By default, coccicheck utilizes all available threads to implement
parallelisation. However, when all available threads are used,
a decrease in performance is noted. The elapsed time is minimum
when at most one thread per core is used.
For example, on benchmarking the semantic patch kfree.cocci for
usb/serial using hyperfine, the outputs obtained for J=5 and J=2
are 1.32 and 1.90 times faster than those for J=10 and J=9
respectively for two separate runs. For the larger drivers/staging
directory, minimium elapsed time is obtained for J=3 which is 1.86
times faster than that for J=12. The optimal J value does not
exceed 6 in any of the test runs. The benchmarks are run on a machine
with 6 cores, with 2 threads per core, i.e, 12 hyperthreads in all.
To improve performance, modify coccicheck to use at most only
one thread per core by default.
Signed-off-by: Sumera Priyadarsini <[email protected]>
---
scripts/coccicheck | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/scripts/coccicheck b/scripts/coccicheck
index e04d328210ac..dd228dcc915e 100755
--- a/scripts/coccicheck
+++ b/scripts/coccicheck
@@ -75,8 +75,17 @@ else
OPTIONS="--dir $KBUILD_EXTMOD $COCCIINCLUDE"
fi
+ # Use only one thread per core by default if hyperthreading is enabled
+ THREADS_PER_CORE=$(lscpu | grep "Thread(s) per core: " | tr -cd [:digit:])
if [ -z "$J" ]; then
NPROC=$(getconf _NPROCESSORS_ONLN)
+ if [ $THREADS_PER_CORE -gt 1 -a $NPROC -gt 2 ] ; then
+ if [ $NPROC -gt 8 ] ; then
+ NPROC=$((NPROC/4))
+ else
+ NPROC=$((NPROC/2))
+ fi
+ fi
else
NPROC="$J"
fi
--
2.17.1