You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@doris.apache.org by "kevinw66 (via GitHub)" <gi...@apache.org> on 2025/10/30 07:52:18 UTC

[PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

kevinw66 opened a new pull request, #57507:
URL: https://github.com/apache/doris/pull/57507

   ### What problem does this PR solve?
   
   Issue Number: close #xxx
   
   Related PR: #xxx
   
   Problem Summary:
   
   ### Release note
   
   None
   
   ### Check List (For Author)
   
   - Test <!-- At least one of them must be included. -->
       - [ ] Regression test
       - [ ] Unit Test
       - [ ] Manual test (add detailed scripts or steps below)
       - [ ] No need to test or manual test. Explain why:
           - [ ] This is a refactor/code format and no logic has been changed.
           - [ ] Previous test can cover this change.
           - [ ] No code files have been changed.
           - [ ] Other reason <!-- Add your reason?  -->
   
   - Behavior changed:
       - [ ] No.
       - [ ] Yes. <!-- Explain the behavior change -->
   
   - Does this need documentation?
       - [ ] No.
       - [ ] Yes. <!-- Add document PR link here. eg: https://github.com/apache/doris-website/pull/1214 -->
   
   ### Check List (For Reviewer who merge this PR)
   
   - [ ] Confirm the release note
   - [ ] Confirm test cases
   - [ ] Confirm document
   - [ ] Add branch pick label <!-- Add branch pick label that this PR should merge into -->
   
   


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "kevinw66 (via GitHub)" <gi...@apache.org>.
kevinw66 commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3475329024

   run buildall


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "doris-robot (via GitHub)" <gi...@apache.org>.
doris-robot commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3475388934

   
   <details>
   <summary>TPC-DS: <b>Total hot run time: 188384 ms</b></summary>
   
   ```
   machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
   scripts: https://github.com/apache/doris/tree/master/tools/tpcds-tools
   TPC-DS sf100 test result on commit fe65edd01f1eb2abdb18f3e9b1e5150b6480ca95, data reload: false
   
   query1	1062	406	393	393
   query2	6564	1695	1675	1675
   query3	6761	225	220	220
   query4	26767	23824	23675	23675
   query5	5622	636	475	475
   query6	350	236	216	216
   query7	4649	498	299	299
   query8	355	262	255	255
   query9	8703	2594	2600	2594
   query10	565	354	306	306
   query11	15449	15224	15003	15003
   query12	198	124	116	116
   query13	1696	577	450	450
   query14	11541	9467	9426	9426
   query15	247	197	183	183
   query16	7746	675	537	537
   query17	1632	813	631	631
   query18	2043	435	334	334
   query19	272	209	180	180
   query20	141	123	127	123
   query21	216	137	117	117
   query22	3991	4071	4075	4071
   query23	33984	32976	32883	32883
   query24	8333	2437	2455	2437
   query25	566	519	422	422
   query26	1232	274	154	154
   query27	2715	491	358	358
   query28	4374	2200	2184	2184
   query29	776	610	471	471
   query30	294	227	199	199
   query31	898	796	747	747
   query32	85	69	73	69
   query33	595	388	320	320
   query34	798	850	523	523
   query35	835	891	750	750
   query36	974	999	924	924
   query37	125	120	97	97
   query38	3598	3463	3468	3463
   query39	1494	1428	1421	1421
   query40	228	129	121	121
   query41	65	60	60	60
   query42	134	108	111	108
   query43	501	495	463	463
   query44	1231	761	742	742
   query45	190	192	179	179
   query46	882	987	664	664
   query47	1812	1795	1691	1691
   query48	397	413	312	312
   query49	794	541	436	436
   query50	637	693	404	404
   query51	3886	3987	3845	3845
   query52	112	112	104	104
   query53	249	269	198	198
   query54	310	308	273	273
   query55	90	84	82	82
   query56	324	318	318	318
   query57	1186	1180	1127	1127
   query58	290	272	280	272
   query59	2573	2628	2603	2603
   query60	360	348	332	332
   query61	160	158	155	155
   query62	783	734	679	679
   query63	228	199	199	199
   query64	4521	1161	858	858
   query65	4032	3949	3974	3949
   query66	1059	452	358	358
   query67	15396	15012	14842	14842
   query68	6579	916	604	604
   query69	509	333	304	304
   query70	1432	1249	1332	1249
   query71	437	339	314	314
   query72	5788	4825	4752	4752
   query73	647	559	362	362
   query74	9186	8871	8997	8871
   query75	3319	3345	2856	2856
   query76	3170	1153	719	719
   query77	498	408	324	324
   query78	9389	10277	8879	8879
   query79	1904	876	594	594
   query80	1538	579	499	499
   query81	572	265	228	228
   query82	364	159	133	133
   query83	277	261	251	251
   query84	259	117	99	99
   query85	925	482	436	436
   query86	448	309	313	309
   query87	3737	3796	3606	3606
   query88	2853	2269	2263	2263
   query89	386	334	301	301
   query90	1872	216	214	214
   query91	167	166	137	137
   query92	78	68	67	67
   query93	1195	956	648	648
   query94	739	445	333	333
   query95	392	320	315	315
   query96	474	570	275	275
   query97	2887	2982	2859	2859
   query98	242	212	209	209
   query99	1297	1409	1299	1299
   Total cold run time: 274236 ms
   Total hot run time: 188384 ms
   ```
   </details>
   


