overview
schur / original
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 92.0 ms | 94.3 ms | 95.8 ms |
| R² goodness-of-fit | 0.999 | 0.999 | 1.00 |
| Mean execution time | 92.8 ms | 94.3 ms | 96.4 ms |
| Standard deviation | 686 μs | 2.93 ms | 4.69 ms |
Outlying measurements have a slight (9.88%) effect on estimated standard deviation.
schur / hasktorch-cpu
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 51.0 ms | 87.9 ms | 139 ms |
| R² goodness-of-fit | 0.577 | 0.694 | 0.996 |
| Mean execution time | 51.4 ms | 59.2 ms | 81.9 ms |
| Standard deviation | 5.99 ms | 23.0 ms | 38.8 ms |
Outlying measurements have a severe (90.1%) effect on estimated standard deviation.
schur / hasktorch-cuda
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 22.6 ms | 22.9 ms | 23.3 ms |
| R² goodness-of-fit | 0.999 | 0.999 | 1.00 |
| Mean execution time | 23.6 ms | 24.5 ms | 26.8 ms |
| Standard deviation | 540 μs | 2.94 ms | 5.04 ms |
Outlying measurements have a severe (53.4%) effect on estimated standard deviation.
woodbury / original
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 42.3 ms | 43.0 ms | 43.9 ms |
| R² goodness-of-fit | 0.999 | 0.999 | 1.00 |
| Mean execution time | 42.5 ms | 42.9 ms | 43.7 ms |
| Standard deviation | 496 μs | 1.07 ms | 1.48 ms |
Outlying measurements have a slight (6.63%) effect on estimated standard deviation.
woodbury / hasktorch-cpu
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 54.1 ms | 60.7 ms | 68.9 ms |
| R² goodness-of-fit | 0.874 | 0.957 | 0.996 |
| Mean execution time | 60.8 ms | 65.8 ms | 71.1 ms |
| Standard deviation | 6.02 ms | 9.12 ms | 14.8 ms |
Outlying measurements have a moderate (43.9%) effect on estimated standard deviation.
woodbury / hasktorch-cuda
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 814 μs | 816 μs | 818 μs |
| R² goodness-of-fit | 1.00 | 1.00 | 1.00 |
| Mean execution time | 816 μs | 817 μs | 818 μs |
| Standard deviation | 3.13 μs | 4.03 μs | 5.02 μs |
Outlying measurements have a slight (1.33%) effect on estimated standard deviation.
direct-inversion / original
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 82.3 ms | 96.9 ms | 116 ms |
| R² goodness-of-fit | 0.859 | 0.939 | 0.986 |
| Mean execution time | 65.1 ms | 73.3 ms | 84.4 ms |
| Standard deviation | 12.8 ms | 16.0 ms | 19.6 ms |
Outlying measurements have a severe (68.4%) effect on estimated standard deviation.
direct-inversion / hasktorch-cpu
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 74.6 ms | 86.5 ms | 103 ms |
| R² goodness-of-fit | 0.854 | 0.938 | 0.991 |
| Mean execution time | 78.3 ms | 87.3 ms | 94.6 ms |
| Standard deviation | 8.78 ms | 14.2 ms | 21.6 ms |
Outlying measurements have a severe (57.9%) effect on estimated standard deviation.
direct-inversion / hasktorch-cuda
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 23.7 ms | 24.3 ms | 25.0 ms |
| R² goodness-of-fit | 0.997 | 0.999 | 1.00 |
| Mean execution time | 25.2 ms | 26.4 ms | 29.2 ms |
| Standard deviation | 1.07 ms | 3.85 ms | 5.80 ms |
Outlying measurements have a severe (61.6%) effect on estimated standard deviation.
indirect-inversion / original
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 55.7 ms | 64.9 ms | 77.9 ms |
| R² goodness-of-fit | 0.896 | 0.952 | 0.992 |
| Mean execution time | 52.0 ms | 58.2 ms | 63.0 ms |
| Standard deviation | 5.89 ms | 9.94 ms | 16.1 ms |
Outlying measurements have a severe (56.8%) effect on estimated standard deviation.
indirect-inversion / hasktorch-cpu
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 34.6 ms | 57.4 ms | 73.9 ms |
| R² goodness-of-fit | 0.623 | 0.826 | 0.975 |
| Mean execution time | 49.8 ms | 56.7 ms | 64.3 ms |
| Standard deviation | 10.1 ms | 13.4 ms | 19.5 ms |
Outlying measurements have a severe (73.7%) effect on estimated standard deviation.
indirect-inversion / hasktorch-cuda
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 972 μs | 975 μs | 977 μs |
| R² goodness-of-fit | 1.00 | 1.00 | 1.00 |
| Mean execution time | 975 μs | 977 μs | 979 μs |
| Standard deviation | 5.26 μs | 6.77 μs | 9.55 μs |
Outlying measurements have a slight (1.39%) effect on estimated standard deviation.
dense-inversion / original
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 62.4 ms | 65.0 ms | 67.5 ms |
| R² goodness-of-fit | 0.993 | 0.997 | 1.00 |
| Mean execution time | 63.3 ms | 64.2 ms | 65.5 ms |
| Standard deviation | 1.39 ms | 1.94 ms | 2.45 ms |
Outlying measurements have a slight (8.26%) effect on estimated standard deviation.
dense-inversion / hasktorch-cpu
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 60.1 ms | 63.1 ms | 64.7 ms |
| R² goodness-of-fit | 0.984 | 0.995 | 1.00 |
| Mean execution time | 59.9 ms | 61.6 ms | 62.7 ms |
| Standard deviation | 973 μs | 2.54 ms | 3.70 ms |
Outlying measurements have a slight (7.80%) effect on estimated standard deviation.
dense-inversion / hasktorch-cuda
| lower bound | estimate | upper bound | |
|---|---|---|---|
| OLS regression | 23.5 ms | 23.9 ms | 24.4 ms |
| R² goodness-of-fit | 0.998 | 0.999 | 1.00 |
| Mean execution time | 24.4 ms | 25.5 ms | 27.7 ms |
| Standard deviation | 86.4 μs | 3.30 ms | 5.16 ms |
Outlying measurements have a severe (56.3%) effect on estimated standard deviation.