01:44:25 -!- stassats` [~stassats@wikipedia/stassats] has quit [Ping timeout: 252 seconds] 02:28:58 lichtblau: how's the fixed thread tests going? 04:23:51 attila_lendvai [~attila_le@37.99.50.32] has joined #sbcl 04:23:52 -!- attila_lendvai [~attila_le@37.99.50.32] has quit [Changing host] 04:23:52 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 04:59:03 angavrilov [~angavrilo@217.71.227.190] has joined #sbcl 05:19:43 edgar-rft [~GOD@HSI-KBW-078-043-123-191.hsi4.kabel-badenwuerttemberg.de] has joined #sbcl 05:31:55 sdemarre [~serge@91.176.6.113] has joined #sbcl 06:03:24 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Quit: Leaving.] 07:00:16 -!- edgar-rft [~GOD@HSI-KBW-078-043-123-191.hsi4.kabel-badenwuerttemberg.de] has quit [Quit: this computer sucks] 08:33:47 -!- wbooze [~wbooze@xdsl-78-35-134-112.netcologne.de] has quit [Quit: none] 08:35:12 wbooze [~wbooze@xdsl-78-35-134-112.netcologne.de] has joined #sbcl 08:42:33 woudshoo [~user@ironhead.xs4all.nl] has joined #sbcl 09:31:40 SBCL needs to get faster bignums :C 09:42:55 -!- sdemarre [~serge@91.176.6.113] has quit [Ping timeout: 246 seconds] 09:56:53 sdemarre [~serge@91.176.71.210] has joined #sbcl 10:19:11 -!- wbooze [~wbooze@xdsl-78-35-134-112.netcologne.de] has quit [Quit: none] 10:19:38 wbooze [~wbooze@xdsl-78-35-134-112.netcologne.de] has joined #sbcl 11:12:23 -!- wbooze [~wbooze@xdsl-78-35-134-112.netcologne.de] has quit [Ping timeout: 252 seconds] 11:22:54 -!- woudshoo [~user@ironhead.xs4all.nl] has quit [Ping timeout: 276 seconds] 11:49:21 woudshoo [~user@ironhead.xs4all.nl] has joined #sbcl 12:27:37 gko [~user@114-34-168-13.HINET-IP.hinet.net] has joined #sbcl 12:46:05 LiamH [~healy@96.231.222.72] has joined #sbcl 13:09:36 attila_lendvai [~attila_le@37.99.50.32] has joined #sbcl 13:09:36 -!- attila_lendvai [~attila_le@37.99.50.32] has quit [Changing host] 13:09:36 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 13:10:36 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Client Quit] 13:19:49 -!- woudshoo [~user@ironhead.xs4all.nl] has quit [Remote host closed the connection] 13:20:02 woudshoo [~user@ironhead.xs4all.nl] has joined #sbcl 13:21:40 -!- LiamH [~healy@96.231.222.72] has quit [Ping timeout: 246 seconds] 13:22:14 LiamH [~healy@96.231.222.72] has joined #sbcl 13:25:45 -!- woudshoo [~user@ironhead.xs4all.nl] has quit [Ping timeout: 276 seconds] 13:54:11 woudshoo [~user@ironhead.xs4all.nl] has joined #sbcl 13:56:34 stassats [~stassats@wikipedia/stassats] has joined #sbcl 13:59:07 attila_lendvai [~attila_le@37.99.50.32] has joined #sbcl 13:59:08 -!- attila_lendvai [~attila_le@37.99.50.32] has quit [Changing host] 13:59:08 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 14:06:58 -!- easye [~user@213.33.70.157] has quit [Ping timeout: 268 seconds] 14:42:51 easye [~user@213.33.70.157] has joined #sbcl 14:51:33 -!- sdemarre [~serge@91.176.71.210] has quit [Ping timeout: 276 seconds] 15:08:38 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Ping timeout: 252 seconds] 15:21:17 leuler [~user@p548FC316.dip.t-dialin.net] has joined #sbcl 15:26:00 -!- easye [~user@213.33.70.157] has quit [Remote host closed the connection] 15:27:54 Quadrescence: Just curious: How large are the bignums you are interested in and which operations on them specifically? 15:49:23 easye [~user@213.33.70.157] has joined #sbcl 16:00:13 -!- stassats [~stassats@wikipedia/stassats] has quit [Ping timeout: 246 seconds] 16:02:35 attila_lendvai [~attila_le@37.99.50.32] has joined #sbcl 16:02:35 -!- attila_lendvai [~attila_le@37.99.50.32] has quit [Changing host] 16:02:35 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 16:14:46 -!- woudshoo [~user@ironhead.xs4all.nl] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] 16:29:34 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Quit: Leaving.] 16:35:48 stassats [~stassats@wikipedia/stassats] has joined #sbcl 16:40:09 *froydnj* is curious to hear Quadrescence's answer to this too 16:54:56 attila_lendvai [~attila_le@37.99.50.32] has joined #sbcl 16:54:56 -!- attila_lendvai [~attila_le@37.99.50.32] has quit [Changing host] 16:54:56 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 17:02:22 -!