Messages in this thread | | | Date | Sat, 30 Oct 1999 23:26:01 +0200 (CEST) | From | Andrea Arcangeli <> | Subject | Re: benchmark of the new buffer hash |
| |
On Sat, 30 Oct 1999, Manfred Spraul wrote:
>Have you thought about using a simple "sleep(1)"? This will synchonize
Yes this is a good idea of course ;). I only wanted a figure of the numbers and so I didn't bothered to do all the stuff necessary for a very precise bench.
>I've attached a modified version: (Pentium II/350 and Amd K6/200 (*)) >* hash13: 16 / 13 cpu ticks >* hash14pre: 46 / 33 cpu ticks > >(*) the K6 values vary, it seems the benchmark cannot force all cpu >caches into the 'hot' state. These are the lowest values.
There is a better way than the sleep one: in 2.2.x on i386 you can iopl(2) and asm("cli" :: ) by hand to do a precise bench. On 2.3.x you'll have to boot with the nonmi option to achieve the same.
Andrea
- 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/
| |