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "kevinw66 (via GitHub)" <gi...@apache.org>.
kevinw66 commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3466519012

   ## Improvements
   
   ### BM_Vstring_RTrim
   | Size | Origin (ns) | SVE (ns) | Improvement |
   |------|-------------|----------|-------------|
   | 16   | 8.06        | 7.94     | **+1.5%**   |
   | 32   | 17.4        | 1.56     | **+91.0%**  |
   | 64   | 37.1        | 2.08     | **+94.4%**  |
   | 256  | 167         | 5.21     | **+96.9%**  |
   | 1024 | 646         | 21.1     | **+96.7%**  |
   
   ### BM_Vstring_LTrim
   | Size | Origin (ns) | SVE (ns) | Improvement |
   |------|-------------|----------|-------------|
   | 16   | 7.48        | 7.48     | 0.0%        |
   | 32   | 14.3        | 1.04     | **+92.7%**  |
   | 64   | 27.8        | 1.49     | **+94.6%**  |
   | 256  | 114         | 4.09     | **+96.4%**  |
   | 1024 | 475         | 15.1     | **+96.8%**  |
   
   ### BM_Vstring_IsAscii
   | Size | Origin (ns) | SVE (ns) | Improvement |
   |------|-------------|----------|-------------|
   | 16   | 2.02        | 0.842    | **+58.3%**  |
   | 32   | 2.77        | 0.841    | **+69.6%**  |
   | 64   | 4.33        | 1.54     | **+64.4%**  |
   | 256  | 14.0        | 5.23     | **+62.6%**  |
   | 1024 | 52.8        | 20.6     | **+61.0%**  |
   
   **Summary:** SVE optimization provides dramatic performance improvements across all tested `vstring` functions. `LTrim` and `RTrim` show exceptional gains of over **90%** for string sizes of 32 bytes and larger. `IsAscii` also demonstrates significant and consistent improvements, with performance boosts of **58-70%**. The benefits of SVE are most pronounced on string sizes that are multiples of the vector length.


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "doris-robot (via GitHub)" <gi...@apache.org>.
doris-robot commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3475397078

