clintropolis commented on PR #12277: URL: https://github.com/apache/druid/pull/12277#issuecomment-1255538371
the slowdown on select * is also no longer quite as dramatic as in my first screenshots: <img width="1182" alt="Screen Shot 2022-09-22 at 1 50 23 PM" src="https://user-images.githubusercontent.com/1577461/191848586-d0211d6a-0510-465d-8909-a16bfbbeb7c7.png";> <img width="1176" alt="Screen Shot 2022-09-22 at 1 49 59 PM" src="https://user-images.githubusercontent.com/1577461/191848610-96e8e431-2eb6-4dfc-8b75-7989fe23318b.png";> <img width="1210" alt="Screen Shot 2022-09-22 at 1 49 32 PM" src="https://user-images.githubusercontent.com/1577461/191848633-f47f8b8a-13f1-4d13-86af-de927577431c.png";> For reference, these were the first set of benchmarks when I opened the PR (because i have edited the PR description with the updated benchmarks): ``` Benchmark (indexType) (numElements) (numOperations) (width) Mode Cnt Score Error Units FrontCodedIndexedBenchmark.get generic 10000 10000 16 avgt 5 78.819 ± 2.541 ns/op FrontCodedIndexedBenchmark.get:encoded size generic 10000 10000 16 avgt 5 240010.000 bytes FrontCodedIndexedBenchmark.get front-coded-4 10000 10000 16 avgt 5 220.887 ± 19.267 ns/op FrontCodedIndexedBenchmark.get:encoded size front-coded-4 10000 10000 16 avgt 5 169894.000 bytes FrontCodedIndexedBenchmark.get front-coded-16 10000 10000 16 avgt 5 629.621 ± 59.667 ns/op FrontCodedIndexedBenchmark.get:encoded size front-coded-16 10000 10000 16 avgt 5 164428.000 bytes FrontCodedIndexedBenchmark.indexOf generic 10000 10000 16 avgt 5 1228.012 ± 100.691 ns/op FrontCodedIndexedBenchmark.indexOf:encoded size generic 10000 10000 16 avgt 5 240010.000 bytes FrontCodedIndexedBenchmark.indexOf front-coded-4 10000 10000 16 avgt 5 2101.530 ± 209.465 ns/op FrontCodedIndexedBenchmark.indexOf:encoded size front-coded-4 10000 10000 16 avgt 5 169908.000 bytes FrontCodedIndexedBenchmark.indexOf front-coded-16 10000 10000 16 avgt 5 2364.248 ± 156.352 ns/op FrontCodedIndexedBenchmark.indexOf:encoded size front-coded-16 10000 10000 16 avgt 5 164321.000 bytes FrontCodedIndexedBenchmark.iterator generic 10000 10000 16 avgt 5 80.247 ± 1.639 ns/op FrontCodedIndexedBenchmark.iterator:encoded size generic 10000 10000 16 avgt 5 240010.000 bytes FrontCodedIndexedBenchmark.iterator front-coded-4 10000 10000 16 avgt 5 101.921 ± 6.225 ns/op FrontCodedIndexedBenchmark.iterator:encoded size front-coded-4 10000 10000 16 avgt 5 169889.000 bytes FrontCodedIndexedBenchmark.iterator front-coded-16 10000 10000 16 avgt 5 105.235 ± 2.333 ns/op FrontCodedIndexedBenchmark.iterator:encoded size front-coded-16 10000 10000 16 avgt 5 164221.000 bytes FrontCodedIndexedBenchmark.get generic 100000 10000 16 avgt 5 95.947 ± 3.842 ns/op FrontCodedIndexedBenchmark.get:encoded size generic 100000 10000 16 avgt 5 2400010.000 bytes FrontCodedIndexedBenchmark.get front-coded-4 100000 10000 16 avgt 5 239.630 ± 40.148 ns/op FrontCodedIndexedBenchmark.get:encoded size front-coded-4 100000 10000 16 avgt 5 1636718.000 bytes FrontCodedIndexedBenchmark.get front-coded-16 100000 10000 16 avgt 5 648.001 ± 28.965 ns/op FrontCodedIndexedBenchmark.get:encoded size front-coded-16 100000 10000 16 avgt 