May 2016 - aarch64-port-dev

Request for clarification: hs or hs-comp
by Edward Nevill 25 May '16

25 May '16

Hi, Could someone please clarify which repo we should be pushing aarch64 compiler specific bug fixes to now that we are post FC. Should we continue pushing to hs-comp as before FC, or should we now be pushing to hs? Thanks, Ed.

3 2

RFR: 8157834: aarch64: Hello World crashes with fastdebug build
by Edward Nevill 25 May '16

25 May '16

Hi, Please review the following webrev http://cr.openjdk.java.net/~enevill/8157834/webrev/ jira issue: https://bugs.openjdk.java.net/browse/JDK-8157834 This crash is caused by StrIndexOf being called with argument encoding != UU even though CompactStrings is disabled. Fix is to implement LL, UL and LU encodings for StrIndexOf. All the best, Ed.

2 1

RFR: 8157841: aarch64: prefetch ignores cache line size
by Edward Nevill 25 May '16

25 May '16

Hi, Please review the following webrev http://cr.openjdk.java.net/~enevill/8157841/webrev The aarch64 implementation ignores the actual dcache line size and sets the various prefetch variables based on an assumed cache line size of 64. One of our partners has a cache line size of 128. The above patch reads the cache line size from the ctr_el0 register and sets the prefetch variables based on multiples of that instead. Regards, Ed.

2 1

Thursday afternoon fireside chat (2)
by Edward Nevill 25 May '16

25 May '16

Hi, Our second fireside chat will take place tomorrow (Thursday) afternoon at 15.00 UTC (16:00 Local UK time). This is an open conference call where anyone who is involved in or interested in the aarch64 port of OpenJDK can join in to voice their suggestions, ask questions, air their grievances or just tell us how well we are doing. This chat will be hosted by Stuart Monteith of Linaro as I have a meeting tomorrow which may run into the chat so I might be late. The call will be held using 'BlueJeans' conferencing. Please use the link below. https://bluejeans.com/285836006 If you have not used BlueJeans before please try the above link in advance as it needs to do some downloads first time it starts. Alternatively you may dial in using one of the following numbers http://bluejeans.com/numbers and enter the Meeting ID: 285836006 followed by '#'. Please do not use any of the 'freefone' numbers, because although they may be free for you they cost us $$$$. Talk to you tomorrow, Ed.

1 0

Re: [aarch64-port-dev ] PING: Re: RFR: 8156731: aarch64: java/util/Arrays/Correct.java fails due to _generic_arraycopy stub routine
by Andrew Haley 25 May '16

25 May '16

On 05/25/2016 09:28 AM, Edward Nevill wrote: > Can you please confirm that it is now OK for Felix to push this to hs-comp. No, to hs. OK. Andrew.

1 0

hg: aarch64-port/jdk8u/hotspot: 3 new changesets
by aph＠redhat.com 24 May '16

24 May '16

Changeset: 6569981b0146 Author: aph Date: 2016-04-26 14:38 +0000 URL: http://hg.openjdk.java.net/aarch64-port/jdk8u/hotspot/rev/6569981b0146 8155100: AArch64: Relax alignment requirement for byte_map_base Reviewed-by: roland ! src/cpu/aarch64/vm/macroAssembler_aarch64.cpp Changeset: 204125f031ec Author: aph Date: 2016-04-20 11:14 +0000 URL: http://hg.openjdk.java.net/aarch64-port/jdk8u/hotspot/rev/204125f031ec 8154739: AArch64: TemplateTable::fast_xaccess loads in wrong mode Reviewed-by: roland ! src/cpu/aarch64/vm/templateTable_aarch64.cpp Changeset: a10787e56d77 Author: aph Date: 2016-05-23 15:39 +0000 URL: http://hg.openjdk.java.net/aarch64-port/jdk8u/hotspot/rev/a10787e56d77 8150045: arraycopy causes segfaults in SATB during garbage collection Reviewed-by: roland ! src/cpu/aarch64/vm/stubGenerator_aarch64.cpp