   
   <details>
   <summary>ClickBench: <b>Total hot run time: 27.57 s</b></summary>
   
   ```
   machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
   scripts: https://github.com/apache/doris/tree/master/tools/clickbench-tools
   ClickBench test result on commit fe65edd01f1eb2abdb18f3e9b1e5150b6480ca95, data reload: false
   
   query1	0.05	0.05	0.05
   query2	0.09	0.05	0.05
   query3	0.25	0.08	0.08
   query4	1.61	0.12	0.11
   query5	0.28	0.26	0.25
   query6	1.15	0.66	0.64
   query7	0.04	0.03	0.03
   query8	0.05	0.04	0.04
   query9	0.62	0.53	0.52
   query10	0.57	0.58	0.57
   query11	0.16	0.11	0.12
   query12	0.16	0.12	0.12
   query13	0.61	0.60	0.60
   query14	1.00	0.99	1.00
   query15	0.85	0.84	0.83
   query16	0.41	0.40	0.41
   query17	1.05	1.08	1.02
   query18	0.21	0.20	0.20
   query19	1.86	1.80	1.75
   query20	0.02	0.01	0.02
   query21	15.46	0.20	0.13
   query22	5.00	0.07	0.04
   query23	15.68	0.26	0.11
   query24	3.42	0.58	0.44
   query25	0.07	0.07	0.06
   query26	0.13	0.13	0.12
   query27	0.07	0.06	0.05
   query28	5.28	1.13	0.92
   query29	12.56	3.94	3.28
   query30	0.28	0.14	0.12
   query31	2.81	0.59	0.38
   query32	3.22	0.54	0.47
   query33	2.99	3.09	3.05
   query34	15.61	5.15	4.60
   query35	4.58	4.59	4.62
   query36	0.67	0.50	0.50
   query37	0.10	0.06	0.06
   query38	0.06	0.04	0.04
   query39	0.04	0.03	0.02
   query40	0.17	0.14	0.14
   query41	0.08	0.03	0.02
   query42	0.04	0.04	0.03
   query43	0.04	0.04	0.03
   Total cold run time: 99.4 s
   Total hot run time: 27.57 s
   ```
   </details>
   


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3510453165

   PR approved by anyone and no changes requested.


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3510452996

   PR approved by at least one committer and no changes requested.


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "zclllyybb (via GitHub)" <gi...@apache.org>.
zclllyybb merged PR #57507:
URL: https://github.com/apache/doris/pull/57507


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "Thearas (via GitHub)" <gi...@apache.org>.
Thearas commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3466512281

   
   Thank you for your contribution to Apache Doris.
   Don't know what should be done next? See [How to process your PR](https://cwiki.apache.org/confluence/display/DORIS/How+to+process+your+PR).
   
   Please clearly describe your PR:
   1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
   2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
   3. What features were added. Why was this function added?
   4. Which code was refactored and why was this part of the code refactored?
   5. Which functions were optimized and what is the difference before and after the optimization?
       


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "kevinw66 (via GitHub)" <gi...@apache.org>.
kevinw66 commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3466770352

   run buildall


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "kevinw66 (via GitHub)" <gi...@apache.org>.
kevinw66 commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3466684632

   run buildall


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "hello-stephen (via GitHub)" <gi...@apache.org>.
hello-stephen commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3475718730

   # BE Regression && UT Coverage Report
   Increment line coverage `100% (0/0)` :tada:
   
   [Increment coverage report](http://coverage.selectdb-in.cc/coverage/57507_fe65edd01f1eb2abdb18f3e9b1e5150b6480ca95_merge/increment_report/index.html)
   [Complete coverage report](http://coverage.selectdb-in.cc/coverage/57507_fe65edd01f1eb2abdb18f3e9b1e5150b6480ca95_merge/report/index.html)
   | Category          | Coverage           |
   |-------------------|--------------------|
   | Function Coverage | 71.42% (24046/33667) |
   | Line Coverage     | 57.83% (250126/432517) |
   | Region Coverage   | 52.78% (207189/392530) |
   | Branch Coverage   | 54.57% (89098/163274) |


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "doris-robot (via GitHub)" <gi...@apache.org>.
doris-robot commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3475453288

   # BE UT Coverage Report
   Increment line coverage ` ` :tada:
   
   [Increment coverage report](http://coverage.selectdb-in.cc/coverage/fe65edd01f1eb2abdb18f3e9b1e5150b6480ca95_fe65edd01f1eb2abdb18f3e9b1e5150b6480ca95/increment_report/index.html)
   [Complete coverage report](http://coverage.selectdb-in.cc/coverage/fe65edd01f1eb2abdb18f3e9b1e5150b6480ca95_fe65edd01f1eb2abdb18f3e9b1e5150b6480ca95/report/index.html)
   | Category          | Coverage           |
   |-------------------|--------------------|
   | Function Coverage | 52.68% (18045/34252) |
   | Line Coverage     | 37.97% (163957/431822) |
   | Region Coverage   | 32.31% (125013/386966) |
   | Branch Coverage   | 33.71% (54664/162171) |


