[Cs-dev] ParCS and request

[Cs-dev] ParCS and request

Date	2010-08-17 17:28
From	John ff
Subject	[Cs-dev] ParCS and request
	I have moved a little code around and optimised a little. The request is to try trapped with more than one thread. I am getting an "impossible" error message with 2 threads, and would like to know if this is repeatable. Those concerned bout speed, set #define TRACE 0 in H/cs_par_base.h and also remove the call to cs_dag_print in Engine/cs_par_dispatch.c function csp_dag_optimization At present I am setting various traces and not sure what value arrives in CVS. And,... it is the elapsed time that matters, not CPU as the code uses spinlocks which cost CPU but do not cause a process swap ==John ffitch ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2010-08-17 17:35
From	Michael Gogins
Subject	Re: [Cs-dev] ParCS and request
	Will do as soon as I can build ParCS! Regards, Mike On Tue, Aug 17, 2010 at 12:28 PM, John ff wrote: > I have moved a little code around and optimised a little. > > The request is to try trapped with more than one thread. I am getting > an "impossible" error message with 2 threads, and would like to know > if this is repeatable. > > Those concerned bout speed, set > #define TRACE 0 > in H/cs_par_base.h and also remove the call to cs_dag_print in > Engine/cs_par_dispatch.c function csp_dag_optimization > > At present I am setting various traces and not sure what value arrives > in CVS. > > And,... it is the elapsed time that matters, not CPU as the code uses > spinlocks which cost CPU but do not cause a process swap > > ==John ffitch > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by > > Make an app they can't live without > Enter the BlackBerry Developer Challenge > http://p.sf.net/sfu/RIM-dev2dev > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > -- Michael Gogins Irreducible Productions http://www.michael-gogins.com Michael dot Gogins at gmail dot com ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2010-08-17 17:42
From	Victor Lazzarini
Subject	Re: [Cs-dev] ParCS and request
	Here it is: coltrane:csound5 victor$ ./csound -n examples/trapped.csd -j 2 time resolution is 1000.000 ns PortMIDI real time MIDI plugin for Csound PortAudio real-time audio module for Csound 0dBFS level = 32768.0 Csound version 5.12 beta (float samples) Aug 17 2010 libsndfile-1.0.21 UnifiedCSD: examples/trapped.csd STARTING FILE Creating options Creating orchestra Creating score orchname: /var/folders/5P/5PSrqM3mGnCIxRlBhio0cE+++TM/-Tmp-//csound- m0V167.orc scorename: /var/folders/5P/5PSrqM3mGnCIxRlBhio0cE+++TM/-Tmp-//csound- pZclcl.sco rtaudio: PortAudio module enabled ... using callback interface rtmidi: PortMIDI module enabled ******************** * USING NEW PARSER * ******************** Add global read, write lists without any instruments Add global read, write lists without any instruments MINUS!!: make_num: .7 make_num: .3 make_num: .6 make_num: .4 make_num: .9 make_num: 1.4 make_num: .9 make_num: .1 make_num: .009 make_num: .2 make_num: .7 make_num: .3 make_num: 1.2 make_num: .7 make_num: .5 make_num: .333 make_num: .007 make_num: .3 make_num: .99 make_num: 3.1 make_num: .01 make_num: 3.1 make_num: .97 make_num: .6 make_num: .001 make_num: .8 make_num: .2 make_num: .001 make_num: .005 make_num: .5 make_num: .4 make_num: .6 make_num: .02 make_num: .99 make_num: .4 make_num: .001 make_num: .5 make_num: .1 make_num: .6 make_num: .2 make_num: .97 make_num: .2 make_num: .001 make_num: .003 make_num: .8 make_num: .91 make_num: .09 make_num: 1.4 make_num: .2 make_num: .001 make_num: .6 make_num: .2 make_num: .8 make_num: .1 make_num: .98 make_num: .1 make_num: .001 make_num: .001 make_num: 1.3 make_num: .5 make_num: .5 make_num: .005 make_num: .71 make_num: .015 make_num: .29 make_num: .01 make_num: .2 MINUS!!: make_num: .001 make_num: .01 make_num: .99 make_num: .001 make_num: .5 make_num: .5 make_num: .7 make_num: .3 make_num: .333 make_num: .333 MINUS!!: make_num: .5 make_num: .05 make_num: .1 make_num: .8 make_num: .2 make_num: .4 make_num: .9 make_num: .96 make_num: .1 make_num: .2 make_num: .76 make_num: .001 make_num: .4 make_num: .99 make_num: .6 make_num: .0001 make_num: .009 make_num: .2 make_num: .25 make_num: .4 make_num: 1.4 make_num: 2.1 make_num: .2 make_num: .1 make_num: .3 make_num: .6 make_num: 11.3 make_num: .1 make_num: .001 make_num: .001 make_num: .999 make_num: .001 make_num: .8 make_num: .6 make_num: .7 make_num: .8 make_num: .2 make_num: .3 make_num: .03 make_num: .2 make_num: .1 make_num: .1 make_num: 1.001 make_num: .7 make_num: .3 MINUS!!: make_num: .01 make_num: .05 make_num: .5 make_num: .5 make_num: .4 make_num: .5 make_num: .2 make_num: .93 make_num: .9 make_num: .8 make_num: 1.4 make_num: .2 make_num: 1.4 make_num: .97 make_num: .9 make_num: .15 make_num: .5 make_num: .7 make_num: .3 make_num: .3 make_num: .7 make_num: .1 make_num: .98 make_num: .91 make_num: .2 make_num: 1.2 make_num: .96 make_num: .3 make_num: .9 make_num: 1.3 make_num: .1 make_num: .8 make_num: .2 make_num: .91 make_num: .004 make_num: .3 make_num: .85 make_num: .006 make_num: .5 make_num: .95 make_num: .009 make_num: .8 make_num: .1 make_num: .98 make_num: .91 make_num: .2 make_num: 1.2 make_num: .96 make_num: .3 make_num: .9 make_num: 1.3 make_num: .1 make_num: .8 make_num: .2 make_num: .91 make_num: .004 make_num: .3 make_num: .85 make_num: .006 make_num: .5 make_num: .95 make_num: .009 make_num: .8 make_num: .5 make_num: .5 make_num: .72 make_num: .28 make_num: .333 make_num: .333 make_num: .5 make_num: .05 make_num: .1 make_num: .4 make_num: .9 make_num: .1 make_num: .24 make_num: .76 make_num: .5 make_num: .46 make_num: .4 make_num: .03 make_num: .3 make_num: .7 make_num: .3 make_num: .3 make_num: .7 make_num: .25 make_num: .25 make_num: .5 make_num: .02 make_num: .1 make_num: .4 make_num: .6 make_num: .4 make_num: .1 make_num: .9 make_num: .2 make_num: .3 make_num: .8 make_num: .01 make_num: .3 make_num: .8 make_num: .2 make_num: .7 make_num: .3 make_num: .9 make_num: .001 make_num: .1 make_num: .1 make_num: .5 make_num: .3 make_num: .8 make_num: .5 make_num: .001 make_num: .8 make_num: .98 make_num: .3 make_num: .7 make_num: .3 make_num: .6 make_num: .995 make_num: .4 make_num: .9 make_num: .1 make_num: .9 make_num: .1 make_num: .009 make_num: .2 make_num: .985 make_num: .7 make_num: .3 make_num: 1.2 make_num: .7 make_num: .5 make_num: .333 make_num: .007 make_num: .5 make_num: .8 make_num: .2 make_num: .97 make_num: .6 make_num: .001 make_num: .99 make_num: .97 make_num: .01 make_num: .97 make_num: .005 make_num: .8 make_num: .002 make_num: .91 make_num: .99 make_num: .09 make_num: .99 make_num: .99 make_num: .4 make_num: .001 make_num: .5 make_num: .1 