.59n ± 0% 22.97n ± 0% -2.63% (p\u003d0.000 n\u003d10)\n4/16/0/2 23.59n ± 0% 23.08n ± 1% -2.17% (p\u003d0.000 n\u003d10)\n4/16/0/3 23.64n ± 0% 22.97n ± 2% -2.85% (p\u003d0.000 n\u003d10)\n4/16/1/0 23.58n ± 0% 22.97n ± 0% -2.61% (p\u003d0.000 n\u003d10)\n4/16/1/1 23.60n ± 0% 23.00n ± 0% -2.53% (p\u003d0.000 n\u003d10)\n4/16/1/2 23.61n ± 0% 23.06n ± 0% -2.35% (p\u003d0.000 n\u003d10)\n4/16/1/3 23.61n ± 0% 23.00n ± 0% -2.60% (p\u003d0.000 n\u003d10)\n4/16/2/0 23.63n ± 1% 22.96n ± 0% -2.85% (p\u003d0.000 n\u003d10)\n4/16/2/1 23.58n ± 1% 23.02n ± 1% -2.38% (p\u003d0.000 n\u003d10)\n4/16/2/2 23.59n ± 0% 23.04n ± 1% -2.32% (p\u003d0.000 n\u003d10)\n4/16/2/3 23.59n ± 0% 23.00n ± 1% -2.50% (p\u003d0.000 n\u003d10)\n4/16/3/0 23.64n ± 0% 23.04n ± 2% -2.57% (p\u003d0.002 n\u003d10)\n4/16/3/1 23.59n ± 0% 23.06n ± 1% -2.23% (p\u003d0.000 n\u003d10)\n4/16/3/2 23.57n ± 1% 23.00n ± 1% -2.40% (p\u003d0.000 n\u003d10)\n4/16/3/3 23.62n ± 1% 22.98n ± 1% -2.73% (p\u003d0.000 n\u003d10)\n4/16/4/4 82.87n ± 0% 80.25n ± 1% -3.16% (p\u003d0.000 n\u003d10)\n16/4/0/0 33.00n ± 1% 15.07n ± 0% -54.32% (p\u003d0.000 n\u003d10)\n16/4/0/1 32.98n ± 1% 15.05n ± 0% -54.37% (p\u003d0.000 n\u003d10)\n16/4/0/2 33.13n ± 3% 15.06n ± 1% -54.53% (p\u003d0.000 n\u003d10)\n16/4/0/3 32.96n ± 4% 15.07n ± 1% -54.27% (p\u003d0.000 n\u003d10)\n16/4/1/0 32.98n ± 1% 15.08n ± 1% -54.26% (p\u003d0.000 n\u003d10)\n16/4/1/1 32.97n ± 2% 15.06n ± 0% -54.33% (p\u003d0.000 n\u003d10)\n16/4/1/2 33.16n ± 1% 15.06n ± 0% -54.58% (p\u003d0.000 n\u003d10)\n16/4/1/3 33.13n ± 1% 15.08n ± 1% -54.49% (p\u003d0.000 n\u003d10)\n16/4/2/0 33.02n ± 1% 15.07n ± 0% -54.35% (p\u003d0.000 n\u003d10)\n16/4/2/1 33.03n ± 2% 15.06n ± 1% -54.40% (p\u003d0.000 n\u003d10)\n16/4/2/2 33.17n ± 1% 15.07n ± 1% -54.57% (p\u003d0.000 n\u003d10)\n16/4/2/3 33.21n ± 3% 15.06n ± 0% -54.64% (p\u003d0.000 n\u003d10)\n16/4/3/0 33.14n ± 3% 15.07n ± 0% -54.52% (p\u003d0.000 n\u003d10)\n16/4/3/1 33.11n ± 3% 15.07n ± 1% -54.47% (p\u003d0.000 n\u003d10)\n16/4/3/2 33.07n ± 3% 15.06n ± 0% -54.46% (p\u003d0.000 n\u003d10)\n16/4/3/3 33.12n ± 2% 15.10n ± 0% -54.41% (p\u003d0.000 n\u003d10)\n16/4/4/4 108.0n ± 1% 105.9n ± 0% -1.97% (p\u003d0.000 n\u003d10)\n8/32/0/0 38.86n ± 0% 41.71n ± 0% +7.34% (p\u003d0.000 n\u003d10)\n8/32/0/1 38.70n ± 1% 41.75n ± 0% +7.89% (p\u003d0.000 n\u003d10)\n8/32/0/2 38.82n ± 1% 41.70n ± 0% +7.42% (p\u003d0.000 n\u003d10)\n8/32/0/3 38.70n ± 1% 41.74n ± 0% +7.84% (p\u003d0.000 n\u003d10)\n8/32/1/0 