> I've seen about 5% variability in compile times, and various
> simple tests that use a "medium" amount of memory, on various x86
> systems over the years.
>
> Once a program has started "slow", it stays slow while it runs.
>
> I've always assumed it was L2 cache issues or similar. I don't
> know if there's a practical way to fix the problem, or if modern
> chipsets don't have this problem.
I am also experiencing this. My dual P-120 with Neptune
chipset is sometimes slower on tasks (x11amp in particular)
than my single P-100 with Triton chipset.
I suspect that at least part of the reason is due to the RT
process switching CPUs too much though. The rest is probably
due to the fact that this Neptune board has crappy L2 cache
management and worse behaviour for memory > 64MB (I have 72).
Maybe cache colouring would be worth it after all :))
cheers,
Rik -- All 2.2-pre kernels available at 10Mbit rates...
+-------------------------------------------------------------------+
| Linux memory management tour guide. riel@humbolt.geo.uu.nl |
| Scouting Vries cubscout leader. http://humbolt.geo.uu.nl/~riel |
+-------------------------------------------------------------------+
-
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/