make_num: .6 make_num: .2 make_num: .97 make_num: .2 make_num: .001 make_num: .003 make_num: .9 make_num: .98 make_num: .81 make_num: .003 make_num: .19 make_num: .003 make_num: 1.4 make_num: .2 make_num: .001 make_num: .6 make_num: .2 make_num: .8 make_num: .1 make_num: .98 make_num: .1 make_num: .001 make_num: .001 make_num: .1 make_num: .08 make_num: .5 make_num: .5 MINUS!!: make_num: 2.1 MINUS!!: Semantic Analysis Instr: 1 read: { } write: { } read_write: { } weight: 1 Instr: 2 read: { } write: { } read_write: { garvb } weight: 1 Instr: 3 read: { } write: { } read_write: { garvb } weight: 1 Instr: 4 read: { } write: { } read_write: { garvb } weight: 1 Instr: 5 read: { } write: { } read_write: { garvb } weight: 1 Instr: 6 read: { } write: { } read_write: { garvb } weight: 1 Instr: 7 read: { } write: { } read_write: { garvb } weight: 1 Instr: 8 read: { } write: { } read_write: { garvb } weight: 1 Instr: 9 read: { } write: { } read_write: { garvb, gadel } weight: 1 Instr: 10 read: { } write: { } read_write: { garvb } weight: 1 Instr: 11 read: { } write: { } read_write: { garvb, gadel } weight: 1 Instr: 12 read: { } write: { } read_write: { garvb } weight: 1 Instr: 13 read: { } write: { } read_write: { } weight: 1 Instr: 98 read: { gadel } write: { gadel } read_write: { } weight: 1 Instr: 99 read: { garvb } write: { garvb } read_write: { } weight: 1 Semantic Analysis Ends Parsing successful! Verifying AST (NEED TO IMPLEMENT) Inserting Parallelism Constructs into AST Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] Inserting Parallelism Constructs into AST [End Inserting Parallelism Constructs into AST] [End Inserting Parallelism Constructs into AST] make_num: .5 make_num: .5 make_num: .5 make_num: .5 make_num: .5 make_num: .5 make_num: .5 make_num: .5 make_num: .5 make_num: .5 make_num: .5 Calculating Instrument weights from AST Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] Calculating Instrument weight from AST WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 WARNING: Unexpected node type in weight calculation walk 277 [End Calculating Instrument weight from AST] [End Calculating Instrument weights from AST] Unknown TREE node of type 289 found in root. Unknown TREE node of type 289 found in root. create_instrument: instr num 1 Starting to install instruments create_instrument: instr num 2 Starting to install instruments create_instrument: instr num 3 Starting to install instruments create_instrument: instr num 4 Starting to install instruments create_instrument: instr num 5 Starting to install instruments create_instrument: instr num 6 Starting to install instruments create_instrument: instr num 7 Starting to install instruments create_instrument: instr num 8 Starting to install instruments create_instrument: instr num 9 Starting to install instruments create_instrument: instr num 10 Starting to install instruments create_instrument: instr num 11 Starting to install instruments create_instrument: instr num 12 Starting to install instruments create_instrument: instr num 13 Starting to install instruments create_instrument: instr num 98 Starting to install instruments create_instrument: instr num 99 Starting to install instruments Elapsed time at end of orchestra compile: real: 0.045s, CPU: 0.021s sorting score ... ... done Elapsed time at end of score sort: real: 0.052s, CPU: 0.022s