00:18:03 slyrus [~chatzilla@adsl-75-36-215-204.dsl.pltn13.sbcglobal.net] has joined #sbcl 00:52:43 -!- stassats` [~stassats@wikipedia/stassats] has quit [Ping timeout: 240 seconds] 01:02:15 -!- slyrus [~chatzilla@adsl-75-36-215-204.dsl.pltn13.sbcglobal.net] has quit [Ping timeout: 252 seconds] 02:14:48 -!- rbarraud_ [~rbarraud@118-92-2-40.dsl.dyn.ihug.co.nz] has quit [Read error: Connection reset by peer] 02:49:57 rbarraud [~rbarraud@118-92-2-40.dsl.dyn.ihug.co.nz] has joined #sbcl 03:20:49 slyrus [~chatzilla@adsl-75-36-215-204.dsl.pltn13.sbcglobal.net] has joined #sbcl 13:47:38 ccl-logbot [~ccl-logbo@setf.clozure.com] has joined #sbcl 13:47:38 -!- names: ccl-logbot @nikodemus Kae hargettp tcr ThomasP1 Blkt @Krystof slyrus cmm christop` luis lnostdal `micro jsnell froydnj fe[nl]ix pkhuong @Xof holycow angavrilov chandler foom _3b` tokenrove joshe 13:49:41 -!- hargettp [~anonymous@pool-71-174-139-221.bstnma.east.verizon.net] has quit [Quit: hargettp] 13:56:42 hargettp [~anonymous@pool-71-174-139-221.bstnma.east.verizon.net] has joined #sbcl 14:04:09 no, it doesn't. could be made to work, but doesn't currently 14:04:35 that patch doesn't even build-as-is, as it runs into a metacircle during bootstrap 14:05:57 Krystof: but then, #'(LABELS FOO) and #"(LAMBDA (SB-PCL:.ARG0.)) don't work either, so... 14:07:31 Krystof: do you have any comments on the sb-debug:backtrace suggestion? 14:09:50 nikodemus: if #'(dfun foo) doesn't work, then you don't need at least some of the define-pcl-internal-name-syntax 14:09:57 or whatever the operator is called 14:17:12 Now I've defined almost all non-memory SSE intrinsics, but register allocation sometimes seems very strange: http://paste.lisp.org/display/114468 14:17:46 I wonder what's going wrong 14:19:34 Krystof: NAMED-LAMBDA is a suboptimal hammer here, really, because it requires valid-fun-name-p (or whatever) to pass 14:22:58 ah, ok 14:42:57 Is that kind of extraneous moving around a known issue, or did I mess up something in my vop definitions? 14:51:02 angavrilov: not sure. could be either. possibly lifetime issue in vops 14:54:09 http://who-t.blogspot.com/2009/12/on-commit-messages.html # we need to clone this and put it in our patch submission guidelines 15:54:44 ccl-logbot [~ccl-logbo@setf.clozure.com] has joined #sbcl 15:54:44 -!- names: ccl-logbot nyef hargettp Kae tcr ThomasP1 Blkt @Krystof slyrus cmm christop` luis lnostdal `micro jsnell froydnj fe[nl]ix pkhuong @Xof holycow angavrilov joshe tokenrove _3b` foom chandler 15:55:41 stassats [~stassats@wikipedia/stassats] has joined #sbcl 16:50:00 -!- christop` is now known as christoph_debian 17:35:02 Hmm, it seems that chaining vop arg -> tmp -> result via :target is too much for the register allocator. After I tweaked the definition to resemble the one used for standard float ops, most of the stupid moves disappeared. 17:38:59 stassats` [~stassats@pppoe.178-66-35-160.dynamic.avangarddsl.ru] has joined #sbcl 17:39:09 -!- stassats` [~stassats@pppoe.178-66-35-160.dynamic.avangarddsl.ru] has quit [Changing host] 17:39:09 stassats` [~stassats@wikipedia/stassats] has joined #sbcl 17:40:05 -!- Blkt [~user@dynamic-adsl-94-37-246-235.clienti.tiscali.it] has quit [Read error: Connection reset by peer] 17:41:26 -!- stassats [~stassats@wikipedia/stassats] has quit [Ping timeout: 264 seconds] 17:45:57 angavrilov: http://www.lisphacker.com/codex/001-sbcl-vops.txt 17:45:58 Blkt [~user@dynamic-adsl-94-37-246-235.clienti.tiscali.it] has joined #sbcl 17:46:43 (In short, :target is very close to stupidly useless, but not quite.) 