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "kevinw66 (via GitHub)" <gi...@apache.org>.
kevinw66 commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3466515234

   Benchmark results on Huawei kC2 instance:
   
   Original:
   ```
   Running ./output/be/lib/benchmark_test
   Run on (32 X 200 MHz CPU s)
   CPU Caches:
     L1 Data 64 KiB (x32)
     L1 Instruction 64 KiB (x32)
     L2 Unified 512 KiB (x32)
     L3 Unified 32768 KiB (x1)
   Load Average: 0.77, 0.20, 0.24
   ---------------------------------------------------------------------------------------------------
   Benchmark                                         Time             CPU   Iterations UserCounters...
   ---------------------------------------------------------------------------------------------------
   BM_Vstring_RTrim/16/repeats:5_mean             8.06 ns         8.06 ns            5 bytes_per_second=1.96472G/s
   BM_Vstring_RTrim/16/repeats:5_median           8.06 ns         8.06 ns            5 bytes_per_second=1.96478G/s
   BM_Vstring_RTrim/16/repeats:5_stddev          0.001 ns        0.001 ns            5 bytes_per_second=230.973k/s
   BM_Vstring_RTrim/16/repeats:5_cv               0.01 %          0.01 %             5 bytes_per_second=0.01%
   BM_Vstring_RTrim/32/repeats:5_mean             17.4 ns         17.4 ns            5 bytes_per_second=1.76777G/s
   BM_Vstring_RTrim/32/repeats:5_median           17.4 ns         17.4 ns            5 bytes_per_second=1.76776G/s
   BM_Vstring_RTrim/32/repeats:5_stddev          0.001 ns        0.001 ns            5 bytes_per_second=116.729k/s
   BM_Vstring_RTrim/32/repeats:5_cv               0.01 %          0.01 %             5 bytes_per_second=0.01%
   BM_Vstring_RTrim/64/repeats:5_mean             37.1 ns         37.1 ns            5 bytes_per_second=1.63265G/s
   BM_Vstring_RTrim/64/repeats:5_median           37.1 ns         37.1 ns            5 bytes_per_second=1.63274G/s
   BM_Vstring_RTrim/64/repeats:5_stddev          0.012 ns        0.011 ns            5 bytes_per_second=525.265k/s
   BM_Vstring_RTrim/64/repeats:5_cv               0.03 %          0.03 %             5 bytes_per_second=0.03%
   BM_Vstring_RTrim/256/repeats:5_mean             167 ns          167 ns            5 bytes_per_second=1.43452G/s
   BM_Vstring_RTrim/256/repeats:5_median           167 ns          167 ns            5 bytes_per_second=1.43452G/s
   BM_Vstring_RTrim/256/repeats:5_stddev         0.007 ns        0.006 ns            5 bytes_per_second=52.8043k/s
   BM_Vstring_RTrim/256/repeats:5_cv              0.00 %          0.00 %             5 bytes_per_second=0.00%
   BM_Vstring_RTrim/1024/repeats:5_mean            646 ns          646 ns            5 bytes_per_second=1.47886G/s
   BM_Vstring_RTrim/1024/repeats:5_median          646 ns          646 ns            5 bytes_per_second=1.47877G/s
   BM_Vstring_RTrim/1024/repeats:5_stddev        0.078 ns        0.082 ns            5 bytes_per_second=197.965k/s
   BM_Vstring_RTrim/1024/repeats:5_cv             0.01 %          0.01 %             5 bytes_per_second=0.01%
   BM_Vstring_LTrim/16/repeats:5_mean             7.48 ns         7.48 ns            5 bytes_per_second=2.11783G/s
   BM_Vstring_LTrim/16/repeats:5_median           7.48 ns         7.48 ns            5 bytes_per_second=2.11797G/s
   BM_Vstring_LTrim/16/repeats:5_stddev          0.001 ns        0.001 ns            5 bytes_per_second=432.119k/s
   BM_Vstring_LTrim/16/repeats:5_cv               0.02 %          0.02 %             5 bytes_per_second=0.02%
   BM_Vstring_LTrim/32/repeats:5_mean             14.3 ns         14.3 ns            5 bytes_per_second=2.14762G/s
   BM_Vstring_LTrim/32/repeats:5_median           14.3 ns         14.3 ns            5 bytes_per_second=2.14778G/s
   BM_Vstring_LTrim/32/repeats:5_stddev          0.016 ns        0.016 ns            5 bytes_per_second=2.40854M/s
