On Wednesday, 6 May 2020 at 07:47:59 UTC, drug wrote:
06.05.2020 10:42, data pulverizer пишет:
On Wednesday, 6 May 2020 at 07:27:19 UTC, data pulverizer
wrote:
On Wednesday, 6 May 2020 at 06:54:07 UTC, drug wrote:
Thing are really interesting. So there is a space to improve
performance in 2.5 times :-)
Yes, `array` is smart enough and if you call it on another
array it is no op.
What means `--fast` in Chapel? Do you try `--fast-math` in
ldc? Don't know if 05 use this flag
I tried `--fast-math` in ldc but it didn't make any
difference the documentation of `--fast` in Chapel says
"Disable checks; optimize/specialize".
Just tried removing the boundscheck and got 1.5 seconds in D!
Congrats! it looks like the thriller!
What about cpu usage? the same 40%?
CPU usage now revs up almost has time to touch 100% before the
process is finished! Interestingly using `--boundscheck=off`
without `--ffast-math` gives a timing of around 4 seconds and,
whereas using `--ffast-math` without `--boundscheck=off` made no
difference, having both gives us the 1.5 seconds. As Jacob
Carlborg suggested I tried adding `-mcpu=native -flto=full
-defaultlib=phobos2-ldc-lto,druntime-ldc-lto` but I didn't see
any difference.
Current Julia time is still around 35 seconds even when using
@inbounds @simd, and running julia -O3 --check-bounds=no but I'll
probably need to run the code by the Julia community to see
whether it can be further optimized but it's pretty interesting
to see D so far in front. Interesting when I attempt to switch
off the garbage collector in Julia, the process gets killed
because my computer runs out of memory (I have over 26 GB of
memory free) whereas in D the memory I'm using barely registers
(max 300MB) - it uses even less than Chapel (max 500MB) - which
doesn't use much at all. It's exactly the same computation, D and
Julia's timing were similar before the GC optimization and
compiler flag magic in D.
