Jon Christiansen wrote:
If you do this, just be sure to query the system to be sure there are
1 CPU :)
Runtime.getRuntime().availableProcessors()
And allow the user to override via a System property in case they dont
want to overrun all their CPU's when running findbugs.....
You probably get a better results if FB internally would use SEDA
architecture with several processing conveyors. Disk IO / class parsing
/ analysis. And then allow to tweak thread pool for each of those
individually.
regards,
Eugene
On 12/4/06, Andrei Loskutov <loskutov@xxxxxx> wrote:
Hi all,
I've just asked myself why findbugs uses only one CPU core if running on
multicore systems? I saw in my process monitor that one core sleeps and
another one was 100% loaded.
Many users have 2x core systems already. E.g. in our lab we have ~50%
multicore systems now, next year probably all 100% of workstations, and
then we expect to see 4x core workstations soon.
Unlike compiler, findbugs does not really requires *sequentual*
analysis,
so are there any plans to parallelize bug search task? I could help with
Eclipse integration of this feature, if it is needed.
| 