   BM_Vstring_LTrim/32/repeats:5_cv               0.11 %          0.11 %             5 bytes_per_second=0.11%
   BM_Vstring_LTrim/64/repeats:5_mean             27.8 ns         27.8 ns            5 bytes_per_second=2.17405G/s
   BM_Vstring_LTrim/64/repeats:5_median           27.8 ns         27.8 ns            5 bytes_per_second=2.17568G/s
   BM_Vstring_LTrim/64/repeats:5_stddev          0.057 ns        0.057 ns            5 bytes_per_second=4.56898M/s
   BM_Vstring_LTrim/64/repeats:5_cv               0.21 %          0.21 %             5 bytes_per_second=0.21%
   BM_Vstring_LTrim/256/repeats:5_mean             114 ns          114 ns            5 bytes_per_second=2.09789G/s
   BM_Vstring_LTrim/256/repeats:5_median           114 ns          114 ns            5 bytes_per_second=2.09789G/s
   BM_Vstring_LTrim/256/repeats:5_stddev         0.017 ns        0.017 ns            5 bytes_per_second=329.94k/s
   BM_Vstring_LTrim/256/repeats:5_cv              0.02 %          0.01 %             5 bytes_per_second=0.01%
   BM_Vstring_LTrim/1024/repeats:5_mean            475 ns          475 ns            5 bytes_per_second=2.01149G/s
   BM_Vstring_LTrim/1024/repeats:5_median          475 ns          475 ns            5 bytes_per_second=2.01151G/s
   BM_Vstring_LTrim/1024/repeats:5_stddev        0.018 ns        0.021 ns            5 bytes_per_second=92.1246k/s
   BM_Vstring_LTrim/1024/repeats:5_cv             0.00 %          0.00 %             5 bytes_per_second=0.00%
   BM_Vstring_IsAscii/16/repeats:5_mean           2.02 ns         2.02 ns            5 bytes_per_second=7.37832G/s
   BM_Vstring_IsAscii/16/repeats:5_median         2.02 ns         2.02 ns            5 bytes_per_second=7.37847G/s
   BM_Vstring_IsAscii/16/repeats:5_stddev        0.000 ns        0.000 ns            5 bytes_per_second=293.773k/s
   BM_Vstring_IsAscii/16/repeats:5_cv             0.00 %          0.00 %             5 bytes_per_second=0.00%
   BM_Vstring_IsAscii/32/repeats:5_mean           2.77 ns         2.77 ns            5 bytes_per_second=10.7601G/s
   BM_Vstring_IsAscii/32/repeats:5_median         2.77 ns         2.77 ns            5 bytes_per_second=10.7602G/s
   BM_Vstring_IsAscii/32/repeats:5_stddev        0.000 ns        0.000 ns            5 bytes_per_second=479.111k/s
   BM_Vstring_IsAscii/32/repeats:5_cv             0.00 %          0.00 %             5 bytes_per_second=0.00%
   BM_Vstring_IsAscii/64/repeats:5_mean           4.33 ns         4.33 ns            5 bytes_per_second=13.7731G/s
   BM_Vstring_IsAscii/64/repeats:5_median         4.33 ns         4.33 ns            5 bytes_per_second=13.773G/s
   BM_Vstring_IsAscii/64/repeats:5_stddev        0.000 ns        0.000 ns            5 bytes_per_second=254.18k/s
   BM_Vstring_IsAscii/64/repeats:5_cv             0.00 %          0.00 %             5 bytes_per_second=0.00%
   BM_Vstring_IsAscii/256/repeats:5_mean          14.0 ns         14.0 ns            5 bytes_per_second=17.0032G/s
   BM_Vstring_IsAscii/256/repeats:5_median        14.0 ns         14.0 ns            5 bytes_per_second=17.0033G/s
   BM_Vstring_IsAscii/256/repeats:5_stddev       0.000 ns        0.000 ns            5 bytes_per_second=313.07k/s
   BM_Vstring_IsAscii/256/repeats:5_cv            0.00 %          0.00 %             5 bytes_per_second=0.00%
   BM_Vstring_IsAscii/1024/repeats:5_mean         52.8 ns         52.8 ns            5 bytes_per_second=18.0632G/s
   BM_Vstring_IsAscii/1024/repeats:5_median       52.8 ns         52.8 ns            5 bytes_per_second=18.063G/s
   BM_Vstring_IsAscii/1024/repeats:5_stddev      0.001 ns        0.001 ns            5 bytes_per_second=527.829k/s
   BM_Vstring_IsAscii/1024/repeats:5_cv           0.00 %          0.00 %             5 bytes_per_second=0.00%
   ```
   