17:58:34 Blkt` [~user@dynamic-adsl-94-34-31-251.clienti.tiscali.it] has joined #sbcl 18:02:30 -!- Blkt [~user@dynamic-adsl-94-37-246-235.clienti.tiscali.it] has quit [Ping timeout: 272 seconds] 18:25:40 ... Hunh. There's a hole in that argument. 18:26:11 Guess I'll have to look over the register allocator again. 18:48:46 A factor might be that using plain temporaries without lifetime adjustments inserts some randomizing factor into the allocator that on its own seems biased to stuff everything into xmm0... 18:50:32 Incidentally, many floating-point ops in x86-64 at least use something like "x :scs (single-reg single-float-immediate) :load-if (not (sc-is x single-float-immediate))". Isn't that load-if redundant as that variant is already included in scs? 18:51:29 I copied that pattern blindly, but want to understand the purpose. 18:56:27 You know, I've never actually looked into the :load-if stuff. 18:58:28 I sort of understand it in arith.lisp, where it is sometimes used to suppress loading in some unusual combinations that the op can handle better. But that stuff in float.lisp looks suspicious to me. Of course, I don't know a lot of stuff... 19:12:58 -!- Blkt` is now known as Blkt 19:33:42 -!- cmm [~cmm@bzq-79-181-203-193.red.bezeqint.net] has quit [Ping timeout: 272 seconds] 19:34:09 cmm [~cmm@bzq-79-181-203-193.red.bezeqint.net] has joined #sbcl 19:45:54 -!- ThomasP1 [~thomas@88-134-14-187-dynip.superkabel.de] has quit [Ping timeout: 272 seconds] 20:53:54 nikodemus [~nikodemus@cs181199216.pp.htv.fi] has joined #sbcl 20:53:54 -!- ChanServ has set mode +o nikodemus 20:55:45 ThomasP1 [~thomas@p54A5981C.dip0.t-ipconnect.de] has joined #sbcl 21:17:09 dbrunner [~user@p4FF7517C.dip.t-dialin.net] has joined #sbcl 21:22:14 minion [~minion@common-lisp.net] has joined #sbcl 21:23:39 specbot [~specbot@common-lisp.net] has joined #sbcl 21:25:03 -!- tcr [~tcr@cpc1-bour2-0-0-cust414.15-1.cable.virginmedia.com] has quit [Quit: Leaving.] 21:25:40 tcr [~tcr@cpc1-bour2-0-0-cust414.15-1.cable.virginmedia.com] has joined #sbcl 21:27:26 minion: say something 21:27:27 please stop playing with me... i am not a toy 21:57:55 -!- nikodemus [~nikodemus@cs181199216.pp.htv.fi] has quit [Ping timeout: 240 seconds] 21:58:23 nikodemus [~nikodemus@cs181199216.pp.htv.fi] has joined #sbcl 21:58:23 -!- ChanServ has set mode +o nikodemus 22:38:24 rbarraud [~rbarraud@118-92-19-211.dsl.dyn.ihug.co.nz] has joined #sbcl 22:43:23 -!- stassats` [~stassats@wikipedia/stassats] has quit [Ping timeout: 276 seconds] 22:52:09 -!- dbrunner [~user@p4FF7517C.dip.t-dialin.net] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] 22:53:05 dbrunner [~user@p4FF7517C.dip.t-dialin.net] has joined #sbcl 22:53:36 -!- dbrunner [~user@p4FF7517C.dip.t-dialin.net] has quit [Client Quit] 23:14:52 -!- christoph_debian [christoph@cl-1281.dus-01.de.sixxs.net] has quit [Read error: Operation timed out] 23:31:26 christoph_debian [christoph@sf-ogame.de] has joined #sbcl 23:33:55 -!- ThomasP1 [~thomas@p54A5981C.dip0.t-ipconnect.de] has quit [Remote host closed the connection] 23:41:05 -!- rbarraud [~rbarraud@118-92-19-211.dsl.dyn.ihug.co.nz] has quit [Read error: Operation timed out] 23:46:56 -!- nikodemus [~nikodemus@cs181199216.pp.htv.fi] has quit [Quit: This computer has gone to sleep] 23:46:59 looks like the result of a stupid translation from before constant float scs 23:48:22 -!- Krystof [~csr21@84-51-132-95.christ977.adsl.metronet.co.uk] has quit [Ping timeout: 252 seconds] 23:52:05 Wouldn't surprise me, given some of the other stuff I've seen in the backends. 23:58:20 it's all my code (: