Re: Kaffe/Jigsaw

Here is a little bench.

this test is using ptester, running on an ss20/SunOS5.5 and
Jigsaw running on a P90/Linux 2.0.25

With Kaffe-0.6.0 JIT

ptester -h138.96.152.39 -p8007 -c5 -k5 /test1k.html /test4k.html
Running test: >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

URL                  NRq  Min Ct Avg Ct Max Ct  Min Tx Avg Tx Max Tx  
Bytes
/test4k.html        1698  0.0010 0.0003 0.0126  0.0779 0.0351 0.4193  4355
/test1k.html        1710  0.0010 0.0002 0.0101  0.0661 0.0349 0.4169  1286

Actual test time.. 60 seconds
Total requests.... 3408 (56 requests/sec)
Total failed...... 0 (0 requests/sec)
Total bytes....... 9596176 (159936 bytes/sec)

With Sun's VM ported to Linux:

ptester -h138.96.152.39 -p8007 -c5 -k5 /test1k.html /test4k.html
Running test: >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

URL                  NRq  Min Ct Avg Ct Max Ct  Min Tx Avg Tx Max Tx  
Bytes
/test1k.html        1056  0.0010 0.0003 0.0727  0.1508 0.0562 0.6733  1286
/test4k.html         930  0.0010 0.0002 0.0041  0.2598 0.0639 3.2828  4355

Actual test time.. 60 seconds
Total requests.... 1986 (33 requests/sec)
Total failed...... 0 (0 requests/sec)
Total bytes....... 5409458 (90157 bytes/sec)

So, a JIT improves the performances ;)

I made more or less the same test with Kaffe on sparc/solaris (without 
JIT) and the result was... the opposite! Kaffe was two time slower than 
Sun's VM.

Of course it is just a fast test. i didn't set up the optimal 
configuration for jigsaw performance ;)

	~~Yves

Received on Thursday, 28 November 1996 10:23:03 UTC