   SVE:
   ```
   Running ./output/be/lib/benchmark_test
   Run on (32 X 200 MHz CPU s)
   CPU Caches:
     L1 Data 64 KiB (x32)
     L1 Instruction 64 KiB (x32)
     L2 Unified 512 KiB (x32)
     L3 Unified 32768 KiB (x1)
   Load Average: 3.15, 1.88, 2.27
   ---------------------------------------------------------------------------------------------------
   Benchmark                                         Time             CPU   Iterations UserCounters...
   ---------------------------------------------------------------------------------------------------
   BM_Vstring_RTrim/16/repeats:5_mean             7.94 ns         7.94 ns            5 bytes_per_second=1.99411G/s
   BM_Vstring_RTrim/16/repeats:5_median           7.94 ns         7.94 ns            5 bytes_per_second=1.99394G/s
   BM_Vstring_RTrim/16/repeats:5_stddev          0.003 ns        0.002 ns            5 bytes_per_second=637.853k/s
   BM_Vstring_RTrim/16/repeats:5_cv               0.03 %          0.03 %             5 bytes_per_second=0.03%
   BM_Vstring_RTrim/32/repeats:5_mean             1.56 ns         1.56 ns            5 bytes_per_second=19.7217G/s
   BM_Vstring_RTrim/32/repeats:5_median           1.56 ns         1.56 ns            5 bytes_per_second=19.7224G/s
   BM_Vstring_RTrim/32/repeats:5_stddev          0.000 ns        0.000 ns            5 bytes_per_second=2.28825M/s
   BM_Vstring_RTrim/32/repeats:5_cv               0.02 %          0.01 %             5 bytes_per_second=0.01%
   BM_Vstring_RTrim/64/repeats:5_mean             2.08 ns         2.08 ns            5 bytes_per_second=29.1386G/s
   BM_Vstring_RTrim/64/repeats:5_median           2.08 ns         2.08 ns            5 bytes_per_second=29.1388G/s
   BM_Vstring_RTrim/64/repeats:5_stddev          0.000 ns        0.000 ns            5 bytes_per_second=1.83039M/s
   BM_Vstring_RTrim/64/repeats:5_cv               0.01 %          0.01 %             5 bytes_per_second=0.01%
   BM_Vstring_RTrim/256/repeats:5_mean            5.21 ns         5.21 ns            5 bytes_per_second=45.9438G/s
   BM_Vstring_RTrim/256/repeats:5_median          5.19 ns         5.19 ns            5 bytes_per_second=46.0813G/s
   BM_Vstring_RTrim/256/repeats:5_stddev         0.035 ns        0.035 ns            5 bytes_per_second=316.88M/s
   BM_Vstring_RTrim/256/repeats:5_cv              0.68 %          0.68 %             5 bytes_per_second=0.67%
   BM_Vstring_RTrim/1024/repeats:5_mean           21.1 ns         21.1 ns            5 bytes_per_second=46.6703G/s
   BM_Vstring_RTrim/1024/repeats:5_median         18.3 ns         18.3 ns            5 bytes_per_second=52.1607G/s
   BM_Vstring_RTrim/1024/repeats:5_stddev         4.11 ns         4.11 ns            5 bytes_per_second=8.51247G/s
   BM_Vstring_RTrim/1024/repeats:5_cv            19.52 %         19.52 %             5 bytes_per_second=18.24%
   BM_Vstring_LTrim/16/repeats:5_mean             7.48 ns         7.48 ns            5 bytes_per_second=2.11786G/s
   BM_Vstring_LTrim/16/repeats:5_median           7.48 ns         7.48 ns            5 bytes_per_second=2.11782G/s
   BM_Vstring_LTrim/16/repeats:5_stddev          0.001 ns        0.000 ns            5 bytes_per_second=142.824k/s
   BM_Vstring_LTrim/16/repeats:5_cv               0.01 %          0.01 %             5 bytes_per_second=0.01%
   BM_Vstring_LTrim/32/repeats:5_mean             1.04 ns         1.04 ns            5 bytes_per_second=29.6286G/s
   BM_Vstring_LTrim/32/repeats:5_median           1.04 ns         1.04 ns            5 bytes_per_second=29.6735G/s
   BM_Vstring_LTrim/32/repeats:5_stddev          0.003 ns        0.003 ns            5 bytes_per_second=82.7185M/s
   BM_Vstring_LTrim/32/repeats:5_cv               0.27 %          0.27 %             5 bytes_per_second=0.27%
