02:57:42 mm... the SSB is slower than mprotect because we have to process the addresses. 03:04:53 i.e. it's slower to disable mprotect and update the page tables than to leave both logging and mprotect in place 03:59:45 ok. Coding this right, we have an 6% lower self-build time. 04:40:27 (no build failure ;) 04:40:37 so, still slower when it works. Sigh. 06:20:06 flip214 [~marek@2001:858:107:1:7a2b:cbff:fed0:c11c] has joined #sbcl 06:20:06 -!- flip214 [~marek@2001:858:107:1:7a2b:cbff:fed0:c11c] has quit [Changing host] 06:20:06 flip214 [~marek@unaffiliated/flip214] has joined #sbcl 07:10:44 -!- cmm [~cmm@109.65.214.78] has quit [Ping timeout: 240 seconds] 07:11:48 cmm [~cmm@bzq-109-65-214-78.red.bezeqint.net] has joined #sbcl 08:05:36 hlavaty [~user@91-65-223-81-dynip.superkabel.de] has joined #sbcl 08:06:09 -!- hlavaty [~user@91-65-223-81-dynip.superkabel.de] has left #sbcl 08:06:18 hlavaty [~user@91-65-223-81-dynip.superkabel.de] has joined #sbcl 09:12:54 -!- pchrist [~spirit@gentoo/developer/pchrist] has quit [Quit: leaving] 09:33:07 misterncw [~misterncw@82.71.241.25] has joined #sbcl 09:44:48 -!- cmm [~cmm@bzq-109-65-214-78.red.bezeqint.net] has quit [Ping timeout: 252 seconds] 09:45:47 cmm [~cmm@109.65.214.78] has joined #sbcl 09:56:29 Krystof [~csr21@host86-141-133-140.range86-141.btcentralplus.com] has joined #sbcl 09:56:29 -!- ChanServ has set mode +o Krystof 10:30:30 -!- Krystof [~csr21@host86-141-133-140.range86-141.btcentralplus.com] has quit [Ping timeout: 255 seconds] 10:34:08 Krystof [~csr21@host86-141-133-140.range86-141.btcentralplus.com] has joined #sbcl 10:34:08 -!- ChanServ has set mode +o Krystof 10:55:05 -!- cmm [~cmm@109.65.214.78] has quit [Remote host closed the connection] 10:55:24 cmm [~cmm@109.65.214.78] has joined #sbcl 11:00:43 attila_lendvai [~attila_le@catv-89-133-170-127.catv.broadband.hu] has joined #sbcl 11:00:44 -!- attila_lendvai [~attila_le@catv-89-133-170-127.catv.broadband.hu] has quit [Changing host] 11:00:44 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 11:01:40 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Disconnected by services] 11:01:40 attila_lendvai1 [~attila_le@apn-94-44-175-121.vodafone.hu] has joined #sbcl 11:01:41 -!- attila_lendvai1 is now known as attila_lendvai 11:01:41 -!- attila_lendvai [~attila_le@apn-94-44-175-121.vodafone.hu] has quit [Changing host] 11:01:42 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 11:29:28 -!- Krystof [~csr21@host86-141-133-140.range86-141.btcentralplus.com] has quit [Ping timeout: 258 seconds] 11:38:33 -!- deepfire [~deepfire@80.92.100.69] has quit [Ping timeout: 264 seconds] 11:51:02 -!- cmm [~cmm@109.65.214.78] has quit [Ping timeout: 252 seconds] 11:51:48 cmm [~cmm@109.65.214.78] has joined #sbcl 11:59:11 -!- peddie [~peddie@repl.esden.net] has quit [Quit: peace!] 11:59:41 peddie [~peddie@repl.esden.net] has joined #sbcl 12:09:47 homie`` [~levgue@xdsl-78-35-176-145.netcologne.de] has joined #sbcl 12:11:41 -!- homie` [~levgue@xdsl-78-35-154-33.netcologne.de] has quit [Ping timeout: 240 seconds] 12:32:59 tcr1 [~tcr@217-162-207-189.dynamic.hispeed.ch] has joined #sbcl 12:40:55 -!- homie`` [~levgue@xdsl-78-35-176-145.netcologne.de] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] 12:45:13 homie [~levgue@xdsl-78-35-176-145.netcologne.de] has joined #sbcl 13:15:36 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Connection reset by peer] 13:19:32 -!- antgreen [~user@CPE00222d6c4710-CM00222d6c470d.cpe.net.cable.rogers.com] has quit [Remote host closed the connection] 13:35:42 deepfire [~deepfire@80.92.100.69] has joined #sbcl 13:37:54 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 13:41:02 -!- cmm [~cmm@109.65.214.78] has quit [Ping timeout: 240 seconds] 13:41:06 -!- loke [~elias@bb219-74-91-22.singnet.com.sg] has quit [Read error: Connection reset by peer] 13:42:02 cmm [~cmm@bzq-109-65-214-78.red.bezeqint.net] has joined #sbcl 13:43:53 <_8david> pkhuong: if that's a problem, could you do work to compactify the buffer in a background thread regularly, thereby at least reducing latency due to that address processing (if you take care to process only the parts of the buffer over which memory visibility guarantees that the mutation has moved over already)? 13:45:20 loke [~elias@bb119-74-97-62.singnet.com.sg] has joined #sbcl 13:46:36 -!- flip214 [~marek@unaffiliated/flip214] has quit [Remote host closed the connection] 13:50:59 tcr2 [~tcr@217-162-207-189.dynamic.hispeed.ch] has joined #sbcl 13:50:59 -!- tcr1 [~tcr@217-162-207-189.dynamic.hispeed.ch] has quit [Read error: Connection reset by peer] 13:57:31 -!- tcr2 [~tcr@217-162-207-189.dynamic.hispeed.ch] has quit [Quit: Leaving.] 