5 1564861.000 bytes FrontCodedIndexedBenchmark.indexOf generic 100000 10000 16 avgt 5 1612.575 ± 149.556 ns/op FrontCodedIndexedBenchmark.indexOf:encoded size generic 100000 10000 16 avgt 5 2400010.000 bytes FrontCodedIndexedBenchmark.indexOf front-coded-4 100000 10000 16 avgt 5 2934.297 ± 176.182 ns/op FrontCodedIndexedBenchmark.indexOf:encoded size front-coded-4 100000 10000 16 avgt 5 1636652.000 bytes FrontCodedIndexedBenchmark.indexOf front-coded-16 100000 10000 16 avgt 5 3235.271 ± 438.586 ns/op FrontCodedIndexedBenchmark.indexOf:encoded size front-coded-16 100000 10000 16 avgt 5 1563940.000 bytes FrontCodedIndexedBenchmark.iterator generic 100000 10000 16 avgt 5 787.253 ± 83.715 ns/op FrontCodedIndexedBenchmark.iterator:encoded size generic 100000 10000 16 avgt 5 2400010.000 bytes FrontCodedIndexedBenchmark.iterator front-coded-4 100000 10000 16 avgt 5 1048.532 ± 29.200 ns/op FrontCodedIndexedBenchmark.iterator:encoded size front-coded-4 100000 10000 16 avgt 5 1636546.000 bytes FrontCodedIndexedBenchmark.iterator front-coded-16 100000 10000 16 avgt 5 1054.068 ± 66.667 ns/op FrontCodedIndexedBenchmark.iterator:encoded size front-coded-16 100000 10000 16 avgt 5 1565019.000 bytes ``` ``` Benchmark (query) (rowsPerSegment) (stringEncoding) (vectorize) Mode Cnt Score Error Units SqlBenchmark.querySql 4 5000000 none false avgt 5 152.079 ± 7.234 ms/op SqlBenchmark.querySql 4 5000000 none force avgt 5 76.931 ± 3.635 ms/op SqlBenchmark.querySql 4 5000000 fc4 false avgt 5 151.430 ± 11.542 ms/op SqlBenchmark.querySql 4 5000000 fc4 force avgt 5 77.045 ± 3.972 ms/op SqlBenchmark.querySql 4 5000000 fc16 false avgt 5 160.475 ± 16.692 ms/op SqlBenchmark.querySql 4 5000000 fc16 force avgt 5 75.423 ± 4.572 ms/op SqlBenchmark.querySql 5 5000000 none false avgt 5 49.356 ± 2.468 ms/op SqlBenchmark.querySql 5 5000000 none force avgt 5 49.141 ± 2.853 ms/op SqlBenchmark.querySql 5 5000000 fc4 false avgt 5 49.169 ± 2.025 ms/op SqlBenchmark.querySql 5 5000000 fc4 force avgt 5 49.232 ± 2.801 ms/op SqlBenchmark.querySql 5 5000000 fc16 false avgt 5 49.438 ± 2.585 ms/op SqlBenchmark.querySql 5 5000000 fc16 force avgt 5 49.451 ± 3.924 ms/op SqlBenchmark.querySql 6 5000000 none false avgt 5 199.922 ± 7.085 ms/op SqlBenchmark.querySql 6 5000000 none force avgt 5 105.862 ± 7.491 ms/op SqlBenchmark.querySql 6 5000000 fc4 false avgt 5 209.565 ± 21.137 ms/op SqlBenchmark.querySql 6 5000000 fc4 force avgt 5 106.155 ± 7.080 ms/op SqlBenchmark.querySql 6 5000000 fc16 false avgt 5 199.692 ± 11.608 ms/op SqlBenchmark.querySql 6 5000000 fc16 force avgt 5 107.823 ± 9.708 ms/op SqlBenchmark.querySql 7 5000000 none false avgt 5 154.474 ± 7.260 ms/op SqlBenchmark.querySql 7 5000000 none force avgt 5 84.477 ± 4.757 ms/op SqlBenchmark.querySql 7 5000000 fc4 false avgt 5 150.924 ± 8.694 ms/op SqlBenchmark.querySql 7 5000000 fc4 force avgt 5 82.500 ± 4.569 ms/op SqlBenchmark.querySql 7 5000000 fc16 false avgt 5 151.421 ± 7.497 ms/op SqlBenchmark.querySql 7 5000000 fc16 force avgt 5 86.868 ± 3.030 ms/op SqlBenchmark.querySql 10 5000000 none false avgt 5 780.820 ± 68.020 ms/op SqlBenchmark.querySql 10 5000000 none force avgt 5 432.962 ± 62.454 ms/op SqlBenchmark.querySql 10 