Age | Commit message (Collapse) | Author |
|
591a8fa69de47fdc3876187e0783dc40e32d22e8
Change-Id: Ifbc27a87e5c0a44ce645f4fda885496b5029fb75
|
|
Change-Id: I1c5e4c6b0cc43fbdf2923484b8c643e2a7c9c51e
|
|
The patch adds an option to tcwg-benchmark-results.sh called "metric",
and calls corresponding scripts to convert relevant metric-data to
csv files. And adds metric to rr in tcwg_bmk-build.sh, which is
set to "perf" by default. The rationale behind this change is to
refactor the scripts to simplify adding more metrics in future.
Change-Id: I3dfd2a572faeea4b0034fe6a6bb635824b3d60c3
|
|
It defaults to "yes" so that we parse perf logs by default.
Use "no" when no perf data is available (eg. when benchmarking on
bare-metal).
Change-Id: I0e6c4b4b6e39de5a55808f22154025e435561753
|
|
Change-Id: I49685c1938c44548f88928c7afa455a863af8076
|
|
Now using the version that's
present in the bionic images.
Meaning we have more warnings
but can also set the minimum
level to ignore a lot of them.
Fix the check for whether
shellcheck supports --severity.
(we have pipefail on, but we
expect shellcheck to fail in
this case)
Change-Id: I296b8554591b7d327c75393e3924184abc6512b7
|
|
First round of low hanging fruit
from the files with the fewest
warnings.
Change-Id: I60ad57b5bfb2f0be9bb6506252c820f44d055200
|
|
Ignore symbols in LTO results, and consider only DSOs with 10% and
above contribution to profile.
Change-Id: I14d003baf0c9ba7fbf938657b84048af99eaef73
|
|
Change-Id: I37a5dfa26c4739931dcf138ac3a563f091fb5ae0
|
|
Change-Id: Ie9ea1c24ff103ee12d90b55435d18f6d5ea1fa98
|
|
... by putting results to bash array ${results[@]}.
Change-Id: I8e189f6f14578e8a96a33c0beb2dc8ae13ca2ef2
|
|
Change-Id: If4ae144bedce4c2b5004cdcbe8da171afbb8118a
|
|
... and add option to produce absolute results with "--relative false".
Change-Id: Ice7f735a45ed911655decf203ab09897bd3655c5
|
|
Change-Id: I6dbb63158bfb46fc6e0eab44c17ac1882e54ec4f
|
|
To help debug problems under Jenkins.
Change-Id: I67cda97c0b00c173f4dc5cf32d970a13a11d4628
|
|
Switch from considering top 10 symbols for regressions to only
size of final executable. We will start tracking top 1-2 symbols
after stabilizing tracking of final executables.
This also allows us to switch to quick "verify" benchmarking mode.
Change-Id: If8f1dd14c9ab861e1b0c1d617790df71d4ba25d1
|
|
This patch updates tcwg_bmk-build.sh from the original copy of
tcwg_kernel-build.sh, and adds a couple minor changes to docker-run.sh
and tcwg-benchmark-results.sh.
tcwg-benchmark*: Switch results server to bkp-01.tcwglab
tcwg_bmk-build.sh: Use ABE's sysroot and binutils for LLVM's build
round-robin.sh: Build LLVMgold.so plugin in build_llvm() for LTO
Change-Id: Ibde7a9d004c8b3328cdd36333c14574934b1eeab
|
|
Change-Id: I3af87f6a085ccdc781a67cf4462ebaf6abf393d7
|
|
Change-Id: I4eb163df3cdcd176d13601c2b3830e4f1884a62b
|
|
Instead of hardcoding "1" as reference.
Change-Id: I57efe5c3203b242f556c09d310d85c1124175fb1
|
|
This patch re-organizes benchmarking results as the following
/home/tcwg-benchmark/
results-tk1/<job_name>-<build_num>/
tcwg-bmk-tk1-06/
<files>
tcwg-bmk-tk1-07/
<files>
tcwg-bmk-tk1-08/
<files>
results-tx1/<job_name>-<build_num>/
tcwg-bmk-tx1-06/
<files>
tcwg-bmk-tx1-07/
<files>
tcwg-bmk-tx1-08/
<files>
This allows us to control access to results per board by
setting permissions on top-level directory. Additionally,
we hard-code host for results as dev-01.tcwglab to disallow
forwarding of results to other hosts.
Parameter "resultsdest" is replaced by "results_id", which is
in form "<hw_type>/<job_name>-<build_num>", e.g.,
"tk1/tcwg-benchmark-christophe-123". Results_id
is appended to dev-01.tcwglab:/home/tcwg-benchmark/results-$results_id
to construct path to the results.
The tcwg-benchmark-results job is being reworked to upload only
relative results to jenkins web UI. The new "--relative" option
to csvs2table.py is not yet implemented, so the job is not yet working.
Change-Id: Ib0ff755ad122714bd9af3ba19cd57672ea081321
|
|
Change-Id: I41cd27697d38d06773a346d525b1f53cc5b77f50
|
|
This script is processing result files from tcwg-benchmark
job, which passes it resultsdest parameter. The results are,
normally, located in dev-01.tcwglab:/home/tcwg-benchmark/results/$BUILD_TAG/.
The output is 2 files: time.csv has time data from "official" SPEC2k6
results, and sample.csv has 10HZ sample data from perf.data files.
Change-Id: I90ac529d7265194f779184d51913adaa16322470
|