Welcome to the main channel on the development of MoarVM, a virtual machine for NQP and Rakudo (moarvm.org). This channel is being logged for historical purposes.
Set by lizmat on 24 May 2021.
00:02 reportable6 left 03:05 reportable6 joined 03:51 frost left 04:49 squashable6 left 05:00 [Coke] left 05:51 squashable6 joined 06:02 reportable6 left 06:09 frost joined 06:39 [Coke] joined 07:39 evalable6 left, linkable6 left, linkable6 joined 07:42 evalable6 joined 08:04 reportable6 joined 09:40 nativecallable6 left, sourceable6 left, unicodable6 left, linkable6 left, statisfiable6 left, reportable6 left, squashable6 left, shareable6 left, greppable6 left, bisectable6 left, quotable6 left, bloatable6 left, coverable6 left, committable6 left, notable6 left, benchable6 left, tellable6 left, evalable6 left, releasable6 left 09:41 tellable6 joined, nativecallable6 joined, greppable6 joined 09:42 notable6 joined, squashable6 joined, linkable6 joined 09:43 statisfiable6 joined, bisectable6 joined, coverable6 joined, unicodable6 joined 10:28 frost left 10:29 dogbert17 left, dogbert11 joined 10:34 frost joined 10:37 dogbert17 joined 10:38 dogbert11 left 10:40 dogbert11 joined 10:41 bloatable6 joined, dogbert17 left, reportable6 joined 10:43 shareable6 joined 11:00 nebuchadnezzar joined 11:04 dogbert17 joined 11:06 dogbert11 left, dogbert11 joined 11:09 dogbert17 left 11:41 quotable6 joined 11:43 sourceable6 joined 12:02 reportable6 left
timo SDL_SetRenderDrawColor($render, 0x00, 0x00, 0x00, 0xff) ends up compiling to four getspeshslots followed by getting the bigint out of the objects, then passing them as an integer argument. but that's not on the nativecall dispatcher invoke branch, so perhaps it's better there? it's also possible that i didn't do the multi sub correctly such that it decides to take the Int rather than int candidates 12:47
12:52 frost left 12:57 frost joined 13:04 reportable6 joined
MasterDuke huh. i did a --profile-compile of a smaller version of the script with lots of variables (only 20k) and it reported 120417 deoptimizations, 20002 global deoptimization, and 40854 OSRs 13:10
those numbers seem higher than i would expect
timo depends on where the deopts happen, of course. but if you have a tight loop that deopts and osr's every time it goes around the loop that could happen 13:39
MasterDuke yeah, looks like compile_all_the_stmts gen/moar/stage2/QAST.nqp:6450 has 20k of the OSRs and 60k of the deopts 13:41
13:43 benchable6 joined
timo that's very funny 13:46
dogbert11 ===SORRY!=== Error while compiling /home/dogbert/repos/rakudo/t/spec/S06-routine-modifiers/proxy.t
Container spec 'code_pair' must be configured with a fetch
happens when running with a (very) small nursery 13:47
14:38 frost left 14:42 committable6 joined 14:43 evalable6 joined, releasable6 joined 18:00 dogbert17 joined 18:02 reportable6 left 18:03 dogbert11 left 18:05 reportable6 joined 18:06 dogbert17 left, dogbert17 joined 21:29 dogbert17 left, dogbert17 joined 21:40 gabriel80546 joined 22:48 benchable6 left, nativecallable6 left, notable6 left, squashable6 left, reportable6 left, evalable6 left, statisfiable6 left, linkable6 left, committable6 left, coverable6 left, sourceable6 left, unicodable6 left, bisectable6 left, shareable6 left, quotable6 left, greppable6 left, releasable6 left, bloatable6 left, tellable6 left, evalable6 joined, coverable6 joined 22:49 notable6 joined, committable6 joined 22:50 unicodable6 joined, tellable6 joined, sourceable6 joined 22:51 reportable6 joined 23:41 gabriel80546 left 23:48 benchable6 joined, squashable6 joined 23:49 shareable6 joined 23:50 quotable6 joined 23:51 bisectable6 joined 23:52 kjp left