On 01/18/12 08:22 PM, David Terei wrote:
On 18 January 2012 03:13, Karel Gardas<[email protected]> wrote:
Hello,
if this is desired, then is it possible to do this in non-standard nofib run
as this would probably kill any benchmarking on not so fast platform like
ARM...
I assume you mean by this that my aims with nofib will increase the
runtime of the suite and for ARM it's already very long? nofib already
has a design of different 'modes' where a mode is meant to allow
controlling of the total runtime. So there are 'fast', 'normal' and
'slow' modes. I think we should get all three of these modes working.
ARM can use 'fast' mode. I defiantly don't want to do this in a branch
/ forked nofib. There is no reason we can't accommodate your
requirements in the standard build.
Ah, yes, well then if fast is available and runnable in matter of hours
on ARM, then everything is all right. I've just felt a need to note that
not all platforms do have performance of recent Intel boxes... :-)
Thanks,
Karel
_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc
