[jhc] Benchmarks including JHC
john at repetae.net
Wed Jul 28 18:08:13 EDT 2010
On Sun, Jul 25, 2010 at 08:33:10PM -0300, Felipe Lessa wrote:
> In case someone on this list missed this blog post:
> It is nice to see JHC being superior than GHC in some tests. =)
Cool, I had seen an earlier version of it, good to see it updated with
0.7.5. I am working on getting all the compile errors exposed by this
fixed, there were quite a few minor parsing/typechecker bugs that were
minor in that they had easy workarounds, but given a large enough amount
of code in the wild you were bound to run into one or another of them. I
have been able to fix the vast majority of them for 0.7.6.
I have not looked much into the ones where jhc performs slowly, but I
can pretty much guarentee at least a few of them are due to
deforestation, hitting the right list deforestation can make or break
your compiler when it comes to speed. I plan on addressing this issue in
an upcoming release, probably not 0.7.6 though. I need to make the RULES
mechanism a bit more robust and less resource intensive before I throw
the large number of RULES a good deforester requires at it.
It also looks like a few of the tests are failing due to using
non-portable libraries when they can be trivially made haskell 98 by
replacing a few imports.
John Meacham - ⑆repetae.net⑆john⑈ - http://notanumber.net/
More information about the jhc