- gko [~user@114-34-168-13.HINET-IP.hinet.net] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] 17:14:00 luis- [~luis@nhop.r42.eu] has joined #sbcl 17:17:42 -!- luis` [~luis@nhop.r42.eu] has quit [Ping timeout: 246 seconds] 17:17:42 -!- LiamH [~healy@96.231.222.72] has quit [Ping timeout: 246 seconds] 17:18:05 LiamH [~healy@96.231.222.72] has joined #sbcl 17:20:01 woudshoo [~user@ironhead.xs4all.nl] has joined #sbcl 17:25:15 -!- woudshoo [~user@ironhead.xs4all.nl] has quit [Ping timeout: 244 seconds] 17:46:25 woudshoo [~user@ironhead.xs4all.nl] has joined #sbcl 17:50:34 wbooze [~wbooze@xdsl-87-79-194-178.netcologne.de] has joined #sbcl 18:15:55 edgar-rft [~GOD@HSI-KBW-078-043-123-191.hsi4.kabel-badenwuerttemberg.de] has joined #sbcl 18:49:24 -!- christop` is now known as christoph_debian 18:57:45 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Quit: Leaving.] 19:16:35 -!- edgar-rft [~GOD@HSI-KBW-078-043-123-191.hsi4.kabel-badenwuerttemberg.de] has quit [Quit: nuclear meltdown] 19:23:30 -!- woudshoo [~user@ironhead.xs4all.nl] has quit [Quit: going to bed] 21:02:14 leuler, I was doing some large digit computations for integer lattice reductions 21:02:40 (also messing about with general binary splitting of hypergeometric series) 21:16:00 Quadrescence: So your integers are large enough that the quadratic time algorithms of SBCL's bignums hurt? I was just wondering whether your cause could be helped by anything short of basing SBCL's bignums on the gmp library. 21:18:12 leuler, Well I've been trying to profile without a lot of success. I know a somewhat equivalent program written in C++ using GMP (sloppy C++ mind you) is on the order of 10 times faster (actually seemingly quadratically faster, but 10x for some 500k digit stuff) 21:19:12 are you sure that the bignums are the bottleneck? 21:20:21 stassats, According to the profiling, it seems so. For the series computations, the other other major thing going on is creation of structs (containing four bignums), funcalls of pure functions, and log(N) recursive divisions (where the total size of what's divided fits in a fixnum) 21:23:00 They're all integer computations. I actually had one final major / computation, which did a gcd, and increased the length of time taken by 75% (so GCD was a real bottleneck) 21:27:23 I happen to have tried Karatsuba multiplication in SBCL a while ago just for fun (a much too simple implementation just using bignums, consing madly). It started being faster than the current implementation above 1000 or 10000 bits length or something, I don't remember exactly. If your digits are 64 bits each (?) a factor of ten should be easily achievable (for multiplication). 21:27:40 hm, why is there a (sb!xc:deftype bignum-type () 'bignum) ? 21:28:10 Concerning GCD I thought that Juho's GCD implementation is already quite good? 21:29:03 stassats: I think sb-bignum was meant to be portable 21:30:17 leuler: although, just calling out to gmp with a conversion to sign/magnitude would be simpler (: 21:32:00 i'd rather accept faster small bignums 21:32:04 pkhuong: yes. 21:33:03 stassats: "small" like one or two digits long? 21:33:29 somewhere there, yes 21:34:28 some way of reusing already constructed bignums would be interesting 21:37:18 it'd break EQL hash tables in an interesting way. 21:38:08 pkhuong: no, i mean in a single function, where it can be proved (or declared) that it cannot escape 21:44:45 leuler, the GCD might be fine, it was just a bottleneck. Doesn't imply it's necessarily slow. I haven't tested it specifically. 21:49:58 Also, I somehow doubt SBCL is doing fixnum*bignum optimizations. That'd be nice. (But maybe it is.) 21:58:36 bignum times fixnum is implemented as its own function, but not bignum divided by fixnum. 22:00:16 ah okay 22:04:03 If anyone's curious, this is the series code I was messing about with https://bitbucket.org/tarballs_are_good/lisp-random/src/45d3c3d6c389/binsplit.lisp 22:15:14 lggr [~lggr@84-73-159-126.dclient.hispeed.ch] has joined #sbcl 23:17:41 -!- leuler [~user@p548FC316.dip.t-dialin.net] has quit [Quit: ERC Version 5.1.2 $Revision: 1.796.2.6 $ (IRC client for Emacs)] 23:37:50 -!- wbooze [~wbooze@xdsl-87-79-194-178.netcologne.de] has quit [Read error: Operation timed out]