   BM_Vstring_LTrim/64/repeats:5_mean             1.49 ns         1.49 ns            5 bytes_per_second=40.7235G/s
   BM_Vstring_LTrim/64/repeats:5_median           1.49 ns         1.49 ns            5 bytes_per_second=40.7199G/s
   BM_Vstring_LTrim/64/repeats:5_stddev          0.001 ns        0.001 ns            5 bytes_per_second=41.8824M/s
   BM_Vstring_LTrim/64/repeats:5_cv               0.10 %          0.10 %             5 bytes_per_second=0.10%
   BM_Vstring_LTrim/256/repeats:5_mean            4.09 ns         4.09 ns            5 bytes_per_second=58.4778G/s
   BM_Vstring_LTrim/256/repeats:5_median          4.09 ns         4.09 ns            5 bytes_per_second=58.4872G/s
   BM_Vstring_LTrim/256/repeats:5_stddev         0.005 ns        0.005 ns            5 bytes_per_second=77.9197M/s
   BM_Vstring_LTrim/256/repeats:5_cv              0.13 %          0.13 %             5 bytes_per_second=0.13%
   BM_Vstring_LTrim/1024/repeats:5_mean           15.1 ns         15.1 ns            5 bytes_per_second=63.3362G/s
   BM_Vstring_LTrim/1024/repeats:5_median         15.1 ns         15.1 ns            5 bytes_per_second=63.3427G/s
   BM_Vstring_LTrim/1024/repeats:5_stddev        0.010 ns        0.010 ns            5 bytes_per_second=43.5754M/s
   BM_Vstring_LTrim/1024/repeats:5_cv             0.07 %          0.07 %             5 bytes_per_second=0.07%
   BM_Vstring_IsAscii/16/repeats:5_mean          0.842 ns        0.842 ns            5 bytes_per_second=17.7057G/s
   BM_Vstring_IsAscii/16/repeats:5_median        0.842 ns        0.842 ns            5 bytes_per_second=17.7036G/s
   BM_Vstring_IsAscii/16/repeats:5_stddev        0.000 ns        0.000 ns            5 bytes_per_second=4.45343M/s
   BM_Vstring_IsAscii/16/repeats:5_cv             0.02 %          0.02 %             5 bytes_per_second=0.02%
   BM_Vstring_IsAscii/32/repeats:5_mean          0.841 ns        0.841 ns            5 bytes_per_second=35.4248G/s
   BM_Vstring_IsAscii/32/repeats:5_median        0.841 ns        0.841 ns            5 bytes_per_second=35.427G/s
   BM_Vstring_IsAscii/32/repeats:5_stddev        0.000 ns        0.000 ns            5 bytes_per_second=8.12242M/s
   BM_Vstring_IsAscii/32/repeats:5_cv             0.02 %          0.02 %             5 bytes_per_second=0.02%
   BM_Vstring_IsAscii/64/repeats:5_mean           1.54 ns         1.54 ns            5 bytes_per_second=38.9129G/s
   BM_Vstring_IsAscii/64/repeats:5_median         1.45 ns         1.45 ns            5 bytes_per_second=41.0031G/s
   BM_Vstring_IsAscii/64/repeats:5_stddev        0.117 ns        0.117 ns            5 bytes_per_second=2.86714G/s
   BM_Vstring_IsAscii/64/repeats:5_cv             7.57 %          7.57 %             5 bytes_per_second=7.37%
   BM_Vstring_IsAscii/256/repeats:5_mean          5.23 ns         5.23 ns            5 bytes_per_second=45.5621G/s
   BM_Vstring_IsAscii/256/repeats:5_median        5.23 ns         5.23 ns            5 bytes_per_second=45.5566G/s
   BM_Vstring_IsAscii/256/repeats:5_stddev       0.001 ns        0.001 ns            5 bytes_per_second=12.6037M/s
   BM_Vstring_IsAscii/256/repeats:5_cv            0.03 %          0.03 %             5 bytes_per_second=0.03%
   BM_Vstring_IsAscii/1024/repeats:5_mean         20.6 ns         20.6 ns            5 bytes_per_second=46.394G/s
   BM_Vstring_IsAscii/1024/repeats:5_median       20.6 ns         20.6 ns            5 bytes_per_second=46.3891G/s
   BM_Vstring_IsAscii/1024/repeats:5_stddev      0.003 ns        0.003 ns            5 bytes_per_second=7.73003M/s
   BM_Vstring_IsAscii/1024/repeats:5_cv           0.02 %          0.02 %             5 bytes_per_second=0.02%
   ```


