Re: [Haskell-cafe] ANNOUNCE: taffybar: an alternative status bar for xmonad
On Sat, Aug 13, 2011 at 09:54:13PM -0700, Joel Burget wrote: This sounds really intriguing. Since I'm temporarily not using xmonad, and I'm sure others would like to see as well, could we get a screenshot? Oops, how could I forget. http://pages.cs.wisc.edu/~travitch/taffybar.jpg I have the xmonad log on the left, a CPU graph, memory graph, date/time, weather, and then system tray visible there. signature.asc Description: Digital signature ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] Invitation to connect on LinkedIn
LinkedIn I'd like to add you to my professional network on LinkedIn. - Andrew Andrew Smith B.Sc(Hons),MBA Founder and CEO at VTRL - Value Technology Research Ltd Edinburgh, United Kingdom Confirm that you know Andrew Smith B.Sc(Hons),MBA https://www.linkedin.com/e/uc6lxc-grc8ge3y-3d/isd/3853824536/atjCCKYr/ -- (c) 2011, LinkedIn Corporation___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Invitation to connect on LinkedIn
lol. I don't know Andrew Smith. How about y'all? On Aug 14, 2011, at 12:32 PM, Andrew Smith B.Sc(Hons),MBA wrote: LinkedIn I'd like to add you to my professional network on LinkedIn. - Andrew Andrew Smith B.Sc(Hons),MBA Founder and CEO at VTRL - Value Technology Research Ltd Edinburgh, United Kingdom Confirm that you know Andrew © 2011, LinkedIn Corporation ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Invitation to connect on LinkedIn
That's a clever way to build your resume. :)) Someone looking at his numerous contacts with haskell community may think Wow, this guys is some sort of haskell guru. Better bring him in for our stock market division. On Sunday, August 14, 2011 01:02:42 PM Daniel Patterson wrote: lol. I don't know Andrew Smith. How about y'all? On Aug 14, 2011, at 12:32 PM, Andrew Smith B.Sc(Hons),MBA wrote: LinkedIn I'd like to add you to my professional network on LinkedIn. - Andrew Andrew Smith B.Sc(Hons),MBA Founder and CEO at VTRL - Value Technology Research Ltd Edinburgh, United Kingdom Confirm that you know Andrew © 2011, LinkedIn Corporation ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Invitation to connect on LinkedIn
On Sun, Aug 14, 2011 at 13:09, vagif.ve...@gmail.com wrote: That's a clever way to build your resume. :)) No, just LinkedIn being stupid; the sign-up form is designed to spam your contacts, and you actually have to take some care to avoid it. -- brandon s allbery allber...@gmail.com wandering unix systems administrator (available) (412) 475-9364 vm/sms ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Invitation to connect on LinkedIn
Not so sure; his company's website is under construction for more than a year and after brief google'ing I still don't understand even what kind of business are they supposed to be in. Seems more likely that it's actually Andrew who does the spamming. Отправлено с iPad 14.08.2011, в 21:18, Brandon Allbery allber...@gmail.com написал(а): On Sun, Aug 14, 2011 at 13:09, vagif.ve...@gmail.com wrote: That's a clever way to build your resume. :)) No, just LinkedIn being stupid; the sign-up form is designed to spam your contacts, and you actually have to take some care to avoid it. -- brandon s allbery allber...@gmail.com wandering unix systems administrator (available) (412) 475-9364 vm/sms ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Proposal #3339: Add (+) as a synonym for mappend
On Sun, Aug 14, 2011 at 13:38, Yitzchak Gale g...@sefer.org wrote: Brandon Allbery wrote: Umm, I think the semigroups package will break everything that creates Monoid instances anyway. It has never broken anything for me. What do you mean? Anything useful has to be modified to depend on SemiGroup as well to get mconcat or its replacement; that's why you jumped the proposal to begin with As others have noted, this is a rather intrusive change to the Haskell ecosystem. *If* it has in fact been decided to go ahead with it, *then* this proposal should be merged with yours since the best time to do it is when mconcat is being relocated anyway. I don't think we've made it past the if, especially given the comments so far. -- brandon s allbery allber...@gmail.com wandering unix systems administrator (available) (412) 475-9364 vm/sms ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Proposal #3339: Add (+) as a synonym for mappend
Brandon Allbery wrote: Anything useful has to be modified to depend on SemiGroup as well to get mconcat or its replacement; that's why you jumped the proposal to begin with Not at all. Types with Monoid instances need an additional instance, a Semgroup instance, in order to be able to use '' instead of mappend. mconcat is not involved in this discussion. That is the current situation. I am advocating leaving it that way. As others have noted, this is a rather intrusive change to the Haskell ecosystem. Exporting from Data.Monoid is the intrusive change. I am strongly against it. In the long run, it will create ugliness and inconvenience to our class system, which has enough problems as it is. I advocate leaving things as they are. If individual library authors wish to add a Semigroup instance to their Monoid instances as a convenience, which is harmless, I think that would be wonderful. But that is a separate issue. Thanks, Yitz ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] how to read CPU time vs wall time report from GHC?
Hi guys, I'm new in parallel programming with Haskell. I made a simple test program using that par combinator etc, and was a bit unhappy that it turns out to be slower than its sequential version. But firstly, I dont fully understand how to read the runtime report produced by GHC with -s option: SPARKS: 5 (5 converted, 0 pruned) INIT time0.02s ( 0.01s elapsed) MUT time3.46s ( 0.89s elapsed) GCtime5.49s ( 1.46s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time8.97s ( 2.36s elapsed) As I understand it from the documentation, the left time-column is the CPU time, whereas the right one is elapses wall time. But how come that the wall time is less than the CPU time? Isn't wall time = user's perspective of time; so that is CPU time + IO + etc? Any help? --Wish. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Proposal #3339: Add (+) as a synonym for mappend
On Sun, 2011-08-14 at 21:05 +0300, Yitzchak Gale wrote: Brandon Allbery wrote: Anything useful has to be modified to depend on SemiGroup as well to get mconcat or its replacement; that's why you jumped the proposal to begin with Not at all. Types with Monoid instances need an additional instance, a Semgroup instance That does require depending on semigroups though, and I think that's what Brandon was saying. Of course, the obvious solution to this would be to promote semigroups, e.g., by adding it to the Haskell Platform or including it in base... but the current semigroups package is a bit heavyweight for that; it exports four new modules for what is really a very simple concept! -- Chris Smith ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
Hello Wishnu, That is slightly odd. What CPU and operating system are you running on? Include Kernel versions if Linux. Cheers, Edward Excerpts from Wishnu Prasetya's message of Sun Aug 14 14:11:36 -0400 2011: Hi guys, I'm new in parallel programming with Haskell. I made a simple test program using that par combinator etc, and was a bit unhappy that it turns out to be slower than its sequential version. But firstly, I dont fully understand how to read the runtime report produced by GHC with -s option: SPARKS: 5 (5 converted, 0 pruned) INIT time0.02s ( 0.01s elapsed) MUT time3.46s ( 0.89s elapsed) GCtime5.49s ( 1.46s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time8.97s ( 2.36s elapsed) As I understand it from the documentation, the left time-column is the CPU time, whereas the right one is elapses wall time. But how come that the wall time is less than the CPU time? Isn't wall time = user's perspective of time; so that is CPU time + IO + etc? Any help? --Wish. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
On Sun, Aug 14, 2011 at 08:11:36PM +0200, Wishnu Prasetya wrote: Hi guys, I'm new in parallel programming with Haskell. I made a simple test program using that par combinator etc, and was a bit unhappy that it turns out to be slower than its sequential version. But firstly, I dont fully understand how to read the runtime report produced by GHC with -s option: SPARKS: 5 (5 converted, 0 pruned) INIT time0.02s ( 0.01s elapsed) MUT time3.46s ( 0.89s elapsed) GCtime5.49s ( 1.46s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time8.97s ( 2.36s elapsed) As I understand it from the documentation, the left time-column is the CPU time, whereas the right one is elapses wall time. But how come that the wall time is less than the CPU time? Isn't wall time = user's perspective of time; so that is CPU time + IO + etc? Yes, but if you have multiple CPUs, then CPU time accumulates faster than wall-clock time. Based on the above example, I guess you have or you run the program on 4 cores (2.36 * 4 = 9.44, which means you got a very nice ~95% efficiency). regards, iustin ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
Hello Edward, I'm using Windows 7 on Intel i7 (4 cores with hyperthreading)... --Wish. Hello Wishnu, That is slightly odd. What CPU and operating system are you running on? Include Kernel versions if Linux. Cheers, Edward Excerpts from Wishnu Prasetya's message of Sun Aug 14 14:11:36 -0400 2011: Hi guys, I'm new in parallel programming with Haskell. I made a simple test program using that par combinator etc, and was a bit unhappy that it turns out to be slower than its sequential version. But firstly, I dont fully understand how to read the runtime report produced by GHC with -s option: SPARKS: 5 (5 converted, 0 pruned) INIT time0.02s ( 0.01s elapsed) MUT time3.46s ( 0.89s elapsed) GCtime5.49s ( 1.46s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time8.97s ( 2.36s elapsed) As I understand it from the documentation, the left time-column is the CPU time, whereas the right one is elapses wall time. But how come that the wall time is less than the CPU time? Isn't wall time = user's perspective of time; so that is CPU time + IO + etc? Any help? --Wish. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
Ah, good catch. :-) Edward Excerpts from Iustin Pop's message of Sun Aug 14 14:25:02 -0400 2011: On Sun, Aug 14, 2011 at 08:11:36PM +0200, Wishnu Prasetya wrote: Hi guys, I'm new in parallel programming with Haskell. I made a simple test program using that par combinator etc, and was a bit unhappy that it turns out to be slower than its sequential version. But firstly, I dont fully understand how to read the runtime report produced by GHC with -s option: SPARKS: 5 (5 converted, 0 pruned) INIT time0.02s ( 0.01s elapsed) MUT time3.46s ( 0.89s elapsed) GCtime5.49s ( 1.46s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time8.97s ( 2.36s elapsed) As I understand it from the documentation, the left time-column is the CPU time, whereas the right one is elapses wall time. But how come that the wall time is less than the CPU time? Isn't wall time = user's perspective of time; so that is CPU time + IO + etc? Yes, but if you have multiple CPUs, then CPU time accumulates faster than wall-clock time. Based on the above example, I guess you have or you run the program on 4 cores (2.36 * 4 = 9.44, which means you got a very nice ~95% efficiency). regards, iustin ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
On 14-8-2011 20:25, Iustin Pop wrote: On Sun, Aug 14, 2011 at 08:11:36PM +0200, Wishnu Prasetya wrote: Hi guys, I'm new in parallel programming with Haskell. I made a simple test program using that par combinator etc, and was a bit unhappy that it turns out to be slower than its sequential version. But firstly, I dont fully understand how to read the runtime report produced by GHC with -s option: SPARKS: 5 (5 converted, 0 pruned) INIT time0.02s ( 0.01s elapsed) MUT time3.46s ( 0.89s elapsed) GCtime5.49s ( 1.46s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time8.97s ( 2.36s elapsed) As I understand it from the documentation, the left time-column is the CPU time, whereas the right one is elapses wall time. But how come that the wall time is less than the CPU time? Isn't wall time = user's perspective of time; so that is CPU time + IO + etc? Yes, but if you have multiple CPUs, then CPU time accumulates faster than wall-clock time. Based on the above example, I guess you have or you run the program on 4 cores (2.36 * 4 = 9.44, which means you got a very nice ~95% efficiency). regards, iustin That makes sense... But are you sure thats how i should read this? I dont want to jump happy too early. --Wish. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Haskell Actors, Linda, publish / subscribe models?
On Sat, Aug 13, 2011 at 3:54 PM, dokondr doko...@gmail.com wrote: Hi, I am trying to figure out what Haskell libraries can be used to build publish / subscribe communication between threads running both in the same and different address spaces on the net. For my needs any of these models will work: - Actors [ http://en.wikipedia.org/wiki/Actor_model ] - Linda tuple space [ http://en.wikipedia.org/wiki/Linda_%28coordination_language%29 ] - Publish / subscribe [ http://en.wikipedia.org/wiki/Java_Message_Service#Publish.2Fsubscribe_model] I need to build a framework to coordinate task producers / consumers distributed in the same and different address spaces. I need to scale a data processing application somewhat Hadoop-like way yet in more flexible manner, without Hadoop-specific distributed FS constraints. Looking through Applications and libraries/Concurrency and parallelism: http://www.haskell.org/haskellwiki/Applications_and_libraries/Concurrency_and_parallelism I found Haskell actor package [ http://hackage.haskell.org/cgi-bin/hackage-scripts/package/actor ] that fails to build with ghc 7.0. Please advise on latest working libraries. Thanks! Have anybody used Haskell actor package [ http://hackage.haskell.org/cgi-bin/hackage-scripts/package/actor ]? ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
On Sun, Aug 14, 2011 at 08:32:36PM +0200, Wishnu Prasetya wrote: On 14-8-2011 20:25, Iustin Pop wrote: On Sun, Aug 14, 2011 at 08:11:36PM +0200, Wishnu Prasetya wrote: Hi guys, I'm new in parallel programming with Haskell. I made a simple test program using that par combinator etc, and was a bit unhappy that it turns out to be slower than its sequential version. But firstly, I dont fully understand how to read the runtime report produced by GHC with -s option: SPARKS: 5 (5 converted, 0 pruned) INIT time0.02s ( 0.01s elapsed) MUT time3.46s ( 0.89s elapsed) GCtime5.49s ( 1.46s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time8.97s ( 2.36s elapsed) As I understand it from the documentation, the left time-column is the CPU time, whereas the right one is elapses wall time. But how come that the wall time is less than the CPU time? Isn't wall time = user's perspective of time; so that is CPU time + IO + etc? Yes, but if you have multiple CPUs, then CPU time accumulates faster than wall-clock time. Based on the above example, I guess you have or you run the program on 4 cores (2.36 * 4 = 9.44, which means you got a very nice ~95% efficiency). regards, iustin That makes sense... But are you sure thats how i should read this? As far as I know, this is correct. I dont want to jump happy too early. Well, you algorithm does work in parallel, but if you look at the GC/MUT time, ~60% of the total runtime is spent in GC, so you have a space leak or an otherwise inefficient algorithm. The final speedup is just 3.46s/2.36s, i.e. 1.46x instead of ~4x, so you still have some work to do to make this better. At least, this is how I read those numbers. regards, iustin ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Invitation to connect on LinkedIn
Andrew Smith B.Sc(Hons),MBA asmith9...@gmail.com writes: I'd like to add you to my professional network on LinkedIn. Since LinkedIn tends to spam even more ambitiously than the other social network sites, I have a procmail rule sending any mail from Linkedin to /dev/null. But it doesn't work when it's sent via other mailing lists, when some person of dubious faculty and questionable morals wants to add an entire list to his professional network... -k -- If I haven't seen further, it is by standing in the footprints of giants ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
On Sunday 14 August 2011, 21:53:21, Iustin Pop wrote: On Sun, Aug 14, 2011 at 08:32:36PM +0200, Wishnu Prasetya wrote: On 14-8-2011 20:25, Iustin Pop wrote: On Sun, Aug 14, 2011 at 08:11:36PM +0200, Wishnu Prasetya wrote: Hi guys, I'm new in parallel programming with Haskell. I made a simple test program using that par combinator etc, and was a bit unhappy that it turns out to be slower than its sequential version. But firstly, I dont fully understand how to read the runtime report produced by GHC with -s option: SPARKS: 5 (5 converted, 0 pruned) INIT time0.02s ( 0.01s elapsed) MUT time3.46s ( 0.89s elapsed) GCtime5.49s ( 1.46s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time8.97s ( 2.36s elapsed) As I understand it from the documentation, the left time-column is the CPU time, whereas the right one is elapses wall time. But how come that the wall time is less than the CPU time? Isn't wall time = user's perspective of time; so that is CPU time + IO + etc? Yes, but if you have multiple CPUs, then CPU time accumulates faster than wall-clock time. Based on the above example, I guess you have or you run the program on 4 cores (2.36 * 4 = 9.44, which means you got a very nice ~95% efficiency). regards, iustin That makes sense... But are you sure thats how i should read this? As far as I know, this is correct. It is indeed. CPU time is the sum of CPU time for all threads, which is typically larger than elapsed time when several threads run in parallel. I dont want to jump happy too early. Well, you algorithm does work in parallel, but if you look at the GC/MUT time, ~60% of the total runtime is spent in GC, so you have a space leak or an otherwise inefficient algorithm. Not enough data to make more than guesses concerning the cause, but 60% GC definitely indicates a problem with the algorithm (resp. its implementation), The final speedup is just 3.46s/2.36s, i.e. 1.46x instead of ~4x, so you still have some work to do to make this better. We don't know the times for a non-threaded run (or an -N1 run), so it could be anything from a slowdown to a 4× speedup (but it's likely to be a speedup by a factor 4×). At least, this is how I read those numbers. regards, iustin ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Proposal #3339: Add (+) as a synonym for mappend
I originally didn't have the package exporting those things. I would be amenable to standardization without them, but I use them in about 20 packages that are built on top of semigroups, and naturals and non-empty lists come up when talking about semigroups a lot. Rather than having them live way up in extension land with the rest of my algebra libraries i moved them down to where they could do some good and admit some optimizations. On Sun, Aug 14, 2011 at 2:21 PM, Chris Smith cdsm...@gmail.com wrote: On Sun, 2011-08-14 at 21:05 +0300, Yitzchak Gale wrote: Brandon Allbery wrote: Anything useful has to be modified to depend on SemiGroup as well to get mconcat or its replacement; that's why you jumped the proposal to begin with Not at all. Types with Monoid instances need an additional instance, a Semgroup instance That does require depending on semigroups though, and I think that's what Brandon was saying. Of course, the obvious solution to this would be to promote semigroups, e.g., by adding it to the Haskell Platform or including it in base... but the current semigroups package is a bit heavyweight for that; it exports four new modules for what is really a very simple concept! -- Chris Smith ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Invitation to connect on LinkedIn
On Sun, Aug 14, 2011 at 16:15, Ketil Malde ke...@malde.org wrote: Andrew Smith B.Sc(Hons),MBA asmith9...@gmail.com writes: I'd like to add you to my professional network on LinkedIn. Since LinkedIn tends to spam even more ambitiously than the other social network sites, I have a procmail rule sending any mail from Linkedin to I should actually mention that this is a double infelicity: the OP has a gmail.com address. gmail has an annoying tendency to add every address you've ever sent mail to to your primary address book; so telling something like LinkedIn to send invites to your address book, with the reasonable expectation that that means your actual contacts, suddenly turns into spamming the (your) known universe. -- brandon s allbery allber...@gmail.com wandering unix systems administrator (available) (412) 475-9364 vm/sms ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
On 14-8-2011 22:17, Daniel Fischer wrote: On Sunday 14 August 2011, 21:53:21, Iustin Pop wrote: On Sun, Aug 14, 2011 at 08:32:36PM +0200, Wishnu Prasetya wrote: On 14-8-2011 20:25, Iustin Pop wrote: On Sun, Aug 14, 2011 at 08:11:36PM +0200, Wishnu Prasetya wrote: Hi guys, I'm new in parallel programming with Haskell. I made a simple test program using that par combinator etc, and was a bit unhappy that it turns out to be slower than its sequential version. But firstly, I dont fully understand how to read the runtime report produced by GHC with -s option: SPARKS: 5 (5 converted, 0 pruned) INIT time0.02s ( 0.01s elapsed) MUT time3.46s ( 0.89s elapsed) GCtime5.49s ( 1.46s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time8.97s ( 2.36s elapsed) As I understand it from the documentation, the left time-column is the CPU time, whereas the right one is elapses wall time. But how come that the wall time is less than the CPU time? Isn't wall time = user's perspective of time; so that is CPU time + IO + etc? Yes, but if you have multiple CPUs, then CPU time accumulates faster than wall-clock time. Based on the above example, I guess you have or you run the program on 4 cores (2.36 * 4 = 9.44, which means you got a very nice ~95% efficiency). regards, iustin That makes sense... But are you sure thats how i should read this? As far as I know, this is correct. It is indeed. CPU time is the sum of CPU time for all threads, which is typically larger than elapsed time when several threads run in parallel. I dont want to jump happy too early. Well, you algorithm does work in parallel, but if you look at the GC/MUT time, ~60% of the total runtime is spent in GC, so you have a space leak or an otherwise inefficient