Messages in this thread | | | Date | Wed, 9 Aug 2000 10:27:25 -0400 | From | Matthew Wilcox <> | Subject | Re: maximum/optimal # of SMP CPUs between 2.2 and 2.4 |
| |
On Tue, Aug 08, 2000 at 09:51:49AM -0700, Zack Brown wrote: > Someone asked me how Linux had improved from 2.2 to 2.4 in terms of the > number of CPUs it could handle at the maximum, and the number it was best > suited for. I grepped through the archives and the sources, but no luck.
As always, it will depend on the exact workload. For example if you have a set of processes which do nothing but apply and remove posix locks to files, I can assure you it's one CPU only. On the other hand, if you're doing webserving, I think http://www.spec.org/osg/web99/results/res2000q2/ tells its own story about Linux's scalability. Oooh! I hadn't noticed before but Dell posted a new one for Q3: http://www.spec.org/osg/web99/results/res2000q3/web99-20000710-00057.asc which manages to score almost double the next highest (non-TUX) score.
Management overview of Linux scalability at webserving:
CPU Score Scaling 1 1270 100% 2 2200 86.6% 4 4200 82.7% 8 6387 62.9%
Management overview of NT scalability at webserving:
CPU Score Scaling 1 732 100% 4 1598 54.6%
so it can be argued that Linux scales to 8 CPUs better than NT scales to 4 CPUs. Have a look at the site yourself for all the `this is a benchmark it is only a benchmark, share prices may go down as well as up, may contain nuts' disclaimers, but the benchmarks were conducted on identical hardware, as far as I can see.
-- Revolutions do not require corporate support.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |