rust-math / intel-mkl-src

Redistribute Intel MKL as a crate
Other
78 stars 22 forks source link

intel-mkl-sys crate #26

Closed termoshtt closed 4 years ago

termoshtt commented 4 years ago

Resolve #11

Rust binding to Intel-MKL including

termoshtt commented 4 years ago

Needs test for linking to vector math functions

termoshtt commented 4 years ago

x4-x8 faster on Intel(R) Core(TM) i7-8700K CPU

cos32_n100              time:   [492.58 ns 506.58 ns 519.05 ns]
                        change: [-16.192% -14.353% -12.600%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 17 outliers among 100 measurements (17.00%)
  16 (16.00%) low severe
  1 (1.00%) low mild

vcos32_n100             time:   [89.886 ns 90.055 ns 90.254 ns]
                        change: [-3.9748% -3.1782% -2.5030%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 8 outliers among 100 measurements (8.00%)
  4 (4.00%) high mild
  4 (4.00%) high severe

cos32_n1000             time:   [5.3297 us 5.3519 us 5.3737 us]
                        change: [-15.456% -13.311% -10.880%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 7 outliers among 100 measurements (7.00%)
  1 (1.00%) low severe
  2 (2.00%) low mild
  2 (2.00%) high mild
  2 (2.00%) high severe

vcos32_n1000            time:   [619.01 ns 620.13 ns 621.28 ns]
                        change: [-0.5284% -0.1526% +0.2074%] (p = 0.43 > 0.05)
                        No change in performance detected.
Found 2 outliers among 100 measurements (2.00%)
  1 (1.00%) high mild
  1 (1.00%) high severe

cos32_n10000            time:   [75.502 us 76.042 us 76.485 us]
                        change: [-4.6132% -4.0370% -3.4214%] (p = 0.00 < 0.05)
                        Performance has improved.

vcos32_n10000           time:   [5.9387 us 5.9669 us 5.9976 us]
                        change: [+0.0897% +0.7740% +1.4877%] (p = 0.03 < 0.05)
                        Change within noise threshold.
Found 3 outliers among 100 measurements (3.00%)
  1 (1.00%) high mild
  2 (2.00%) high severe

cos64_n100              time:   [837.27 ns 840.64 ns 844.42 ns]
                        change: [-17.706% -17.100% -16.452%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
  5 (5.00%) high mild
  1 (1.00%) high severe

vcos64_n100             time:   [192.40 ns 192.68 ns 193.00 ns]
                        change: [-14.937% -13.983% -13.124%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  3 (3.00%) high mild
  2 (2.00%) high severe

cos64_n1000             time:   [12.295 us 12.521 us 12.699 us]
                        change: [-9.0396% -4.8883% -0.6860%] (p = 0.03 < 0.05)
                        Change within noise threshold.
Found 23 outliers among 100 measurements (23.00%)
  23 (23.00%) low mild

vcos64_n1000            time:   [1.7505 us 1.7579 us 1.7659 us]
                        change: [-11.441% -10.646% -9.6345%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  4 (4.00%) high mild
  5 (5.00%) high severe

cos64_n10000            time:   [140.17 us 140.50 us 140.87 us]
                        change: [-0.6191% +0.0270% +0.6612%] (p = 0.94 > 0.05)
                        No change in performance detected.
Found 6 outliers among 100 measurements (6.00%)
  5 (5.00%) high mild
  1 (1.00%) high severe

vcos64_n10000           time:   [17.238 us 17.270 us 17.305 us]
                        change: [-12.466% -11.820% -11.249%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 2 outliers among 100 measurements (2.00%)
  1 (1.00%) high mild
  1 (1.00%) high severe