algorithm. Not enough data to make more than guesses concerning the cause, but 60% GC definitely indicates a problem with the algorithm (resp. its implementation), The final speedup is just 3.46s/2.36s, i.e. 1.46x instead of ~4x, so you still have some work to do to make this better. We don't know the times for a non-threaded run (or an -N1 run), so it could be anything from a slowdown to a 4× speedup (but it's likely to be a speedup by a factor 4×). Well, the -N1 is below. The sequential version of the program has almost the same profile: SPARKS: 5 (1 converted, 4 pruned) INIT time0.00s ( 0.00s elapsed) MUT time2.78s ( 2.99s elapsed) GCtime4.35s ( 4.15s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time7.13s ( 7.14s elapsed) Am I correct then to say that the speed up with respect to sequential is equal to: tot-elapse-time-N1 / tot-elapse-N4 ? So I have 7.14 / 2.36 = 3.0 speed up, and not 1.46 as Iustin said? I'll probably have to do something with that GC :) --Wish. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
On Sunday 14 August 2011, 22:42:13, Wishnu Prasetya wrote: On 14-8-2011 22:17, Daniel Fischer wrote: We don't know the times for a non-threaded run (or an -N1 run), so it could be anything from a slowdown to a 4× speedup (but it's likely to be a speedup by a factor 4×). Well, the -N1 is below. The sequential version of the program has almost the same profile: SPARKS: 5 (1 converted, 4 pruned) INIT time0.00s ( 0.00s elapsed) MUT time2.78s ( 2.99s elapsed) GCtime4.35s ( 4.15s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time7.13s ( 7.14s elapsed) Am I correct then to say that the speed up with respect to sequential is equal to: tot-elapse-time-N1 / tot-elapse-N4 ? So I have 7.14 / 2.36 = 3.0 speed up, and not 1.46 as Iustin said? Yes (with respect to wall-clock time of course). That's not too bad, though it should be possible to increase that factor. I'll probably have to do something with that GC :) But that should be the first target, there's probably some low-hanging fruit there. Maybe increasing the size of the allocation area (+RTS -Ax) or the heap (+RTS -Hx) would already do some good. Also do heap profiling to find out what most likely takes so much GC time (before compiling for profiling, running with +RTS -hT could produce useful information). ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
[Haskell-cafe] Evolving faster Haskell programs - where are the files?
I just read the enlightening article by Don Stewart[1], and wanted to test this method on one of my own programs, but it seems that there some of the attached files are missing, namely [2], [3] and [4]. Is there anybody who knows, where I can get those files? I don't really want to rewrite those file by hand. Yours, Robert Clausecker [1]: http://donsbot.wordpress.com/2009/03/09/evolving-faster-haskell-programs [2]: http://galois.com/~dons/acovea/gcc43_core2.acovea [3]: http://galois.com/~dons/acovea/ghc-basic.acovea [4]: http://galois.com/~dons/acovea/ghc-6_10.acovea ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] Evolving faster Haskell programs - where are the files?
I mentioned this to him on Twitter a while ago. Presumably it has to do with the fact that he's no longer at Galois. Another unfortunate fact is that ACOVEA is at this point unmaintained, and that is why the official homepage for it was removed. When I emailed the author, he said he couldn't afford to spend the time on it it was requiring. :( I've wondered if it would be worth it to make a pure-Haskell replacement specifically devoted to improving the performance of Haskell programs, and not only optimizing a single objective but presenting an efficient frontier between space usage and execution time or whatever other criteria it makes sense to optimize for.. Warren On Sun, Aug 14, 2011 at 2:13 PM, Robert Clausecker fuz...@gmail.com wrote: I just read the enlightening article by Don Stewart[1], and wanted to test this method on one of my own programs, but it seems that there some of the attached files are missing, namely [2], [3] and [4]. Is there anybody who knows, where I can get those files? I don't really want to rewrite those file by hand. Yours, Robert Clausecker [1]: http://donsbot.wordpress.com/2009/03/09/evolving-faster-haskell-programs [2]: http://galois.com/~dons/acovea/gcc43_core2.acovea [3]: http://galois.com/~dons/acovea/ghc-basic.acovea [4]: http://galois.com/~dons/acovea/ghc-6_10.acovea ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] ANNOUNCE: taffybar: an alternative status bar for xmonad