14:10:32 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Connection reset by peer] 14:10:58 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 14:10:59 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 14:10:59 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 14:17:46 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Disconnected by services] 14:18:55 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 14:19:14 attila_lendvai1 [~attila_le@apn-94-44-175-121.vodafone.hu] has joined #sbcl 14:19:14 -!- attila_lendvai1 [~attila_le@apn-94-44-175-121.vodafone.hu] has quit [Client Quit] 14:37:33 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Remote host closed the connection] 14:38:07 attila_lendvai [~attila_le@apn-94-44-175-121.vodafone.hu] has joined #sbcl 14:38:07 -!- attila_lendvai [~attila_le@apn-94-44-175-121.vodafone.hu] has quit [Changing host] 14:38:07 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 14:42:11 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Connection reset by peer] 14:43:01 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 14:43:01 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 14:43:01 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 15:06:33 _8david: yeah. I guess. But the overhead from logging writes is still pretty big. 15:09:21 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Ping timeout: 276 seconds] 15:09:30 Just enabling the write barrier without using it adds 15% runtime. re-enabling mprotect tricks on top of SW write barriers adds ~noise. 15:11:59 (processing the SSB and disabling mprotect adds 5%) 15:13:12 I guess the bootstrap just doesn't really write to old data. 15:13:29 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 15:13:29 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 15:13:29 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 15:18:16 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Connection reset by peer] 15:19:31 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 15:19:38 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 15:19:38 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 15:25:37 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Quit: Leaving.] 15:25:41 attila_lendvai1 [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 15:25:41 -!- attila_lendvai1 is now known as attila_lendvai 15:25:41 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 15:25:41 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 15:34:58 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Connection reset by peer] 15:35:03 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 15:35:03 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 15:35:03 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 15:37:04 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Connection reset by peer] 15:42:02 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 15:42:02 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 15:42:02 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 15:44:27 -!- misterncw [~misterncw@82.71.241.25] has quit [] 15:48:00 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Ping timeout: 244 seconds] 15:52:16 pdlogan [~patrick@174-25-37-137.ptld.qwest.net] has joined #sbcl 15:56:49 -!- scymtym [~user@2001:638:504:2093:21a:a0ff:fe34:2d7d] has quit [Remote host closed the connection] 15:57:03 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 15:57:03 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 15:57:03 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 15:57:54 attila_lendvai1 [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 15:57:54 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Disconnected by services] 15:57:55 -!- attila_lendvai1 is now known as attila_lendvai 15:57:56 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 15:57:56 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 15:58:35 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Connection reset by peer] 15:59:11 -!- loke [~elias@bb119-74-97-62.singnet.com.sg] has quit [Ping timeout: 240 seconds] 15:59:54 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 15:59:54 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 15:59:55 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 16:01:22 loke [~elias@bb220-255-247-251.singnet.com.sg] has joined #sbcl 16:05:55 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Connection reset by peer] 16:06:09 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 16:06:09 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 16:06:09 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 16:12:21 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Connection reset by peer] 16:12:47 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 16:12:47 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 16:12:47 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 16:14:46 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has left #sbcl 16:34:44 well, I've got something that's really good at benchmarketing ;) 16:34:52 give it a loop, and it'll do very well. 16:35:53 basically 0 overhead on array traversals ;) 16:40:24 -!- loke [~elias@bb220-255-247-251.singnet.com.sg] has quit [Ping timeout: 255 seconds] 16:40:46 <_8david> For the build benchmark, does the number of cache misses increase due to SSB writes? (If so, would non-buffer techniques that rely on being to index into a linear table for dynamic space fare better in that respect?) 16:42:09 loke [~elias@bb119-74-201-244.singnet.com.sg] has joined #sbcl 16:43:25 well, we've got one customer who doesn't think that being dying on foreign heap tricks is unacceptable ;) 16:44:56 I'll run perf on it and see. 16:45:23 It shouldn't do (well, not too much), since the SSB is so nicely linear 16:46:57 -!- hlavaty [~user@91-65-223-81-dynip.superkabel.de] has quit [Ping timeout: 264 seconds] 16:48:08 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 16:48:09 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 16:48:09 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 16:50:34 -!- cmm [~cmm@bzq-109-65-214-78.red.bezeqint.net] has quit [Ping timeout: 240 seconds] 16:51:39 cmm [~cmm@bzq-109-65-214-78.red.bezeqint.net] has joined #sbcl 16:53:52 ok. So, it turns out it took me ~80 cycles to process each entry because I sorted them first. 16:54:27 without a sort, it's down to ~8-9. 17:25:21 -!- loke [~elias@bb119-74-201-244.singnet.com.sg] has quit [Ping timeout: 240 seconds] 17:26:15 loke [~elias@bb220-255-249-12.singnet.com.sg] has joined #sbcl 17:27:37 pkhuong pasted "perf stat for self build with and without write barriers" at http://paste.lisp.org/display/122969 17:35:24 yeah. So we process ~1G writes, at 8 bytes each. 17:36:09 that accounts for the cache misses (one miss on logging, another when processing) 17:48:41 we'll see what logging 32 bits only does 18:15:18 -!- cmm [~cmm@bzq-109-65-214-78.red.bezeqint.net] has quit [Ping timeout: 252 seconds] 18:16:20 cmm [~cmm@109.65.214.78] has joined #sbcl 18:43:29 pkhuong annotated #122969 "Log middle 32 bits" at http://paste.lisp.org/display/122969#1 18:51:38 hargettp_ [~hargettp_@dhcp-161.mirrorimage.net] has joined #sbcl 19:01:12 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Operation timed out] 19:05:51 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 19:05:51 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 19:05:52 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 19:15:53 pchrist [~spirit@gentoo/developer/pchrist] has joined #sbcl 19:27:05 tcr1 [~tcr@80-218-247-218.dclient.hispeed.ch] has joined #sbcl 19:33:47 rpg [~rpg@mpls.sift.info] has joined #sbcl 19:40:17 sigh. On one hand, 10% slowdown is obviously bad. On the other hand, processes running in other cores have a much greater effect on build times :| 19:48:26 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Read error: Operation timed out] 19:49:36 attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has joined #sbcl 19:49:38 -!- attila_lendvai [~attila_le@catv-89-132-188-166.catv.broadband.hu] has quit [Changing host] 19:49:38 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 19:51:27 -!- cmm [~cmm@109.65.214.78] has quit [Ping timeout: 258 seconds] 19:52:26 cmm [~cmm@bzq-109-65-214-78.red.bezeqint.net] has joined #sbcl 20:08:42 antgreen [~user@CPE00222d6c4710-CM00222d6c470d.cpe.net.cable.rogers.com] has joined #sbcl 20:30:36 -!- cmm [~cmm@bzq-109-65-214-78.red.bezeqint.net] has quit [Ping timeout: 252 seconds] 20:31:24 cmm [~cmm@109.65.214.78] has joined #sbcl 20:43:11 prxq [~mommer@mnhm-5f75c1ef.pool.mediaWays.net] has joined #sbcl 21:57:53 slyrus_ [~chatzilla@173-228-44-88.dsl.static.sonic.net] has joined #sbcl 22:29:55 pkhuong annotated #122969 "in-line slow path" at http://paste.lisp.org/display/122969#2 23:00:27 -!- prxq [~mommer@mnhm-5f75c1ef.pool.mediaWays.net] has quit [Quit: Leaving] 23:01:58 -!- slyrus_ [~chatzilla@173-228-44-88.dsl.static.sonic.net] has quit [Ping timeout: 258 seconds] 23:09:38 -!- tcr1 [~tcr@80-218-247-218.dclient.hispeed.ch] has quit [Quit: Leaving.] 23:19:16 -!- rpg [~rpg@mpls.sift.info] has quit [Ping timeout: 252 seconds] 23:27:23 yeah. Here's the thing: pages that have been mutated don't get re-protected until they're GCed 23:30:39 -!- pdlogan [~patrick@174-25-37-137.ptld.qwest.net] has left #sbcl 23:34:37 also, interestingly enough, there are usually less than ~300k logged writes per GC (about half of which are for literals)