00:42:18 dto [~user@pool-96-252-62-13.bstnma.fios.verizon.net] has joined #ccl 00:48:30 Vivitron [~Vivitron@pool-98-110-213-33.bstnma.fios.verizon.net] has joined #ccl 01:03:31 erikc [~erikc@CPE00222d53fe78-CM00222d53fe75.cpe.net.cable.rogers.com] has joined #ccl 01:27:58 -!- Vivitron [~Vivitron@pool-98-110-213-33.bstnma.fios.verizon.net] has quit [Remote host closed the connection] 01:45:06 -!- dented42 [~dented42@opengroove.org] has quit [Ping timeout: 264 seconds] 01:51:30 dented42 [~dented42@opengroove.org] has joined #ccl 02:38:09 -!- dto [~user@pool-96-252-62-13.bstnma.fios.verizon.net] has quit [Remote host closed the connection] 02:54:40 bfulgham_ [~brent@cpe-76-173-170-144.socal.res.rr.com] has joined #ccl 03:09:40 -!- bhyde [~bhyde@c-24-61-81-138.hsd1.ma.comcast.net] has quit [Quit: bhyde] 03:34:51 -!- DataLinkDroid [~DataLinkD@120.154.70.10] has quit [Ping timeout: 256 seconds] 03:46:34 -!- erikc [~erikc@CPE00222d53fe78-CM00222d53fe75.cpe.net.cable.rogers.com] has quit [Quit: erikc] 03:54:54 DataLinkDroid [~DataLinkD@1.150.178.55] has joined #ccl 04:26:08 -!- bfulgham_ [~brent@cpe-76-173-170-144.socal.res.rr.com] has quit [Quit: bfulgham_] 04:42:01 bfulgham_ [~brent@cpe-76-173-170-144.socal.res.rr.com] has joined #ccl 04:58:54 -!- DataLinkDroid [~DataLinkD@1.150.178.55] has quit [Quit: Bye] 05:35:42 -!- Fare [fare@nat/google/x-aomcdazzzscpvxom] has quit [Ping timeout: 252 seconds] 05:52:02 Vivitron [~Vivitron@pool-98-110-213-33.bstnma.fios.verizon.net] has joined #ccl 06:26:33 -!- bfulgham_ [~brent@cpe-76-173-170-144.socal.res.rr.com] has quit [Quit: bfulgham_] 06:36:35 qh [~user@222.41.8.163] has joined #ccl 10:46:46 -!- qh [~user@222.41.8.163] has quit [Ping timeout: 258 seconds] 11:53:38 bhyde [~bhyde@c-24-61-81-138.hsd1.ma.comcast.net] has joined #ccl 12:49:37 qh [~user@120.128.6.61] has joined #ccl 13:26:20 Fare [fare@nat/google/x-dvvhcvyovunwtugd] has joined #ccl 14:12:20 -!- qh [~user@120.128.6.61] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] 14:22:00 -!- bhyde [~bhyde@c-24-61-81-138.hsd1.ma.comcast.net] has quit [Quit: bhyde] 14:28:49 bhyde [~bhyde@c-24-61-81-138.hsd1.ma.comcast.net] has joined #ccl 14:32:33 -!- sellout- [~Adium@c-98-245-92-119.hsd1.co.comcast.net] has quit [Quit: Leaving.] 14:32:45 gbyers: how about coming on the week of May 6 to 10 ? 14:41:51 -!- Fare [fare@nat/google/x-dvvhcvyovunwtugd] has quit [Ping timeout: 252 seconds] 14:44:39 alms_ [~alms_@209-6-130-32.c3-0.bkl-ubr1.sbo-bkl.ma.cable.rcn.com] has joined #ccl 14:50:25 Fare [fare@nat/google/x-umqguwrllncklsfk] has joined #ccl 15:08:54 sellout- [~Adium@70.96.9.235] has joined #ccl 15:51:47 -!- asedeno [asedeno@nat/google/x-cwjftwvkibmhnxes] has quit [Remote host closed the connection] 16:00:27 asedeno [asedeno@nat/google/x-mrmpgvwswoznwuyk] has joined #ccl 16:02:43 -!- asedeno [asedeno@nat/google/x-mrmpgvwswoznwuyk] has quit [Client Quit] 16:03:19 asedeno [asedeno@nat/google/x-moxlyuctlwwlrgoi] has joined #ccl 16:19:19 erikc [~erikc@209.20.28.194] has joined #ccl 16:36:28 -!- alms_ [~alms_@209-6-130-32.c3-0.bkl-ubr1.sbo-bkl.ma.cable.rcn.com] has quit [Quit: alms_] 16:38:24 -!- bhyde [~bhyde@c-24-61-81-138.hsd1.ma.comcast.net] has quit [Quit: bhyde] 16:40:45 what is the recommended way to profile lisp code with ccl, already? traversing the system for poiu is >10 times slower on ccl than on sbcl. 16:41:06 (22 seconds vs less than 2 seconds) 16:41:33 (which negates the gains from parallelism) 16:43:36 Fare: I think the recommendation is still oprofile: http://ccl.clozure.com/manual/chapter12.1.html 16:45:21 alms_ [~alms_@209-6-130-32.c3-0.bkl-ubr1.sbo-bkl.ma.cable.rcn.com] has joined #ccl 16:45:40 ok 16:47:52 -!- alms_ [~alms_@209-6-130-32.c3-0.bkl-ubr1.sbo-bkl.ma.cable.rcn.com] has quit [Client Quit] 17:07:11 Fare: Hans wrote an advice-based profiler in ccl:contrib;huebner;advice-profiler; that you might want to try. 17:23:40 thanks 17:28:53 -!- erikc [~erikc@209.20.28.194] has left #ccl 17:39:44 -!- sellout- [~Adium@70.96.9.235] has quit [Quit: Leaving.] 18:08:33 alms_ [~alms_@173-162-137-153-NewEngland.hfc.comcastbusiness.net] has joined #ccl 18:09:25 'perf' is another option. 18:11:14 bhyde [~bhyde@c-24-61-81-138.hsd1.ma.comcast.net] has joined #ccl 18:12:49 perf is a great tool 18:13:03 a bit too detailed for most programs, though 18:13:38 Compared to oprofile ? Nor sure. 18:13:53 Not. Sheesh. 18:14:11 I haven't used oprofile 18:21:50 Recent versions of 'perf' have call graph support, so you can ask 'who's calling FOO?' Unfortunately, it's confused by the way that CCL does multiple-value returns (which involves an extra return address) 18:23:00 that's a familiar situation 18:23:22 C tool being confused by CL implementation :D 18:26:06 Fare: try perf stat dd if=/dev/zero of=/dev/null bs=64k count=65536 18:27:19 May involve installing perf, which is sometimes part of a package named 'util-linux' or some other name I never remember. 18:33:52 gbyers, can you come on week of may 6-10 ? 18:34:05 Sure. 18:34:35 ok, will talk to allan 18:34:51 Great. 18:37:40 -!- bhyde [~bhyde@c-24-61-81-138.hsd1.ma.comcast.net] has quit [Quit: bhyde] 18:46:04 -!- xristos is now known as xristos` 18:51:37 bhyde [~bhyde@c-24-61-81-138.hsd1.ma.comcast.net] has joined #ccl 18:54:47 -!- alms_ [~alms_@173-162-137-153-NewEngland.hfc.comcastbusiness.net] has quit [Quit: alms_] 18:57:44 sellout- [~Adium@70.96.9.235] has joined #ccl 19:25:06 alms_ [~alms_@173-162-137-153-NewEngland.hfc.comcastbusiness.net] has joined #ccl 20:14:46 -!- sellout- [~Adium@70.96.9.235] has quit [Read error: Connection reset by peer] 20:16:31 sellout- [~Adium@70.96.9.235] has joined #ccl 20:53:25 -!- Fare [fare@nat/google/x-umqguwrllncklsfk] has quit [Ping timeout: 256 seconds] 21:03:52 -!- alms_ [~alms_@173-162-137-153-NewEngland.hfc.comcastbusiness.net] has quit [Quit: alms_] 21:16:35 alms_ [~alms_@209-6-130-32.c3-0.bkl-ubr1.sbo-bkl.ma.cable.rcn.com] has joined #ccl 22:06:31 Fare [fare@nat/google/x-rjlbryrovwowvmbn] has joined #ccl 22:06:31 -!- Vivitron [~Vivitron@pool-98-110-213-33.bstnma.fios.verizon.net] has quit [Ping timeout: 246 seconds] 22:08:00 -!- jdz [~jdz@85.254.212.34] has quit [Excess Flood] 22:08:36 jdz [~jdz@85.254.212.34] has joined #ccl 22:27:53 -!- sellout- [~Adium@70.96.9.235] has quit [Quit: Leaving.] 22:35:34 DataLinkDroid [~DataLinkD@1.144.81.169] has joined #ccl 22:46:55 dto [~user@pool-96-252-62-13.bstnma.fios.verizon.net] has joined #ccl 22:47:06 hey rme and friends. 22:48:29 rme: does ccl have its own profiler like sbprof? here's the situation: the larger levels in 2x0ng seem to perform much worse in the CCL version, and I suspect a few declarations will fix the issue. 22:49:21 i.e. i think it's something i do in collision detection that happens to optimize alright on sbcl but not on ccl. 22:49:40 so at some point while messing with that i might want to pick your brain / ask a few q's if you have time 22:57:03 I just asked, and the answer was oprofile or perf 23:03:29 -!- bhyde [~bhyde@c-24-61-81-138.hsd1.ma.comcast.net] has quit [Quit: bhyde] 23:04:00 oh hello Fare. thanks 23:04:03 how are you? 23:04:48 super busy and stressed 23:06:04 ok. i won't bug you. 23:07:33 *patrickwonders* managed to adapt this: http://ccl.clozure.com/manual/chapter12.2.html#Profiling-Using-Apples-CHUD-metering-tools to Apple's current toolset with mostly-success a month ago. I should have taken/published good notes. 23:08:03 ah interesting patrickwonders . 23:08:08 CHUD is a great name doo 23:10:14 IIRC, I did pretty much what the 'Usage' said, but I had the Apple Perf tool attach to an existing process. so I stopped at a point before I was interested, attached the Perf tool, and then continued. I don't think I ran inside (chud:meter ) though. 23:11:46 CHUD had ways of stopping/starting the profiler under program control. (As if that might be useful!) 23:29:21 ah. good i can reproduce the performance issue, that means I can fix it :) 23:29:29 *dto* pulls out his copy of PAIP 23:31:54 hmm, what is Prof? i found oprofile, but was hoping to investigate that too 23:32:30 i mean perf. 23:32:34 hmm. 23:32:47 ahh 23:32:54 It's a profiler available in newer versions of Linux. 23:33:30 yes i couldn't find it in my distro, i was looking for the wrong name. 23:33:33 sorry, i'm lame :) 23:33:38 thanks for your help. 23:34:39 It may be included as part of a package named 'linux-utils' or 'util-linux' ; I never remember. 23:36:47 linux-tools-common on here. plus some libs 23:37:15 You basically create an image with ELF symbol info (as per the oprofile doc), then 'perf record ccl -e "(something)" -e "(quit)"', then 'perf report'. 23:37:47 interesting. ok, here goes. :) 23:38:38 It's tied to specific versions of the Linux kernel, so you may have to install the specic version for your kernel too; once that's done, the package system should update it when the kernel is updated. 23:43:12 do you mean 'perf record ccl_image_filename -e ..." 23:43:53 Yes, ... ccl -I image ... 23:43:58 ahhh 23:44:54 ok it works now and wrote some profiling data. thank you. i'll play the level that gets slow... 23:51:39 and compare it with data from a fast level 23:54:34 hmm. what do i do with the data now? i don't have opreport. i'll fin dit. 23:56:08 perf report 23:56:27 You're using 'perf 23:57:07 oh , i was following the documentation too closely. 23:59:10 I don't know if it's changed recently, but 'perf report' used to assume that you're on a Linux console and send escape sequences specific to that terminal type. There's a way of suppressing that that involves a 'stdio' option, and there's a man page ... 23:59:22 im just doing it in a terminal. 23:59:30 thanks so much for your help.