Multithread performance: insno: -1 thread 0 of 2 starting. InstrWeightInfo: [Default] weight_min: 0 weight_max: 0 roots_avail_min: 0 roots_avail_max: 0 Csound version 5.12 beta (float samples) Aug 17 2010 displays suppressed 0dBFS level = 32768.0 orch now loaded audio buffered in 1024 sample-frame blocks not writing to sound disk SECTION 1: ftable 1: ftable 2: ftable 3: ftable 4: ftable 5: ftable 6: ftable 7: ftable 8: ftable 9: ftable 10: ftable 11: ftable 12: ftable 13: ftable 14: ftable 15: ftable 16: ftable 17: ftable 18: ftable 19: ftable 20: ftable 21: ftable 22: new alloc for instr 1: Dag2: 1 [0x3b36d0] roots: 1 [0x3b36d0] root_seen: 2 remaining: 0 remaining: 1 first_root: 0 table: 0 B 0.000 .. 1.000 T 0.998 TT 0.998 M: 33.3 0.0 new alloc for instr 1: Dag2: 1 [0x3b3b30] 1 [0x3b3b80] roots: 1 [0x3b3b30] 1 [0x3b3b80] root_seen: 2 1 remaining: 0 0 remaining: 2 first_root: 0 table: 0 0 0 0 B 1.000 .. 3.600 T 3.605 TT 3.605 M: 205.4 0.1 new alloc for instr 1: Dag2: 1 [0x3b3f70] 1 [0x3b3ff0] 1 [0x3b3fc0] roots: 1 [0x3b3f70] 1 [0x3b3ff0] 1 [0x3b3fc0] root_seen: 2 1 remaining: 0 0 remaining: 2 first_root: 0 table: 0 0 0 0 B 3.600 .. 6.200 T 6.190 TT 6.190 M: 523.6 1.2 new alloc for instr 1: Dag2: 1 [0x3b44a0] 1 [0x3b4520] 1 [0x3b44f0] 1 [0x3b4550] roots: 1 [0x3b44a0] 1 [0x3b4520] 1 [0x3b44f0] 1 [0x3b4550] root_seen: 2 1 remaining: 0 0 remaining: 2 first_root: 0 table: 0 0 0 0 B 6.200 .. 8.400 T 8.390 TT 8.390 M: 961.4 9.1 new alloc for instr 1: Dag2: 1 [0x3b4a00] 1 [0x3b4a80] 1 [0x3b4ae0] 1 [0x3b4a50] 1 [0x3b4ab0] roots: 1 [0x3b4a00] 1 [0x3b4a80] 1 [0x3b4ae0] 1 [0x3b4a50] 1 [0x3b4ab0] root_seen: 2 1 remaining: 0 0 remaining: 2 first_root: 0 table: 0 0 0 0 B 8.400 .. 22.000 T 21.995 TT 21.995 M: 3120.7 2668.0 new alloc for instr 2: new alloc for instr 99: Dag2: 1 [0x3b4d30] 1 [0x3b4e20] 1 [0x3b4df0] 2 [0x3b4e50] 99 [0x3b4e80] roots: 1 [0x3b4d30] 1 [0x3b4e20] 1 [0x3b4df0] 2 [0x3b4e50] root_seen: 2 1 0 remaining: 0 0 1 remaining: 3 first_root: 0 table: 0 0 0 0 0 1 0 0 0 B 22.000 .. 22.130 T 22.132 TT 22.132 M: 815.9 810.0 new alloc for instr 2: Dag2: 1 [0x3b51b0] 1 [0x3b52a0] 2 [0x3b5300] 1 [0x3b5270] 2 [0x3b52d0] 99 [0x3b5360] roots: 1 [0x3b51b0] 1 [0x3b52a0] 2 [0x3b5300] 1 [0x3b5270] 2 [0x3b52d0] root_seen: 2 1 0 remaining: 0 0 2 remaining: 3 first_root: 0 table: 0 0 1 2 0 1 0 0 0 Dag2: 1 [0x3b51b0] 1 [0x3b52a0] 2 [0x3b5300] 1 [0x3b5270] 2 [0x3b52d0] 99 [0x3b5360] roots: root_seen: 2 2 0 remaining: 0 0 1 remaining: 1 first_root: -1 table: 0 0 1 2 0 1 0 0 0 Expected a root to perform. Found none (1 remaining) Csound tidy up: Segmentation fault Csound tidy up: Bus error Csound tidy upCsound tidy up: Bus error On 17 Aug 2010, at 17:28, John ff wrote: > I have moved a little code around and optimised a little. > > The request is to try trapped with more than one thread. I am getting > an "impossible" error message with 2 threads, and would like to know > if this is repeatable. > > Those concerned bout speed, set > #define TRACE 0 > in H/cs_par_base.h and also remove the call to cs_dag_print in > Engine/cs_par_dispatch.c function csp_dag_optimization > > At present I am setting various traces and not sure what value arrives > in CVS. > > And,... it is the elapsed time that matters, not CPU as the code uses > spinlocks which cost CPU but do not cause a process swap > > ==John ffitch > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by > > Make an app they can't live without > Enter the BlackBerry Developer Challenge > http://p.sf.net/sfu/RIM-dev2dev > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2010-08-17 17:48
