diff options
author | Seth, Rohit <rohit.seth@intel.com> | 2005-10-29 18:15:47 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@g5.osdl.org> | 2005-10-29 21:40:35 -0700 |
commit | ba56e91c940146e99ac694c4c7cd7f2b4aaa565d (patch) | |
tree | 5254f1ce8e2f39ccb58a74d0ff0120817b3b5c05 /mm | |
parent | fcdae29aa7a5c79f245110f6680afdc1858d3626 (diff) | |
download | kernel_samsung_smdk4412-ba56e91c940146e99ac694c4c7cd7f2b4aaa565d.zip kernel_samsung_smdk4412-ba56e91c940146e99ac694c4c7cd7f2b4aaa565d.tar.gz kernel_samsung_smdk4412-ba56e91c940146e99ac694c4c7cd7f2b4aaa565d.tar.bz2 |
[PATCH] mm: page_alloc: increase size of per-cpu-pages
Increase the page allocator's per-cpu magazines from 1/4MB to 1/2MB.
Over 100+ runs for a workload, the difference in mean is about 2%. The best
results for both are almost same. Though the max variation in results with
1/2MB is only 2.2%, whereas with 1/4MB it is 12%.
Signed-off-by: Rohit Seth <rohit.seth@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Diffstat (limited to 'mm')
-rw-r--r-- | mm/page_alloc.c | 24 |
1 files changed, 12 insertions, 12 deletions
diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 94c864e..f799217 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -1721,29 +1721,29 @@ static int __devinit zone_batchsize(struct zone *zone) /* * The per-cpu-pages pools are set to around 1000th of the - * size of the zone. But no more than 1/4 of a meg - there's - * no point in going beyond the size of L2 cache. + * size of the zone. But no more than 1/2 of a meg. * * OK, so we don't know how big the cache is. So guess. */ batch = zone->present_pages / 1024; - if (batch * PAGE_SIZE > 256 * 1024) - batch = (256 * 1024) / PAGE_SIZE; + if (batch * PAGE_SIZE > 512 * 1024) + batch = (512 * 1024) / PAGE_SIZE; batch /= 4; /* We effectively *= 4 below */ if (batch < 1) batch = 1; /* - * Clamp the batch to a 2^n - 1 value. Having a power - * of 2 value was found to be more likely to have - * suboptimal cache aliasing properties in some cases. + * We will be trying to allcoate bigger chunks of contiguous + * memory of the order of fls(batch). This should result in + * better cache coloring. * - * For example if 2 tasks are alternately allocating - * batches of pages, one task can end up with a lot - * of pages of one half of the possible page colors - * and the other with pages of the other colors. + * A sanity check also to ensure that batch is still in limits. */ - batch = (1 << fls(batch + batch/2)) - 1; + batch = (1 << fls(batch + batch/2)); + + if (fls(batch) >= (PAGE_SHIFT + MAX_ORDER - 2)) + batch = PAGE_SHIFT + ((MAX_ORDER - 1 - PAGE_SHIFT)/2); + return batch; } |