5000000 fc4 false avgt 5 798.933 ± 71.454 ms/op SqlBenchmark.querySql 10 5000000 fc4 force avgt 5 458.484 ± 37.257 ms/op SqlBenchmark.querySql 10 5000000 fc16 false avgt 5 872.225 ± 49.937 ms/op SqlBenchmark.querySql 10 5000000 fc16 force avgt 5 522.340 ± 35.679 ms/op SqlBenchmark.querySql 12 5000000 none false avgt 5 69.159 ± 4.653 ms/op SqlBenchmark.querySql 12 5000000 none force avgt 5 33.735 ± 2.004 ms/op SqlBenchmark.querySql 12 5000000 fc4 false avgt 5 69.006 ± 4.162 ms/op SqlBenchmark.querySql 12 5000000 fc4 force avgt 5 33.413 ± 1.792 ms/op SqlBenchmark.querySql 12 5000000 fc16 false avgt 5 68.738 ± 3.524 ms/op SqlBenchmark.querySql 12 5000000 fc16 force avgt 5 34.627 ± 2.654 ms/op SqlBenchmark.querySql 19 5000000 none false avgt 5 700.311 ± 100.489 ms/op SqlBenchmark.querySql 19 5000000 none force avgt 5 601.641 ± 39.403 ms/op SqlBenchmark.querySql 19 5000000 fc4 false avgt 5 1056.911 ± 91.692 ms/op SqlBenchmark.querySql 19 5000000 fc4 force avgt 5 963.832 ± 55.223 ms/op SqlBenchmark.querySql 19 5000000 fc16 false avgt 5 2159.028 ± 145.168 ms/op SqlBenchmark.querySql 19 5000000 fc16 force avgt 5 2107.042 ± 123.906 ms/op SqlBenchmark.querySql 21 5000000 none false avgt 5 773.091 ± 46.505 ms/op SqlBenchmark.querySql 21 5000000 none force avgt 5 459.661 ± 36.800 ms/op SqlBenchmark.querySql 21 5000000 fc4 false avgt 5 837.426 ± 41.612 ms/op SqlBenchmark.querySql 21 5000000 fc4 force avgt 5 504.546 ± 18.115 ms/op SqlBenchmark.querySql 21 5000000 fc16 false avgt 5 919.155 ± 106.868 ms/op SqlBenchmark.querySql 21 5000000 fc16 force avgt 5 602.116 ± 7.865 ms/op SqlBenchmark.querySql 22 5000000 none false avgt 5 44.268 ± 2.707 ms/op SqlBenchmark.querySql 22 5000000 none force avgt 5 43.919 ± 2.962 ms/op SqlBenchmark.querySql 22 5000000 fc4 false avgt 5 44.458 ± 3.590 ms/op SqlBenchmark.querySql 22 5000000 fc4 force avgt 5 43.738 ± 2.317 ms/op SqlBenchmark.querySql 22 5000000 fc16 false avgt 5 44.304 ± 2.876 ms/op SqlBenchmark.querySql 22 5000000 fc16 force avgt 5 43.906 ± 2.283 ms/op SqlBenchmark.querySql 23 5000000 none false avgt 5 18260.991 ± 1041.530 ms/op SqlBenchmark.querySql 23 5000000 none force avgt 5 18400.261 ± 1689.495 ms/op SqlBenchmark.querySql 23 5000000 fc4 false avgt 5 30704.667 ± 1624.680 ms/op SqlBenchmark.querySql 23 5000000 fc4 force avgt 5 30603.476 ± 621.976 ms/op SqlBenchmark.querySql 23 5000000 fc16 false avgt 5 48777.535 ± 3223.712 ms/op SqlBenchmark.querySql 23 5000000 fc16 force avgt 5 49966.818 ± 3739.684 ms/op ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
![https://user-images.githubusercontent.com/1577461/191848586-d0211d6a-0510-465d-8909-a16bfbbeb7c7.png"](https://user-images.githubusercontent.com/1577461/191848586-d0211d6a-0510-465d-8909-a16bfbbeb7c7.png"){kind=link}
![https://user-images.githubusercontent.com/1577461/191848610-96e8e431-2eb6-4dfc-8b75-7989fe23318b.png"](https://user-images.githubusercontent.com/1577461/191848610-96e8e431-2eb6-4dfc-8b75-7989fe23318b.png"){kind=link}
![https://user-images.githubusercontent.com/1577461/191848633-f47f8b8a-13f1-4d13-86af-de927577431c.png"](https://user-images.githubusercontent.com/1577461/191848633-f47f8b8a-13f1-4d13-86af-de927577431c.png"){kind=link}