From	jpff@cs.bath.ac.uk
Subject	Re: [Cs-dev] ParCS and request
	> Here it is: > Thanks; as I feared...... Need to understnad a lot more. PS What can I learn from your machine being called coltrane and mine being called xenakis? ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2010-08-17 18:09
From	Victor Lazzarini
Subject	Re: [Cs-dev] ParCS and request
	Maybe that we share some musical appreciation? On 17 Aug 2010, at 17:48, jpff@cs.bath.ac.uk wrote: >> Here it is: >> > > Thanks; as I feared...... > > Need to understnad a lot more. > > PS What can I learn from your machine being called coltrane and mine > being > called xenakis? > > > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by > > Make an app they can't live without > Enter the BlackBerry Developer Challenge > http://p.sf.net/sfu/RIM-dev2dev > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2010-08-17 18:19
From	Victor Lazzarini
Subject	Re: [Cs-dev] ParCS and request
	But in any case, for the first time today I saw a gain in comparison to ordinary Csound. So this is going the right way. Victor On 17 Aug 2010, at 17:48, jpff@cs.bath.ac.uk wrote: >> Here it is: >> > > Thanks; as I feared...... > > Need to understnad a lot more. > > PS What can I learn from your machine being called coltrane and mine > being > called xenakis? > > > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by > > Make an app they can't live without > Enter the BlackBerry Developer Challenge > http://p.sf.net/sfu/RIM-dev2dev > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2010-08-17 21:38
From	Michael Gogins
Subject	Re: [Cs-dev] ParCS and request
Attachments	None None
	Does anyone know how much parallel speedups are limited by irreducible non-parallel code and how much by irreducible threading overhead? MKG from cell phone On Aug 17, 2010 1:14 PM, "Victor Lazzarini" <Victor.Lazzarini@nuim.ie> wrote: But in any case, for the first time today I saw a gain in comparison to ordinary Csound. So this is going the right way. Victor On 17 Aug 2010, at 17:48, jpff@cs.bath.ac.uk wrote: >> Here it is: >> > > Thanks; as I feared...... > > Need to understnad a lot more. > > PS What can I...

Date	2010-08-17 21:49
From	Victor Lazzarini
Subject	Re: [Cs-dev] ParCS and request
Attachments	None None
	In my example, I'm using an obviously parallelisable example, so the slowdown must be caused by threading overhead. Interestingly, as I am using 10 instances, 5 threads are slightly faster than 4, I guess because of evenly spread load balance. Victor On 17 Aug 2010, at 21:38, Michael Gogins wrote: Does anyone know how much parallel speedups are limited by irreducible non-parallel code and how much by irreducible threading overhead? MKG from cell phone On Aug 17, 2010 1:14 PM, "Victor Lazzarini" <Victor.Lazzarini@nuim.ie> wrote: But in any case, for the first time today I saw a gain in comparison to ordinary Csound. So this is going the right way. Victor On 17 Aug 2010, at 17:48, jpff@cs.bath.ac.uk wrote: >> Here it is: >> > > Thanks; as I feared...... > > Need to understnad a lot more. > > PS What can I... ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/csound-devel