Re: [Lse-tech] lockmeter results comparing 2.4.17, 2.5.3, and 2.5.5

Hanna Linder (hannal@us.ibm.com)
Wed, 27 Feb 2002 11:57:57 -0800


--On Wednesday, February 27, 2002 11:45:15 -0800 Andrew Morton <akpm@zip.com.au> wrote:

> The big one is lru_list_lock, of course. I'll be releasing code in
> the next couple of days which should take that off the map. Testing
> would be appreciated.

Ill be glad to run this again with your patch. Also, John Hawkes
has an even bigger system and keeps hitting lru_list_lock too.
>
> I have a concern about the lockmeter results. Lockmeter appears
> to be measuring lock frequency and hold times and contention. But
> is it measuring the cost of the cacheline transfers?

This has come up a few times on lse-tech. Lockmeter doesnt
measure cacheline hits/misses/bouncing. However, someone said
kernprof could be used to access performance registers on the Pentium
chip to get this info. I don't know anyone who has tried that though.
I am working on a patch to decrease cacheline bouncing and it
would be great to see some specific results. Is anyone working on a tool
that could measure cache hits/misses/bouncing?

Hanna

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/