Comments on: GHC’s performance with threads is impressive http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/ Bryan O'Sullivan's blog Mon, 06 Sep 2010 09:50:52 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: Simon Marlow http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/comment-page-1/#comment-123539 Simon Marlow Mon, 11 Feb 2008 21:19:03 +0000 http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/#comment-123539 Lennart's GHC bug report is here: http://hackage.haskell.org/trac/ghc/ticket/1589 the non-linear growth of thread creation is due to the garbage collector not having a proper write barrier on threads. Hopefully I'll get around to fixing it before 6.10. Lennart’s GHC bug report is here:

http://hackage.haskell.org/trac/ghc/ticket/1589

the non-linear growth of thread creation is due to the garbage collector not having a proper write barrier on threads. Hopefully I’ll get around to fixing it before 6.10.

]]> By: Bryan O'Sullivan http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/comment-page-1/#comment-122653 Bryan O'Sullivan Fri, 08 Feb 2008 15:54:16 +0000 http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/#comment-122653 Nick, thanks for mentioning the formatting problem. I wish Wordpress was more predictable. Ben, it is indeed possible that the main thread might sneak in and get the MVar before any of the other threads starts running, but that would be visible in the measured output. Nick, thanks for mentioning the formatting problem. I wish WordPress was more predictable.

Ben, it is indeed possible that the main thread might sneak in and get the MVar before any of the other threads starts running, but that would be visible in the measured output.

]]> By: Ben Moseley http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/comment-page-1/#comment-122650 Ben Moseley Fri, 08 Feb 2008 15:38:47 +0000 http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/#comment-122650 Is there anything in that code which actually ensures that the message will go all the way round? (ie that the measurements will be measuring the correct thing) - 'takeMVar' is documented to be FIFO, but isn't there a danger that the main thread will get its take in before some of the threads? Is there anything in that code which actually ensures that the message will go all the way round? (ie that the measurements will be measuring the correct thing) – ‘takeMVar’ is documented to be FIFO, but isn’t there a danger that the main thread will get its take in before some of the threads?

]]> By: Nick Burlett http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/comment-page-1/#comment-122413 Nick Burlett Thu, 07 Feb 2008 18:13:09 +0000 http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/#comment-122413 Great post, thanks for the simple example! Note: when reading this post over on Planet Haskell, I noticed that the less-than signs are missing. I checked the rss feed directly here, and they appear not to be escaped, nor are they escaped in the HTML for this page (although they display correctly). I recommend updating your code snippit with the "<" mark. Great post, thanks for the simple example!

Note: when reading this post over on Planet Haskell, I noticed that the less-than signs are missing. I checked the rss feed directly here, and they appear not to be escaped, nor are they escaped in the HTML for this page (although they display correctly). I recommend updating your code snippit with the “<” mark.

]]> By: Bryan O'Sullivan http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/comment-page-1/#comment-122369 Bryan O'Sullivan Thu, 07 Feb 2008 15:08:36 +0000 http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/#comment-122369 Lennart, I can't find that ticket. Do you happen to know what its number is? Lennart, I can’t find that ticket. Do you happen to know what its number is?

]]> By: augustss http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/comment-page-1/#comment-122315 augustss Thu, 07 Feb 2008 11:45:14 +0000 http://www.serpentine.com/blog/2008/02/06/ghcs-performance-with-threads-is-impressive/#comment-122315 There's an open bug report (by me) about the non-linear scaling of thread operations. There’s an open bug report (by me) about the non-linear scaling of thread operations.

]]>