Hi all!
After some more cleanups and optimisations work, the MS 32,000 benchmark
first went below 7.15s and after that below 7.1 seconds (on my Core i3
system). This is also significant because it solves 4,500 deals per second
at 7.1111111... (7 and 1 ninth) seconds and we went past that. Here is the
log:
Regards,
-- Shlomi
export FCS_PGO_THEME="-l obf"
ARGS="--worker-step 25 " bash scripts/time-threads-num.bash 4 4 x Several
Times
multi-threaded solver
commit 017c8e18a8d191ffbf20ddf86ee1edb1b2fd5861
(commit Date: Mon May 1 17:11:20 2017 +0300 )
Without (!) Profile Guided Optimisation (PGO).
../source/Tatzer -l ci7b --prefix="/home/shlomif/apps/fcs" -l extra_speed
(With -flto and -fwhole-program).
Without Makefile.gnu
sudo_renice.
-O3
gcc-5.4.0-3.mga6 (Mageia package)
lib64tcmalloc-devel-2.5-3.mga6
With strip.
7.09254908561707s
Highlight: optimizing the positions_by_rank lookup
<DUMP>
Started at 1493644574.814782
Reached Board No. 4000 at 1493644575.736699 (total_num_iters=359513)
Reached Board No. 8000 at 1493644576.612392 (total_num_iters=1056357)
Reached Board No. 12000 at 1493644577.461733 (total_num_iters=3732647)
Unsolved Board No. 11982 at 1493644577.818017
Reached Board No. 16000 at 1493644578.432179 (total_num_iters=4438283)
Reached Board No. 20000 at 1493644579.325626 (total_num_iters=5166294)
Reached Board No. 24000 at 1493644580.172304 (total_num_iters=7849572)
Reached Board No. 28000 at 1493644581.016784 (total_num_iters=8224362)
Reached Board No. 32000 at 1493644581.893775 (total_num_iters=9568033)
Finished at 1493644581.907331 (total_num_iters=11017439)
</DUMP>
( DUMPS-1493644573-t/dump004 )
--
Shlomi Fish http://www.shlomifish.org/
You can never truly appreciate The Gilmore Girls until you've watched it in
the original Klingon.
Please reply to list if it's a mailing list post - http://shlom.in/reply .
Received on Tue May 02 2017 - 16:02:56 IDT