Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2014-01-28

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
02:16 colomon joined #moarvm
03:44 cognominal joined #moarvm
04:32 jnap joined #moarvm
04:43 krunen joined #moarvm
07:11 FROGGS joined #moarvm
08:20 sorear joined #moarvm
08:21 woolfy joined #moarvm
08:57 odc joined #moarvm
10:01 FROGGS joined #moarvm
11:58 krunen joined #moarvm
12:00 FROGGS joined #moarvm
14:26 jnap joined #moarvm
15:21 frankbutt joined #moarvm
15:21 frankbutt left #moarvm
15:51 krunen joined #moarvm
16:50 FROGGS joined #moarvm
17:24 FROGGS joined #moarvm
18:13 tgt joined #moarvm
18:16 benabik joined #moarvm
18:41 jnap joined #moarvm
19:30 diakopter nwc10: patch looks good to me
19:32 FROGGS nwc++
19:32 FROGGS nwc10++
19:46 timotimo so who has measured the speedup?
19:48 nwc10 me, just once and once, on a multi user machine, for the setting
19:49 nwc10 specifically, not the spectests
19:49 nwc10 and not startup time
19:50 timotimo right.
19:50 timotimo so nobody did it properly
19:50 timotimo let me do it, then :)
19:53 * timotimo builds all the backends for nqp
19:55 tgt joined #moarvm
20:02 timotimo first timing with your patch: Stage parse      :  44.563 Stage mast       :  20.349
20:03 nwc10 I'm assuming that the setting compilation has an awful lot of gen2. Atypically high
20:03 nwc10 so will be the best-case win
20:03 timotimo maybe.
20:03 timotimo i'll run spectests afterwards
20:04 timotimo Stage parse      :  44.670 Stage mast       :  20.307
20:05 timotimo one more, then i'll undo the patch and run those again
20:05 timotimo actually, i should probably directly run a spectest
20:05 timotimo since that won't need any extra recompilation
20:06 timotimo Stage parse      :  44.742 Stage mast       :  20.439
20:08 timotimo unfortunately the spectests still take a while :P
20:09 timotimo ah, i already reached S32
20:09 timotimo there's kind of a double hump there :P
20:10 timotimo TEST_JOBS=4 make m-spectest  771,24s user 30,92s system 373% cpu 3:34,72 total
20:10 timotimo unfortunately i didn't get the maxresidentk for some reason :o
20:10 nwc10 I doubt that that will change
20:10 timotimo Files=777, Tests=29654, 205 wallclock secs ( 4.97 usr  0.78 sys + 762.92 cusr 28.87 csys = 797.54 CPU)
20:11 timotimo ISTR that was quite a bit more the last time i tried
20:11 timotimo doing a second run nw.
20:11 timotimo and then i'll do measurements without your patch
20:14 timotimo Files=777, Tests=29654, 206 wallclock secs ( 4.94 usr  0.76 sys + 763.32 cusr 29.49 csys = 798.51 CPU)
20:15 timotimo TEST_JOBS=4 make m-spectest  770,69s user 31,34s system 377% cpu 3:32,50 total
20:15 timotimo so that's not very unstable
20:16 nwc10 is sometimes slower for me, but I don't know what else is running on the machine
20:16 nwc10 seems nothing now I look again
20:16 timotimo Stage parse      :  44.971  Stage mast       :  20.451
20:17 nwc10 OK, I got better than that :-(
20:17 nwc10 but it *is* in the right direction.
20:17 timotimo that's slower than both my previous measurements for stage parse, but no change for stage mast
20:17 timotimo i suppose things in stage mast will die very young?
20:17 nwc10 I don't know enough to be sure
20:18 timotimo Stage parse      :  44.775  Stage mast       :  20.669
20:18 nwc10 I'm also building with -Og, so not maxing out gcc's optmiser, so I might get better speedups than real optimsed runs
20:18 timotimo i'm using -O1
20:19 timotimo Stage parse      :  45.013  Stage mast       :  20.571
20:19 timotimo so it's a little bit better, but not much
20:20 nwc10 OK. :-/
20:20 nwc10 but it's in the right direction.
20:25 timotimo Files=777, Tests=29654, 205 wallclock secs ( 4.97 usr  0.77 sys + 761.93 cusr 29.87 csys = 797.54 CPU)
20:25 timotimo TEST_JOBS=4 make m-spectest  771,09s user 31,94s system 358% cpu 3:44,21 total
20:25 timotimo i see much less cpu percentage
20:25 timotimo but more time all in all?
20:25 timotimo i'll have to re-run that at least once.
20:28 timotimo er, wait
20:28 timotimo i looked the wrong way around of course
20:28 timotimo your change gave a noticable win, actually
20:29 timotimo better cpu utilisation and 12 seconds less run time
20:29 timotimo second run coming up
20:29 nwc10 let's hope that the second run agrees
20:29 timotimo Files=777, Tests=29654, 205 wallclock secs ( 4.94 usr  0.72 sys + 761.93 cusr 29.62 csys = 797.21 CPU)
20:29 timotimo TEST_JOBS=4 make m-spectest  769,11s user 31,57s system 377% cpu 3:32,16 total
20:30 timotimo so the wallclock seconds seem to be the same
20:36 timotimo nwc10: still, good catch!
20:37 timotimo 0.19user 0.03system 0:00.23elapsed 98%CPU (0avgtext+0avgdata 97668maxresident)k ← without patch
20:37 timotimo (perl6-m -e 'say 1')
20:37 timotimo 0.20user 0.02system 0:00.22elapsed 98%CPU (0avgtext+0avgdata 97892maxresident)k
20:37 timotimo with your patch
20:38 timotimo unfortunately it's very noisy at that level, so i'd say there's no change there
20:38 diakopter heh
20:41 FROGGS joined #moarvm
20:41 timotimo well, the startup time being way too small is a *good* problem to have :D
21:27 dalek MoarVM/small_big_ints: 482343c | (Timo Paulssen)++ | src/math/smallbigintsupport. (2 files):
21:27 dalek MoarVM/small_big_ints: use MVMObject in smallbigintsupport.
21:27 dalek MoarVM/small_big_ints: review: https://github.com/MoarVM/MoarVM/commit/482343cc70
21:27 dalek MoarVM/small_big_ints: cc17337 | (Timo Paulssen)++ | src/ (4 files):
21:27 dalek MoarVM/small_big_ints: saner (but still not very sane) bigint coercion in ops
21:27 dalek MoarVM/small_big_ints: review: https://github.com/MoarVM/MoarVM/commit/cc17337794
21:36 dalek MoarVM/small_big_ints: 449ad14 | (Timo Paulssen)++ | src/math/bigintops.c:
21:36 dalek MoarVM/small_big_ints: this may be wrong, but it gets us further
21:36 dalek MoarVM/small_big_ints:
21:36 dalek MoarVM/small_big_ints: i think i'll have to get the data back into
21:36 dalek MoarVM/small_big_ints: the bigint object at some point?
21:36 dalek MoarVM/small_big_ints: review: https://github.com/MoarVM/MoarVM/commit/449ad14a0f
21:36 timotimo i think i need another rethink of this :p
21:48 hoelzro hehe, small big ints...
21:48 timotimo you're not the first ... :)
23:23 tgt joined #moarvm
23:40 lizmat joined #moarvm
23:52 benabik joined #moarvm

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary