[Mono-list] Fwd: Re: SciMark2 benchmark in F#

[Jon Harrop]

On Tuesday 20 January 2009 22:05:23 Alan McGovern wrote:
> Hi,
>
> Here are the results of running that assembly on several platforms/clrs on
> an original Macbook, 1.86ghz core2duo...

Thanks.

> A quick overview tells us that with the old JIT was 2x-6x slower than
> MS.NETand with the new jit performance goes from 0.25x-2x slower. The
> test that used to be 6x slower is now only 0.25x slower. Things have
> improved greatly since you originally ran your benchmark so it's definitely
> not accurate to call mono 3x slower than MS.NET anymore. 1.5x slower would
> be a nice median.

The Mono propaganda I had seen:

  http://www.mono-project.com/Linear_IL

seriously downplayed the performance improvements from Mono 2.0 -> 2.2, saying 
that is was only up to 30% faster when, in fact, it is up to 370% faster. You 
might want to update it. ;-)

Also, I would greatly value benchmarks like SciMark2 over IronPython 
benchmarks.

> Taking a slightly more detailed view tells us a few more interesting
> things. Compare mono 2.0 (windows) to mono 2.0 (opensuse). The exact same
> version of mono ranges between 1.5x faster on linux (LU arrays) to 1.5x
> slower on linux (SPARSE). What this tells us is that there appears to be
> differences in the platform itself which make a difference
> performance-wise.

I don't understand how a tight numerical loop could have such different 
performance characteristics between platforms?!

-- 
Dr Jon Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/?e