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org


Re: [PR] [opt](arm) Improve performance of vstring functions with SVE intrinsics [doris]

Posted by "doris-robot (via GitHub)" <gi...@apache.org>.
doris-robot commented on PR #57507:
URL: https://github.com/apache/doris/pull/57507#issuecomment-3467105220

   
   <details>
   <summary>ClickBench: <b>Total hot run time: 28.19 s</b></summary>
   
   ```
   machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
   scripts: https://github.com/apache/doris/tree/master/tools/clickbench-tools
   ClickBench test result on commit a01d3a204556137b951e4a14d47301d74a1bc114, data reload: false
   
   query1	0.06	0.06	0.05
   query2	0.10	0.05	0.06
   query3	0.25	0.08	0.09
   query4	1.61	0.13	0.12
   query5	0.28	0.26	0.25
   query6	1.18	0.67	0.64
   query7	0.03	0.03	0.03
   query8	0.06	0.05	0.05
   query9	0.62	0.54	0.53
   query10	0.60	0.59	0.59
   query11	0.17	0.12	0.12
   query12	0.18	0.12	0.13
   query13	0.62	0.61	0.61
   query14	1.01	1.02	1.02
   query15	0.85	0.84	0.86
   query16	0.39	0.41	0.40
   query17	1.04	1.03	1.04
   query18	0.22	0.20	0.20
   query19	1.91	1.87	1.90
   query20	0.02	0.02	0.01
   query21	15.45	0.19	0.13
   query22	5.03	0.08	0.05
   query23	15.64	0.27	0.11
   query24	2.92	0.78	0.67
   query25	0.08	0.08	0.07
   query26	0.14	0.13	0.13
   query27	0.07	0.05	0.06
   query28	5.49	1.15	0.93
   query29	12.58	4.13	3.36
   query30	0.28	0.16	0.15
   query31	2.83	0.59	0.38
   query32	3.23	0.56	0.48
   query33	2.98	3.14	3.04
   query34	15.87	5.17	4.60
   query35	4.56	4.58	4.62
   query36	0.69	0.52	0.49
   query37	0.09	0.06	0.07
   query38	0.08	0.05	0.04
   query39	0.04	0.02	0.03
   query40	0.18	0.15	0.14
   query41	0.08	0.04	0.03
   query42	0.04	0.03	0.03
   query43	0.04	0.04	0.04
   Total cold run time: 99.59 s
   Total hot run time: 28.19 s
   ```
   </details>
   


-- 
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: commits-unsubscribe@doris.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org