TechEmpower Benchmarks
614 views
Skip to first unread message
vold
Nov 30, 2015, 4:21:48 AM11/30/15
to Yesod Web Framework
TechEmpower have posted a new run of their benchmarks, and Yesod's performance is awful. Have they done something wrong, or is it really that bad?
Message has been deleted
vold
Nov 30, 2015, 4:35:08 AM11/30/15
to Yesod Web Framework
Incidentally, I can't see which version of Yesod or GHC they used. If it's what comes with Ubuntu 12, that's GHC 7.4 and some really old version of Yesod.
Michael Snoyman
Nov 30, 2015, 4:38:34 AM11/30/15
to yeso...@googlegroups.com
About a year ago, I spent 2 very painful weeks trying to sort out the build process they were using for their Haskell code. GHC 7.8 was out and had introduced some major performance improvements relative to 7.6, but Ubuntu was shipping 7.6 at the time. Eventually I got the changes pushed through, but ever since then, I haven't bothered spending any time trying to win the benchmark game over there, since it's just a game of diminishing returns.
It's possible that there's a real regression that they've detected, but since I haven't seen such regression reported by others, I'm more likely to believe this is yet more busywork waiting for someone who wants to try to optimize their Haskell setup, and I don't intend to be that guy again.
On Mon, Nov 30, 2015 at 11:21 AM, vold <volde...@hotmail.com> wrote:
TechEmpower have posted a new run of their benchmarks, and Yesod's performance is awful. Have they done something wrong, or is it really that bad?
--
You received this message because you are subscribed to the Google Groups "Yesod Web Framework" group.
To unsubscribe from this group and stop receiving emails from it, send an email to yesodweb+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Cody Goodman
Dec 30, 2015, 2:26:12 AM12/30/15
to Yesod Web Framework, mic...@snoyman.com
They are now using an updated GHC/Yesod according to their Github.
I could very well be remembering this wrong, but I recall doing some performance testing and coming to the (potentially wrong) conclusion that persistent was my bottleneck.
At the time I think it was simply because the connection pool was small or something... Actually that's probably unrelated since I wasn't using Yesod and lack of pool resizing would have been my fault.
I could very well be remembering this wrong, but I recall doing some performance testing and coming to the (potentially wrong) conclusion that persistent was my bottleneck.
At the time I think it was simply because the connection pool was small or something... Actually that's probably unrelated since I wasn't using Yesod and lack of pool resizing would have been my fault.
Reply all
Reply to author
Forward
0 new messages