| Summary: | sqlite: upstream performance problem | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] OE-Core | Reporter: | Patrick Ohly <patrick.ohly> | ||||
| Component: | core | Assignee: | Maxin B. John <maxin.john> | ||||
| Status: | RESOLVED FIXED | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | Medium+ | CC: | maxin.john, meta.mr.watcher, meta.watcher, ross.burton, sgw | ||||
| Version: | 2.3 | ||||||
| Target Milestone: | 2.3 M2 | ||||||
| Hardware: | x86 | ||||||
| OS: | Multiple | ||||||
| Whiteboard: | |||||||
| OS type for building Yocto: | --- | Type of Regression: | --- | ||||
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |||||
| Attachments: |
|
||||||
|
Description
Patrick Ohly
2016-11-21 12:53:57 UTC
I have sent an email to sqlite user mailing list, CCing Patrick, pseudo maintainer and Saul. Created attachment 3588 [details]
Correspondence with Sqlite3 team
Refer to the attached correspondence between us and Sqlite3 team. Although the fix is in 3.16.0 but we are suggested to wait for 3.16.2 release which should come out very soon. I think once the Sqlite 3 component in YP is updated to 3.16.2 or any newer versions than it we can re-benchmark without the reverting patch then decide the next to do. Add maxin.john@intel.com the maintainer of sqlite 3 in watch list. Thanks (In reply to comment #3) ... > I think once the Sqlite 3 component in YP is updated to 3.16.2 or any newer > versions than it we can re-benchmark without the reverting patch then decide > the next to do. > > Add maxin.john@intel.com the maintainer of sqlite 3 in watch list. > Patch for sqlite3 version upgrade to 3.16.2 sent to oe-core mailing list: http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/commit/?h=ross/mut&id=41b9dd7ddc0b270638ccfb57429d355b1eea5276 Maxin. Thanks lot! BTW, did you have any chance to verify no performance regression with the new version in the patch? (In reply to comment #5) > Maxin. > Thanks lot! > BTW, did you have any chance to verify no performance regression with the > new version in the patch? # /usr/bin/time -v bitbake core-image-minimal Without the upgrade: Summary: There were 2 WARNING messages shown. Command being timed: "bitbake core-image-minimal" User time (seconds): 11889.38 System time (seconds): 3087.18 Percent of CPU this job got: 300% Elapsed (wall clock) time (h:mm:ss or m:ss): 1:23:08 Average shared text size (kbytes): 0 Average unshared data size (kbytes): 0 Average stack size (kbytes): 0 Average total size (kbytes): 0 Maximum resident set size (kbytes): 802816 Average resident set size (kbytes): 0 Major (requiring I/O) page faults: 7888 Minor (reclaiming a frame) page faults: 721841293 Voluntary context switches: 19513415 Involuntary context switches: 5332647 Swaps: 0 File system inputs: 5243800 File system outputs: 54152904 Socket messages sent: 0 Socket messages received: 0 Signals delivered: 0 Page size (bytes): 4096 Exit status: 0 After upgrade: Summary: There were 2 WARNING messages shown. Command being timed: "bitbake core-image-minimal" User time (seconds): 12026.36 System time (seconds): 3064.97 Percent of CPU this job got: 341% Elapsed (wall clock) time (h:mm:ss or m:ss): 1:13:40 Average shared text size (kbytes): 0 Average unshared data size (kbytes): 0 Average stack size (kbytes): 0 Average total size (kbytes): 0 Maximum resident set size (kbytes): 802860 Average resident set size (kbytes): 0 Major (requiring I/O) page faults: 7551 Minor (reclaiming a frame) page faults: 721237606 Voluntary context switches: 17455269 Involuntary context switches: 7620493 Swaps: 0 File system inputs: 5461224 File system outputs: 53941600 Socket messages sent: 0 Socket messages received: 0 Signals delivered: 0 Page size (bytes): 4096 Exit status: 0 In addition to this, Ross mentioned that this upgrade will be tested on performance cluster before pushing to master. Can this patch be tested also with the Ostro failing issue, I know it's not being supported, but it did show the worst performance regression. Upstream fixed this in 3.16.2, merged in oe-core 05317fe9f11565d40b84ad71300b39c990a53f6d. |