Skip to content

[SPARK-25476][SPARK-25510][TEST] Refactor AggregateBenchmark and add a new trait to better support Dataset and DataFrame API #22484

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 12 commits into from
Closed

[SPARK-25476][SPARK-25510][TEST] Refactor AggregateBenchmark and add a new trait to better support Dataset and DataFrame API #22484

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
649f296
Refactor AggregateBenchmark
wangyum Sep 20, 2018
e0c6d09
Merge remote-tracking branch 'upstream/master' into SPARK-25476
wangyum Sep 21, 2018
42230b6
Add RunBenchmarkWithCodegen
wangyum Sep 22, 2018
2d778a4
Rename RunBenchmarkWithCodegen to SqlBasedBenchmark
wangyum Sep 24, 2018
536d4e9
Add withSQLConf to SqlBasedBenchmark
wangyum Sep 24, 2018
f783099
true.toString -> "true" and false.toString -> "false"
wangyum Sep 25, 2018
d5fecdc
Fix indentation
wangyum Sep 25, 2018
23f1bd8
Merge remote-tracking branch 'upstream/master' into SPARK-25476
wangyum Sep 26, 2018
0fae54d
use SQLHelper
wangyum Sep 26, 2018
45add92
Update result (#12)
dongjoon-hyun Sep 27, 2018
3439992
Fix string variable
wangyum Sep 27, 2018
6c46ad5
Address comments
wangyum Oct 1, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
143 changes: 143 additions & 0 deletions sql/core/benchmarks/AggregateBenchmark-results.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,143 @@
================================================================================================
aggregate without grouping
================================================================================================

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
agg w/o group: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
agg w/o group wholestage off 65374 / 70665 32.1 31.2 1.0X
agg w/o group wholestage on 1178 / 1209 1779.8 0.6 55.5X


================================================================================================
stat functions
================================================================================================

Copy link
Member Author

@wangyum wangyum Sep 20, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@davies Do you know how to generate there benchmark:

spark/sql/core/src/test/scala/org/apache/spark/sql/execution/benchmark/AggregateBenchmark.scala

Lines 70 to 78 in 3c3eebc

Using ImperativeAggregate (as implemented in Spark 1.6):
Intel(R) Core(TM) i7-4558U CPU @ 2.80GHz
stddev: Avg Time(ms) Avg Rate(M/s) Relative Rate
-------------------------------------------------------------------------------
stddev w/o codegen 2019.04 10.39 1.00 X
stddev w codegen 2097.29 10.00 0.96 X
kurtosis w/o codegen 2108.99 9.94 0.96 X
kurtosis w codegen 2090.69 10.03 0.97 X

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
stddev: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
stddev wholestage off 8667 / 8851 12.1 82.7 1.0X
stddev wholestage on 1266 / 1273 82.8 12.1 6.8X

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
kurtosis: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
kurtosis wholestage off 41218 / 41231 2.5 393.1 1.0X
kurtosis wholestage on 1347 / 1357 77.8 12.8 30.6X


================================================================================================
aggregate with linear keys
================================================================================================

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
Aggregate w keys: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
codegen = F 9309 / 9389 9.0 111.0 1.0X
codegen = T hashmap = F 4417 / 4435 19.0 52.7 2.1X
codegen = T hashmap = T 1289 / 1298 65.1 15.4 7.2X


================================================================================================
aggregate with randomized keys
================================================================================================

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
Aggregate w keys: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
codegen = F 11424 / 11426 7.3 136.2 1.0X
codegen = T hashmap = F 6441 / 6496 13.0 76.8 1.8X
codegen = T hashmap = T 2333 / 2344 36.0 27.8 4.9X


================================================================================================
aggregate with string key
================================================================================================

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
Aggregate w string key: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
codegen = F 4751 / 4890 4.4 226.5 1.0X
codegen = T hashmap = F 3146 / 3182 6.7 150.0 1.5X
codegen = T hashmap = T 2211 / 2261 9.5 105.4 2.1X


================================================================================================
aggregate with decimal key
================================================================================================

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
Aggregate w decimal key: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
codegen = F 3029 / 3062 6.9 144.4 1.0X
codegen = T hashmap = F 1534 / 1569 13.7 73.2 2.0X
codegen = T hashmap = T 575 / 578 36.5 27.4 5.3X


================================================================================================
aggregate with multiple key types
================================================================================================

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
Aggregate w multiple keys: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
codegen = F 7506 / 7521 2.8 357.9 1.0X
codegen = T hashmap = F 4791 / 4808 4.4 228.5 1.6X
codegen = T hashmap = T 3553 / 3585 5.9 169.4 2.1X


================================================================================================
max function bytecode size of wholestagecodegen
================================================================================================

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
max function bytecode size: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
codegen = F 608 / 656 1.1 927.1 1.0X
codegen = T hugeMethodLimit = 10000 402 / 419 1.6 613.5 1.5X
codegen = T hugeMethodLimit = 1500 616 / 619 1.1 939.9 1.0X


================================================================================================
cube
================================================================================================

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
cube: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
cube wholestage off 3229 / 3237 1.6 615.9 1.0X
cube wholestage on 1285 / 1306 4.1 245.2 2.5X


================================================================================================
hash and BytesToBytesMap
================================================================================================

OpenJDK 64-Bit Server VM 1.8.0_181-b13 on Linux 3.10.0-862.3.2.el7.x86_64
Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz
BytesToBytesMap: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
UnsafeRowhash 328 / 330 64.0 15.6 1.0X
murmur3 hash 167 / 167 125.4 8.0 2.0X
fast hash 84 / 85 249.0 4.0 3.9X
arrayEqual 192 / 192 109.3 9.1 1.7X
Java HashMap (Long) 144 / 147 145.9 6.9 2.3X
Java HashMap (two ints) 147 / 153 142.3 7.0 2.2X
Java HashMap (UnsafeRow) 785 / 788 26.7 37.4 0.4X
LongToUnsafeRowMap (opt=false) 456 / 457 46.0 21.8 0.7X
LongToUnsafeRowMap (opt=true) 125 / 125 168.3 5.9 2.6X
BytesToBytesMap (off Heap) 885 / 885 23.7 42.2 0.4X
BytesToBytesMap (on Heap) 860 / 864 24.4 41.0 0.4X
Aggregate HashMap 56 / 56 373.9 2.7 5.8X


Loading