38.78n ± 1% 41.76n ± 1% +7.70% (p\u003d0.000 n\u003d10)\n8/32/1/1 38.74n ± 1% 41.78n ± 1% +7.85% (p\u003d0.000 n\u003d10)\n8/32/1/2 38.69n ± 1% 41.84n ± 1% +8.14% (p\u003d0.000 n\u003d10)\n8/32/1/3 38.70n ± 0% 41.76n ± 1% +7.90% (p\u003d0.000 n\u003d10)\n8/32/2/0 38.69n ± 0% 41.74n ± 1% +7.88% (p\u003d0.000 n\u003d10)\n8/32/2/1 38.77n ± 0% 41.81n ± 1% +7.83% (p\u003d0.000 n\u003d10)\n8/32/2/2 38.73n ± 0% 41.69n ± 1% +7.65% (p\u003d0.000 n\u003d10)\n8/32/2/3 38.79n ± 1% 41.76n ± 0% +7.68% (p\u003d0.000 n\u003d10)\n8/32/3/0 38.71n ± 1% 41.82n ± 1% +8.04% (p\u003d0.000 n\u003d10)\n8/32/3/1 38.77n ± 0% 41.93n ± 1% +8.16% (p\u003d0.000 n\u003d10)\n8/32/3/2 38.68n ± 1% 41.76n ± 0% +7.96% (p\u003d0.000 n\u003d10)\n8/32/3/3 38.72n ± 1% 41.76n ± 0% +7.84% (p\u003d0.000 n\u003d10)\n8/32/4/4 181.9n ± 1% 183.9n ± 1% +1.11% (p\u003d0.000 n\u003d10)\n32/8/0/0 67.06n ± 0% 24.48n ± 1% -63.50% (p\u003d0.000 n\u003d10)\n32/8/0/1 67.29n ± 1% 24.49n ± 0% -63.61% (p\u003d0.000 n\u003d10)\n32/8/0/2 67.43n ± 2% 24.50n ± 1% -63.67% (p\u003d0.000 n\u003d10)\n32/8/0/3 67.71n ± 2% 24.48n ± 1% -63.84% (p\u003d0.000 n\u003d10)\n32/8/1/0 67.16n ± 1% 24.48n ± 0% -63.54% (p\u003d0.000 n\u003d10)\n32/8/1/1 67.38n ± 1% 24.49n ± 0% -63.66% (p\u003d0.000 n\u003d10)\n32/8/1/2 67.31n ± 1% 24.49n ± 0% -63.62% (p\u003d0.000 n\u003d10)\n32/8/1/3 67.08n ± 1% 24.46n ± 0% -63.54% (p\u003d0.000 n\u003d10)\n32/8/2/0 67.17n ± 1% 24.48n ± 0% -63.55% (p\u003d0.000 n\u003d10)\n32/8/2/1 67.44n ± 1% 24.46n ± 0% -63.72% (p\u003d0.000 n\u003d10)\n32/8/2/2 66.96n ± 2% 24.46n ± 1% -63.47% (p\u003d0.000 n\u003d10)\n32/8/2/3 67.10n ± 0% 24.46n ± 0% -63.55% (p\u003d0.000 n\u003d10)\n32/8/3/0 67.17n ± 2% 24.49n ± 0% -63.55% (p\u003d0.000 n\u003d10)\n32/8/3/1 67.03n ± 1% 24.46n ± 1% -63.50% (p\u003d0.000 n\u003d10)\n32/8/3/2 67.07n ± 1% 24.49n ± 1% -63.48% (p\u003d0.000 n\u003d10)\n32/8/3/3 67.69n ± 2% 24.49n ± 1% -63.82% (p\u003d0.000 n\u003d10)\n32/8/4/4 275.3n ± 1% 269.0n ± 1% -2.28% (p\u003d0.000 n\u003d10)\n16/64/0/0 117.7n ± 0% 103.7n ± 5% -11.91% (p\u003d0.000 n\u003d10)\n16/64/0/1 117.6n ± 0% 103.2n ± 4% -12.26% (p\u003d0.000 n\u003d10)\n16/64/0/2 117.6n ± 0% 103.3n ± 3% -12.15% (p\u003d0.000 n\u003d10)\n16/64/0/3 117.7n ± 0% 102.7n ± 4% -12.75% (p\u003d0.000 n\u003d10)\n16/64/1/0 117.7n ± 0% 102.6n ± 2% -12.81% (p\u003d0.000 n\u003d10)\n16/64/1/1 117.6n ± 0% 102.5n ± 2% -12.80% (p\u003d0.000 n\u003d10)\n16/64/1/2 117.6n ± 0% 104.3n ± 1% -11.34% (p\u003d0.000 n\u003d10)\n16/64/1/3 117.5n ± 1% 102.7n ± 3% -12.59% (p\u003d0.000 n\u003d10)\n16/64/2/0 117.6n ± 0% 104.2n ± 3% -11.46% (p\u003d0.000 n\u003d10)\n16/64/2/1 117.7n ± 0% 103.1n ± 1% -12.43% (p\u003d0.000 n\u003d10)\n16/64/2/2 117.6n ± 0% 102.5n ± 2% -12.88% (p\u003d0.000 n\u003d10)\n16/64/2/3 117.5n ± 0% 102.8n ± 1% -12.53% (p\u003d0.000 n\u003d10)\n16/64/3/0 117.6n ± 0% 102.3n ± 5% -13.07% (p\u003d0.000 n\u003d10)\n16/64/3/1 117.5n ± 0% 104.1n ± 2% -11.42% (p\u003d0.000 n\u003d10)\n16/64/3/2 117.5n ± 1% 102.8n ± 5% -12.47% (p\u003d0.000 n\u003d10)\n16/64/3/3 117.6n ± 0% 102.6n ± 4% -12.70% (p\u003d0.000 n\u003d10)\n16/64/4/4 640.2n ± 0% 641.0n ± 1% ~ (p\u003d0.971 n\u003d10)\n64/16/0/0 160.98n ± 0% 32.72n ± 1% -79.68% (p\u003d0.000 n\u003d10)\n64/16/0/1 160.69n ± 1% 32.75n ± 0% -79.62% (p\u003d0.000 n\u003d10)\n64/16/0/2 160.67n ± 1% 32.73n ± 0% -79.63% (p\u003d0.000 n\u003d10)\n64/16/0/3 160.72n ± 1% 32.77n ± 0% -79.61% (p\u003d0.000 n\u003d10)\n64/16/1/0 160.40n ± 1% 32.72n ± 0% -79.60% (p\u003d0.000 n\u003d10)\n64/16/1/1 160.56n ± 0% 32.71n ± 0% -79.63% (p\u003d0.000 n\u003d10)\n64/16/1/2 160.52n ± 0% 32.74n ± 1% -79.60% (p\u003d0.000 n\u003d10)\n64/16/1/3 160.37n ± 0% 32.76n ± 0% -79.57% (p\u003d0.000 n\u003d10)\n64/16/2/0 161.31n ± 1% 32.71n ± 0% -79.72% (p\u003d0.000 n\u003d10)\n64/16/2/1 160.38n ± 1% 32.75n ± 1% -79.58% (p\u003d0.000 n\u003d10)\n64/16/2/2 161.01n ± 0% 32.71n ± 0% -79.69% (p\u003d0.000 n\u003d10)\n64/16/2/3 160.20n ± 1% 32.73n ± 1% -79.57% (p\u003d0.000 n\u003d10)\n64/16/3/0 160.57n ± 1% 32.72n ± 1% -79.62% (p\u003d0.000 n\u003d10)\n64/16/3/1 160.45n ± 1% 32.73n ± 0% -79.60% (p\u003d0.000 n\u003d10)\n64/16/3/2 160.41n ± 1% 32.70n ± 0% -79.61% (p\u003d0.000 n\u003d10)\n64/16/3/3 160.19n ± 1% 32.71n ± 0% -79.58% (p\u003d0.000 n\u003d10)\n64/16/4/4 826.2n ± 1% 809.8n ± 1% -1.99% (p\u003d0.000 n\u003d10)\n\nChange-Id: I5cbaf59efaeb5f10b1fc8bf69d2d31db9bf7820b\n", "tree_diff": [ { "type": "modify", "old_id": "ffafa13319e067e52b83f9686cbb59eb63344093", "old_mode": 33188, "old_path": "aom_dsp/x86/convolve_avx2.h", "new_id": "1191a28a67e0aa95e16e2b099b916b32c35dff02", "new_mode": 33188, "new_path": "aom_dsp/x86/convolve_avx2.h" }, { "type": "modify", "old_id": "be53424ac729a38becf6c0cf0d4ece3cf9b59e88", "old_mode": 33188, "old_path": "av1/common/x86/convolve_2d_avx2.c", "new_id": "df4a7d5516d600519d1952c4e514e1b8a6a661fd", "new_mode": 33188, "new_path": "av1/common/x86/convolve_2d_avx2.c" } ] }