lwllvyb commented on PR #2210:
URL:
https://github.com/apache/incubator-uniffle/pull/2210#issuecomment-2425877066
Per event { num: 1000, blockSize: 16384 } eventNum: 1000 total size:
16384000000
write: 75860 72293 73507 73585 73769 Average: 73802
ms, opt: 1.0
writeBuffer(1MB): 68339 69674 67826 68017 69397 Average: 68650 ms, opt:
0.9301916
writeBuffer(8MB): 73166 71645 73885 70285 72046 Average: 72205 ms, opt:
0.978361
writeBuffer(16MB): 71575 70121 70922 70513 70628 Average: 70751 ms, opt:
0.95865965
Per event { num: 1000, blockSize: 32768 } eventNum: 500 total size:
16384000000
write: 35487 35971 36248 37756 38809 Average: 36854
ms, opt: 1.0
writeBuffer(1MB): 33903 33720 35004 33963 33985 Average: 34115 ms, opt:
0.9256797
writeBuffer(8MB): 36286 36067 36605 34224 34378 Average: 35512 ms, opt:
0.96358603
writeBuffer(16MB): 33795 33789 34116 35614 34370 Average: 34336 ms, opt:
0.9316763
Per event { num: 1000, blockSize: 65536 } eventNum: 250 total size:
16384000000
write: 18000 18853 18096 17837 18037 Average: 18164
ms, opt: 1.0
writeBuffer(1MB): 16610 16549 16420 16409 16759 Average: 16549 ms, opt:
0.9110879
writeBuffer(8MB): 17397 17353 17316 17487 16998 Average: 17310 ms, opt:
0.9529839
writeBuffer(16MB): 17518 16970 17526 16971 17320 Average: 17261 ms, opt:
0.95028627
Per event { num: 1000, blockSize: 262144 } eventNum: 125 total size:
32768000000
write: 9059 9083 9162 8962 9588 Average: 9170 ms, opt:
1.0
writeBuffer(1MB): 8283 8254 8221 8187 8305 Average: 8250 ms, opt:
0.89967287
writeBuffer(8MB): 8475 8376 8358 8303 8363 Average: 8375 ms, opt:
0.91330427
writeBuffer(16MB): 8382 8476 8390 8489 8470 Average: 8441 ms, opt:
0.92050165
Per event { num: 1000, blockSize: 524288 } eventNum: 62 total size:
32505856000
write: 4394 4437 4429 4534 4399 Average: 4438 ms, opt:
1.0
writeBuffer(1MB): 4023 4042 4162 4199 4199 Average: 4125 ms, opt:
0.92947274
writeBuffer(8MB): 4178 4260 4081 4109 4294 Average: 4184 ms, opt: 0.942767
writeBuffer(16MB): 4114 6990 4580 4227 4186 Average: 4819 ms, opt:
1.0858495
Per event { num: 1000, blockSize: 1048576 } eventNum: 31 total size:
32505856000
write: 2237 2235 2250 2346 2386 Average: 2290 ms, opt:
1.0
writeBuffer(1MB): 2082 2007 2034 2098 2019 Average: 2048 ms, opt:
0.89432317
writeBuffer(8MB): 2056 2204 2076 2085 2095 Average: 2103 ms, opt:
0.9183406
writeBuffer(16MB): 2105 2163 2078 2088 2145 Average: 2115 ms, opt:
0.92358077
Per event { num: 1000, blockSize: 2097152 } eventNum: 15 total size:
31457280000
write: 1180 1163 1240 1216 1158 Average: 1191 ms, opt:
1.0
writeBuffer(1MB): 989 958 980 973 997 Average: 979 ms, opt: 0.8219983
writeBuffer(8MB): 992 1004 1013 997 979 Average: 997 ms, opt: 0.83711165
writeBuffer(16MB): 999 988 1017 1022 1026 Average: 1010 ms, opt: 0.8480269
After some benchmarks with different block sizes and different buffer sizes,
the performance results are 5% ~ 18% improvement.
--
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]
