JDK version ?

Brian Adkins brian at lojic.com
Wed Mar 24 11:10:33 PST 1999


At 11:29 AM 3/24/99 -0700, Godmar Back wrote:
>> 
>> It might be worthwhile to make whatever changes you need to be able to run
>> the Volano benchmark well:
>> 
>> http://www.javaworld.com/jw-03-1999/jw-03-volanomark.html
>> 
>
>I did that a while ago:
>
>Here's the numbers I got for the simple local loopback test:
>
>FreeBSD 3.1/JDK1.1.7:		410 m/s
>FreeBSD 3.1/Kaffe/jit/select:	700 m/s
>Linux/JDK1.1.7/intrp		808 m/s
>Linux/Kaffe/jit/select		840 m/s
>Linux/Kaffe/jit/poll		860 m/s
>Linux/JDK1.1.7/TYA1.2v4 	1120 m/s

Ok, if I use your FreeBSD figure of 410 and normalize it to the 166 seen in
the URL above (because of different hardware), it looks like the
Linux/Kaffe/jit/poll would be about 348 versus Microsoft's 1412.

Actually, I'm more interested in the amount of concurrent connections Kaffe
can handle - that test separated the men from the boys.

I'm not trying to be critical for the heck of it here.  I need to determine
viable platforms for my product, and performance is an important factor.

>All Linux results are under Debian 2.1.  My RH 5.2 box is a bit slower
>(identical hardware, different kernel.)  All results were obtained
>on a Dual PII/350.
>
>Volanomark doesn't run with native threads, so it's all green threads
>for Sun.  It's all jthreads for Kaffe.
>
>	- Godmar



More information about the kaffe mailing list