On 13 August 2011 08:56, Tristan Ravitch travi...@cs.wisc.edu wrote: I've wanted a slightly fancier status bar than xmobar for a while, so I finally made one. It uses gtk2hs and dbus extensively, so if you hate either of those things it probably isn't for you. Being written in gtk, though, it can have more graphical widgets. http://hackage.haskell.org/package/taffybar Current feature highlights: * It has a system tray * Generic graph widget for things like CPU/memory * XMonad log over DBus so it can be restarted independently of xmonad * Graphical battery widget There is still a lot that I want to add but I figured getting some feedback early would be handy. Documentation is currently at http://pages.cs.wisc.edu/~travitch/taffybar until I figure out how to appease Hackage (see the System.Taffybar module). ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe I apologize if I'm missing something obvious here, but when I try to run taffybar I get Launching custom binary /home/alex/.cache/taffybar/taffybar-linux-i386 taffybar-linux-i386: ConnectionError connectSession: DBUS_SESSION_BUS_ADDRESS is missing or invalid. Is there some D-BUS configuration that needs to happen before the package is usable? Thanks, Alex ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] ANNOUNCE: taffybar: an alternative status bar for xmonad
On Sun, Aug 14, 2011 at 02:45:24PM -0700, Alexander Dunlap wrote: I apologize if I'm missing something obvious here, but when I try to run taffybar I get Launching custom binary /home/alex/.cache/taffybar/taffybar-linux-i386 taffybar-linux-i386: ConnectionError connectSession: DBUS_SESSION_BUS_ADDRESS is missing or invalid. Is there some D-BUS configuration that needs to happen before the package is usable? Sorry, I assumed you would have dbus running already. If you add a line like: eval `dbus-launch --auto-syntax` early in your ~/.xsession (if logging in via some graphical login manager) or ~/.xinitrc (if starting X via startx). That should work for any normal -sh-style or -csh-style shell. That command starts DBus and sets the DBUS_SESSION* environment variables, and both xmonad and taffybar need to have the same settings for that variable, so make sure you execute that command before starting either of them. I'll add some notes in the documentation about this. pgpvc8frKIAwI.pgp Description: PGP signature ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] how to read CPU time vs wall time report from GHC?
On 14-8-2011 23:05, Daniel Fischer wrote: On Sunday 14 August 2011, 22:42:13, Wishnu Prasetya wrote: On 14-8-2011 22:17, Daniel Fischer wrote: We don't know the times for a non-threaded run (or an -N1 run), so it could be anything from a slowdown to a 4× speedup (but it's likely to be a speedup by a factor 4×). Well, the -N1 is below. The sequential version of the program has almost the same profile: SPARKS: 5 (1 converted, 4 pruned) INIT time0.00s ( 0.00s elapsed) MUT time2.78s ( 2.99s elapsed) GCtime4.35s ( 4.15s elapsed) EXIT time0.00s ( 0.00s elapsed) Total time7.13s ( 7.14s elapsed) Am I correct then to say that the speed up with respect to sequential is equal to: tot-elapse-time-N1 / tot-elapse-N4 ? So I have 7.14 / 2.36 = 3.0 speed up, and not 1.46 as Iustin said? Yes (with respect to wall-clock time of course). That's not too bad, though it should be possible to increase that factor. I'll probably have to do something with that GC :) But that should be the first target, there's probably some low-hanging fruit there. Maybe increasing the size of the allocation area (+RTS -Ax) or the heap (+RTS -Hx) would already do some good. Also do heap profiling to find out what most likely takes so much GC time (before compiling for profiling, running with +RTS -hT could produce useful information). Ah ok. Thanks a lot for all the answers and tips. --Wish. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] strictness properties of monoidal folds
Hello Alexey, sorry for my slow response. On Thu, Aug 4, 2011 at 7:10 AM, Alexey Khudyakov alexey.sklad...@gmail.comwrote: On 02.08.2011 08:16, Sebastian Fischer wrote: Data.Foldable also provides the monoidal fold function foldMap. It is left unspecified whether the elements are accumulated leftwards, rightwards or in some other way, which is possible because the combining function is required to be associative. Does this additional freedom for implementors go so far as to allow for strict accumulation? Left and right folds behave identically for finite structures but they are different for infinite ones. I agree that for types like normal lists that allow infinite structure it makes sense to have different folds like foldr and foldl that are explicit about the nesting of the combining function. I don't think that there are laws for foldMap that require it to work for infinite lists. One could even argue that the monoid laws imply that the results for folding leftwards and rightwards should be equal, that is undefined.. For types that only allow finite sequences (like Data.Sequence or Data.Vector) this is not an issue. But you convinced me that a strict and lazy foldMap can be distinguished even for list types that have a strict append function by using a lazy mappend function for accumulation. Best regards, Sebastian ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