1 0

Prefetching parameters
by Stuart Monteith 24 May '16

24 May '16

Hello, I've been doing some investigation of prefetching to decide upon reasonable parameters for each of the platforms we're testing. The results are posted here: http://people.linaro.org/~stuart.monteith/Prefetch/ For each platform: For Platform A: There is no benefit to allocation prefetching, and for purely allocating arrays, it is harmful: -XX:AllocatePrefetchStyle=-1 For array copy prefetching, 3 cache lines are best: -XX:PrefetchCopyIntervalInBytes=384 For Platform B: There is a large benefit to allocation prefetching smaller arrays if they are not being used straight away. This may benefit some workloads. Otherwise but the default should probably be: -XX:AllocatePrefetchStyle=-1 For array copy prefetching, it's best to not do prefetching. For Platform C: For all, except larger arrays, there aren't benefits for allocation prefetching. For arrays that are zero'd but not used, it is best not to turn on prefetching. -XX:AllocatePrefetchStyle=-1 There is no clear benefit to array copy prefetching. For all platforms: -XX:PrefetchScanIntervalInBytes Can be left as is - there are no clear benefits to enabling this. Best regards, Stuart

1 0

Backports to aarch64/jdk8u
by Andrew Haley 24 May '16

24 May '16

comparing with http://hg.openjdk.java.net/aarch64-port/jdk8u/hotspot searching for changes changeset: 8868:25aa7e3f8333 user: aph date: Tue Apr 26 14:38:39 2016 +0000 summary: 8155100: AArch64: Relax alignment requirement for byte_map_base changeset: 8869:7f4d5a501081 tag: tip user: aph date: Wed Apr 20 11:14:10 2016 +0000 summary: 8154739: AArch64: TemplateTable::fast_xaccess loads in wrong mode Andrew.

2 2

JTREG, SPECjbb2013 and Hadoop/Terasort results for OpenJDK 8 on AArch64
by Edward Nevill 24 May '16

24 May '16

Hi All, We have switched the overnight OpenJDK tests to run from the aarch64 jdk8u tree, rather than the aarch64 jdk8 tree, as the jdk8 tree is now frozen. The tests are done nightly is there has been a commit and take ~20 hours to run so the results should be posted later each day. We are also looking at setting up a CI loop for jdk9. All the best, Ed. This is a summary of the JTREG test results =========================================== The build and test results are cycled every 10 days. For detailed information on the test output please refer to: http://openjdk.linaro.org/openjdk8-jtreg-nightly-tests/summary/2016/142/sum… ------------------------------------------------------------------------------- server-release/hotspot ------------------------------------------------------------------------------- Build 0: aarch64/2015/oct/27 pass: 644; fail: 38; error: 5 Build 1: aarch64/2015/oct/29 pass: 645; fail: 37; error: 5 Build 2: aarch64/2015/nov/04 pass: 643; fail: 39; error: 5 Build 3: aarch64/2015/nov/27 pass: 643; fail: 38; error: 6 Build 4: aarch64/2016/jan/07 pass: 644; fail: 38; error: 5 Build 5: aarch64/2016/jan/23 pass: 643; fail: 38; error: 7 Build 6: aarch64/2016/jan/27 pass: 643; fail: 37; error: 8 Build 7: aarch64/2016/may/11 pass: 649; fail: 41; error: 12 Build 8: aarch64/2016/may/18 pass: 642; fail: 49; error: 11 Build 9: aarch64/2016/may/21 pass: 647; fail: 41; error: 14 ------------------------------------------------------------------------------- server-release/jdk ------------------------------------------------------------------------------- Build 0: aarch64/2015/oct/27 pass: 5,495; fail: 234; error: 24 Build 1: aarch64/2015/oct/29 pass: 5,521; fail: 209; error: 23 Build 2: aarch64/2015/nov/04 pass: 5,509; fail: 222; error: 22 Build 3: aarch64/2015/nov/27 pass: 5,517; fail: 214; error: 22 Build 4: aarch64/2016/jan/07 pass: 5,521; fail: 208; error: 24 Build 5: aarch64/2016/jan/23 pass: 5,508; fail: 224; error: 21 Build 6: aarch64/2016/jan/27 pass: 5,507; fail: 213; error: 33 Build 7: aarch64/2016/may/11 pass: 5,515; fail: 221; error: 36 Build 8: aarch64/2016/may/18 pass: 3,750; fail: 289; error: 1,733 Build 9: aarch64/2016/may/21 pass: 5,515; fail: 223; error: 34 ------------------------------------------------------------------------------- server-release/langtools ------------------------------------------------------------------------------- Build 0: aarch64/2015/oct/27 pass: 3,083; error: 8 Build 1: aarch64/2015/oct/29 pass: 3,081; error: 10 Build 2: aarch64/2015/nov/04 pass: 3,080; fail: 1; error: 10 Build 3: aarch64/2015/nov/27 pass: 3,079; error: 12 Build 4: aarch64/2016/jan/07 pass: 3,079; error: 12 Build 5: aarch64/2016/jan/23 pass: 3,079; error: 12 Build 6: aarch64/2016/jan/27 pass: 3,079; fail: 2; error: 10 Build 7: aarch64/2016/may/11 pass: 3,084; error: 10 Build 8: aarch64/2016/may/18 pass: 3,065; fail: 16; error: 13 Build 9: aarch64/2016/may/21 pass: 3,083; error: 11 Previous results can be found here: http://openjdk.linaro.org/openjdk8-jtreg-nightly-tests/index.html SPECjbb2013 composite regression test completed =============================================== This test measures the relative performance of the server compiler running the SPECjbb2013 composite tests and compares the performance against the baseline performance of the server compiler taken on 2014-04-01. In accordance with [1], the SPECjbb2013 tests are run on a system which is not production ready and does not meet all the requirements for publishing compliant results. The numbers below shall be treated as non-compliant (nc) and are for experimental purposes only. Relative performance: Server max-jOPS (nc): 1.21x Relative performance: Server critical-jOPS (nc): 1.44x Details of the test setup and historical results may be found here: http://openjdk.linaro.org/SPECjbb2013-1.00-results/ [1] http://www.spec.org/fairuse.html#Academic Regression test Hadoop-Terasort completed ========================================= This test measures the performance of the server and client compilers running Hadoop sorting a 1GB file using Terasort and compares the performance against the baseline performance of the Zero interpreter and against the baseline performance of the client and server compilers on 2014-04-01. Relative performance: Zero: 1.0, Client: 48.84, Server: 90.48 Client 48.84 / Client 2014-04-01 (43.00): 1.14x Server 90.48 / Server 2014-04-01 (71.00): 1.27x Details of the test setup and historical results may be found here: http://openjdk.linaro.org/hadoop-terasort-benchmark-results/ This is a summary of the jcstress test results ============================================== The build and test results are cycled every 10 days. 2015-10-22 pass rate: 11558/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/295/results/ 2015-10-27 pass rate: 11558/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/300/results/ 2015-10-29 pass rate: 11558/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/302/results/ 2015-11-04 pass rate: 11558/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/308/results/ 2015-11-27 pass rate: 11558/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/331/results/ 2016-01-07 pass rate: 11558/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2016/007/results/ 2016-01-23 pass rate: 11558/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2016/023/results/ 2016-01-27 pass rate: 11558/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2016/027/results/ 2016-05-18 pass rate: 11557/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2016/139/results/ 2016-05-21 pass rate: 11558/11558, results: http://openjdk.linaro.org/jcstress-nightly-runs/2016/142/results/ For detailed information on the test output please refer to: http://openjdk.linaro.org/jcstress-nightly-runs/

1 0

RFR: 8156943: aarch64: string compare does not support CompactStrings
by Edward Nevill 20 May '16

20 May '16

Hi, Please review the following webrev http://cr.openjdk.java.net/~enevill/8156943/webrev.01/ This adds support for LL, LU and UL encodings of StrComp Tested with fastdebug build on jtreg hotspot & langtools. Regards, Ed.

1 0