MarkCiliaVincenti
commented
3 months ago Can you make benchmark for this change?
https://github.com/MarkCiliaVincenti/AsyncKeyedLockBenchmarks/tree/EasyCaching
They have run transparently and openly on Github Actions at https://github.com/MarkCiliaVincenti/AsyncKeyedLockBenchmarks/actions/runs/8796459745/job/24139360286
As you can see, the new version scales better, is faster and uses less memory.
| Results reproduced here (but can be confirmed in the link above) for convenience: | Method | NumberOfLocks | Contention | GuidReversals | Mean | Error | StdDev | Median | Ratio | RatioSD | Gen0 | Gen1 | Gen2 | Allocated | Alloc Ratio |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 'MemoryLock with AsyncKeyedLock' | 200 | 100 | 0 | 15,349.707 us | 942.4231 us | 2,563.9315 us | 15,401.451 us | 0.78 | 0.27 | - | - | - | 5325728 B | 0.79 | |
| 'Current MemoryLock' | 200 | 100 | 0 | 20,258.736 us | 2,389.7323 us | 7,046.1796 us | 18,174.753 us | 1.00 | 0.00 | - | - | - | 6765728 B | 1.00 | |
| 'MemoryLock with AsyncKeyedLock' | 200 | 100 | 1 | 39,420.331 us | 780.4782 us | 1,680.0616 us | 39,058.047 us | 1.03 | 0.06 | - | - | - | 8206016 B | 0.85 | |
| 'Current MemoryLock' | 200 | 100 | 1 | 38,664.992 us | 770.3120 us | 1,484.1297 us | 38,613.519 us | 1.00 | 0.00 | - | - | - | 9646016 B | 1.00 | |
| 'MemoryLock with AsyncKeyedLock' | 200 | 100 | 5 | 104,775.402 us | 3,483.1621 us | 10,270.1821 us | 110,903.831 us | 1.02 | 0.16 | - | - | - | 19726016 B | 0.93 | |
| 'Current MemoryLock' | 200 | 100 | 5 | 103,829.741 us | 3,822.0693 us | 11,269.4577 us | 110,369.754 us | 1.00 | 0.00 | - | - | - | 21166016 B | 1.00 | |
| 'MemoryLock with AsyncKeyedLock' | 200 | 10000 | 0 | 2,088,942.819 us | 35,690.5415 us | 33,384.9546 us | 2,079,745.230 us | 0.93 | 0.02 | 6000.0000 | 5000.0000 | 1000.0000 | 513565136 B | 0.78 | |
| 'Current MemoryLock' | 200 | 10000 | 0 | 2,248,456.686 us | 24,612.5247 us | 23,022.5709 us | 2,245,880.343 us | 1.00 | 0.00 | 9000.0000 | 8000.0000 | 2000.0000 | 657558072 B | 1.00 | |
| 'MemoryLock with AsyncKeyedLock' | 200 | 10000 | 1 | 4,648,699.755 us | 19,148.6526 us | 17,911.6615 us | 4,649,405.561 us | 0.96 | 0.01 | 11000.0000 | 10000.0000 | 2000.0000 | 801564152 B | 0.85 | |
| 'Current MemoryLock' | 200 | 10000 | 1 | 4,844,578.714 us | 33,927.0917 us | 31,735.4226 us | 4,852,854.820 us | 1.00 | 0.00 | 12000.0000 | 11000.0000 | 2000.0000 | 945567296 B | 1.00 | |
| 'MemoryLock with AsyncKeyedLock' | 200 | 10000 | 5 | 11,786,893.676 us | 225,429.2772 us | 231,499.2705 us | 11,836,810.900 us | 0.99 | 0.02 | 26000.0000 | 25000.0000 | 4000.0000 | 1953582520 B | 0.93 | |
| 'Current MemoryLock' | 200 | 10000 | 5 | 11,923,912.383 us | 202,259.5721 us | 189,193.7292 us | 11,929,389.177 us | 1.00 | 0.00 | 28000.0000 | 27000.0000 | 4000.0000 | 2097570712 B | 1.00 | |
| 'MemoryLock with AsyncKeyedLock' | 10000 | 100 | 0 | 1,094,885.648 us | 21,835.5041 us | 21,445.3847 us | 1,100,071.313 us | 0.95 | 0.03 | 4000.0000 | 3000.0000 | 1000.0000 | 287819760 B | 0.80 | |
| 'Current MemoryLock' | 10000 | 100 | 0 | 1,147,084.853 us | 20,032.7784 us | 16,728.2772 us | 1,151,722.032 us | 1.00 | 0.00 | 5000.0000 | 4000.0000 | 1000.0000 | 359819760 B | 1.00 | |
| 'MemoryLock with AsyncKeyedLock' | 10000 | 100 | 1 | 2,303,743.181 us | 15,977.9608 us | 14,945.7944 us | 2,302,793.965 us | 0.98 | 0.01 | 5000.0000 | 4000.0000 | 1000.0000 | 431827584 B | 0.86 | |
| 'Current MemoryLock' | 10000 | 100 | 1 | 2,358,618.870 us | 18,135.4939 us | 16,963.9522 us | 2,355,659.956 us | 1.00 | 0.00 | 6000.0000 | 5000.0000 | 1000.0000 | 503823032 B | 1.00 | |
| 'MemoryLock with AsyncKeyedLock' | 10000 | 100 | 5 | 5,660,138.085 us | 113,166.7531 us | 192,165.7354 us | 5,669,727.557 us | 0.98 | 0.04 | 13000.0000 | 12000.0000 | 2000.0000 | 1007824744 B | 0.93 | |
| 'Current MemoryLock' | 10000 | 100 | 5 | 5,836,943.953 us | 116,096.7156 us | 124,222.1850 us | 5,849,565.337 us | 1.00 | 0.00 | 15000.0000 | 14000.0000 | 3000.0000 | 1079827320 B | 1.00 |
Can you make benchmark for this change?