[abc] almost the same as ajc

As some of you may know - I have been bugged by the fact that we
are slower than ajc on the weka-nullptr benchmark.

Sascha worked long and hard yesterday to make a slightly different
around inliner that makes static methods (like ajc, but better).

I made one more tweak - to remove a limit he had on container size,
and finally I think we have got it.

Here are timings from my computer - have to wait till tomorrow
morning for the tofu timings.

Benchmark, with no apsects

ajc - no aspect
4.287u 0.057s 0:04.68 92.5% 0+0k 0+0io 2pf+0w
4.252u 0.041s 0:04.46 96.1% 0+0k 0+0io 1pf+0w
4.367u 0.058s 0:04.45 99.1% 0+0k 0+0io 1pf+0w
abc - no aspect
4.337u 0.045s 0:04.76 91.8% 0+0k 0+0io 1pf+0w
4.434u 0.061s 0:04.55 98.6% 0+0k 0+0io 1pf+0w
4.298u 0.048s 0:04.50 96.2% 0+0k 0+0io 1pf+0w
javac
4.279u 0.052s 0:04.60 93.9% 0+0k 0+0io 1pf+0w
4.304u 0.053s 0:04.51 96.4% 0+0k 0+0io 1pf+0w
4.250u 0.042s 0:04.45 96.4% 0+0k 0+0io 1pf+0w

Benchmark, with aspects

ajc
6.267u 0.134s 0:07.31 87.4% 0+0k 0+0io 1pf+0w
6.845u 0.146s 0:07.30 95.6% 0+0k 0+0io 1pf+0w
6.792u 0.170s 0:07.26 95.8% 0+0k 0+0io 1pf+0w
7.012u 0.184s 0:07.25 99.1% 0+0k 0+0io 1pf+0w
6.801u 0.134s 0:07.25 95.5% 0+0k 0+0io 1pf+0w

abc (acting like ajc inliner)
6.252u 0.158s 0:07.33 87.3% 0+0k 0+0io 1pf+0w
6.995u 0.214s 0:07.26 99.1% 0+0k 0+0io 1pf+0w
7.026u 0.203s 0:07.25 99.5% 0+0k 0+0io 1pf+0w
6.779u 0.200s 0:07.25 96.1% 0+0k 0+0io 1pf+0w
6.862u 0.175s 0:07.32 96.0% 0+0k 0+0io 1pf+0w

abc force (the original inliner)
7.699u 0.178s 0:08.29 94.8% 0+0k 0+0io 1pf+0w
7.735u 0.162s 0:08.24 95.7% 0+0k 0+0io 1pf+0w
7.707u 0.190s 0:08.25 95.6% 0+0k 0+0io 1pf+0w

So, I think ajc and abc are pretty close, have to wait for tofu
numbers tomorrow.

Cheers, Laurie

+-------------------------------------------------------------+
| Laurie Hendren, Professor, School of Computer Science |
| McGill University |
| 318 McConnell Engineering Building tel: (514) 398-7391 |
| 3480 University Street fax: (514) 398-3883 |
| Montreal, Quebec H3A 2A7 hendren@cs.mcgill.ca |
| CANADA http://www.sable.mcgill.ca/~hendren |
| http://wwww.sable.mcgill.ca http://aspectbench.org |
+-------------------------------------------------------------+
Received on Mon Apr 18 10:47:30 2005