[Mono-devel-list] New Mono compilation engine available.

[Paolo Molaro]

On 04/08/03 Sebastian Bergmann wrote:
> > I filed a bug report about it.
> 
>   Thank you.

Dietmar fixed it yesterday.
FWIW, here are the relative results I get (mini -O=all vs mono):

Relative results for mini -O=all:
index_string_test 	1.693 (32.58 %)
index_test 	3.198 (-7.65 %)
loop_test 	0.024 (-50.00 %)
nested_int_test 0.021 (-47.50 %)
nested_test 	0.050 (-39.76 %)
object_test 	0.058 (-22.67 %)
return_test 	0.009 (-57.14 %)

I get consistent results when increasing the test run time by an order
of magnitude for the tests that take less than 1/10th of a second.
So we get one regression in the index_string_test and quite good results
in the other microbenchmarks (insert quote about the relative
uselfulness of microbenchmarks here:-).

I need to investigate the reason of the regression in index_string_test:
I think it may be related to the use of internal calls, because I saw
already similar regressions that could be explained with that.

lupus

-- 
-----------------------------------------------------------------
lupus at debian.org                                     debian/rules
lupus at ximian.com                             Monkeys do it better