martinus
commented
2 years ago Benchmark results:
Linux, clang++ 13.0.1: ankerl::unordered_dense::map |
ns/op | op/s | err% | ins/op | cyc/op | IPC | bra/op | miss% | total | benchmarking |
|---|---|---|---|---|---|---|---|---|---|---|
| 10,456,378.00 | 95.64 | 0.1% | 101,029,741.00 | 33,226,459.00 | 3.041 | 25,109,968.00 | 0.1% | 0.12 | ankerl::unordered_dense::map<uint64_t, size_t> iterate while adding then removing |
|
| 128,132,576.00 | 7.80 | 0.3% | 542,400,290.00 | 406,468,349.00 | 1.334 | 61,729,425.00 | 12.0% | 1.42 | ankerl::unordered_dense::map<uint64_t, size_t> random insert erase |
|
| 138,092,089.00 | 7.24 | 0.3% | 584,996,547.00 | 438,560,451.00 | 1.334 | 100,922,347.00 | 6.5% | 1.53 | ankerl::unordered_dense::map<uint64_t, size_t> 50% probability to find |
|
| 15,266,999.00 | 65.50 | 0.3% | 106,432,022.00 | 48,693,694.00 | 2.186 | 25,921,308.00 | 0.1% | 0.18 | ankerl::unordered_dense::map<std::string, size_t> iterate while adding then removing |
|
| 522,252,900.00 | 1.91 | 0.8% | 2,889,992,573.00 | 1,656,054,820.00 | 1.745 | 352,929,113.00 | 3.9% | 5.77 | ankerl::unordered_dense::map<std::string, size_t> random insert erase |
|
| 398,917,499.00 | 2.51 | 0.2% | 2,589,100,679.00 | 1,255,316,510.00 | 2.063 | 318,743,586.00 | 2.1% | 4.40 | ankerl::unordered_dense::map<std::string, size_t> 50% probability to find |
robin_hood::unordered_flat_map |
ns/op | op/s | err% | ins/op | cyc/op | IPC | bra/op | miss% | total | benchmarking |
|---|---|---|---|---|---|---|---|---|---|---|
| 88,779,774.00 | 11.26 | 0.2% | 363,295,754.00 | 282,963,884.00 | 1.284 | 52,991,290.00 | 0.8% | 1.00 | robin_hood::unordered_flat_map<uint64_t, size_t> iterate while adding then removing |
|
| 182,156,559.00 | 5.49 | 0.1% | 648,560,823.00 | 580,588,757.00 | 1.117 | 65,227,482.00 | 20.0% | 2.05 | robin_hood::unordered_flat_map<uint64_t, size_t> random insert erase |
|
| 161,153,746.00 | 6.21 | 0.2% | 699,766,270.00 | 511,953,708.00 | 1.367 | 73,791,117.00 | 11.5% | 1.81 | robin_hood::unordered_flat_map<uint64_t, size_t> 50% probability to find |
|
| 90,357,941.00 | 11.07 | 0.2% | 371,978,741.00 | 288,236,935.00 | 1.291 | 54,180,921.00 | 0.8% | 1.01 | robin_hood::unordered_flat_map<std::string, size_t> iterate while adding then removing |
|
| 661,080,369.00 | 1.51 | 0.2% | 4,210,932,698.00 | 2,108,445,010.00 | 1.997 | 469,336,688.00 | 2.6% | 7.35 | robin_hood::unordered_flat_map<std::string, size_t> random insert erase |
|
| 601,181,881.00 | 1.66 | 0.4% | 3,838,032,670.00 | 1,915,156,243.00 | 2.004 | 330,429,865.00 | 2.5% | 6.69 | robin_hood::unordered_flat_map<std::string, size_t> 50% probability to find |
std::unordered_map |
ns/op | op/s | err% | ins/op | cyc/op | IPC | bra/op | miss% | total | benchmarking |
|---|---|---|---|---|---|---|---|---|---|---|
| 98,298,332.00 | 10.17 | 1.0% | 102,527,853.00 | 312,980,521.00 | 0.328 | 25,523,728.00 | 0.1% | 1.06 | std::unordered_map<uint64_t, size_t> iterate while adding then removing |
|
| 222,034,779.00 | 4.50 | 0.1% | 700,586,088.00 | 708,149,988.00 | 0.989 | 132,799,451.00 | 6.6% | 2.47 | std::unordered_map<uint64_t, size_t> random insert erase |
|
| 322,060,122.00 | 3.11 | 0.0% | 506,422,457.00 | 1,026,687,249.00 | 0.493 | 86,441,657.00 | 11.8% | 3.56 | std::unordered_map<uint64_t, size_t> 50% probability to find |
|
| 332,284,553.00 | 3.01 | 0.2% | 108,712,711.00 | 1,059,923,550.00 | 0.103 | 26,476,254.00 | 0.1% | 3.67 | std::unordered_map<std::string, size_t> iterate while adding then removing |
|
| 928,712,123.00 | 1.08 | 0.1% | 4,284,647,292.00 | 2,962,368,185.00 | 1.446 | 578,118,501.00 | 1.6% | 10.24 | std::unordered_map<std::string, size_t> random insert erase |
|
| 1,043,774,629.00 | 0.96 | 0.2% | 4,269,617,190.00 | 3,328,972,577.00 | 1.283 | 494,180,272.00 | 2.0% | 11.64 | std::unordered_map<std::string, size_t> 50% probability to find |
std::hash is really slow on windows, generates a hell of a lot of
imul. Don't fall back to it, until really necessary