Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1479819
| From | Mel Gorman <mgorman@techsingularity.net> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | [PATCH 2/4] block, brd: Treat storage as non-rotational |
| Date | 2016-09-09 12:10 +0200 |
| Message-ID | <sfqUV-1ud-3@gated-at.bofh.it> (permalink) |
| References | <sfqLf-1c3-23@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
Unlike the rims of a punked out car, RAM does not spin. Ramdisk as
implemented by the brd is treated as rotational storage. When used as swap
to simulate fast storage, swap uses the algoritms for minimising seek times
instead of the algorithms optimised for SSD. When the tree_lock contention
was reduced by the previous patch, it was found that the workload was
dominated by scan_swap_map(). This patch has no practical application as
swap-on-ramdisk is dumb is rocks but it's trivial to fix.
4.8.0-rc5 4.8.0-rc5
batch-v1 ramdisknonrot-v1
Amean System-1 192.98 ( 0.00%) 181.00 ( 6.21%)
Amean System-3 198.33 ( 0.00%) 86.19 ( 56.54%)
Amean System-5 105.22 ( 0.00%) 67.43 ( 35.91%)
Amean System-7 97.79 ( 0.00%) 89.55 ( 8.42%)
Amean System-8 149.39 ( 0.00%) 102.92 ( 31.11%)
Amean Elapsd-1 219.95 ( 0.00%) 209.23 ( 4.88%)
Amean Elapsd-3 79.02 ( 0.00%) 36.93 ( 53.26%)
Amean Elapsd-5 29.88 ( 0.00%) 19.52 ( 34.69%)
Amean Elapsd-7 24.06 ( 0.00%) 21.93 ( 8.84%)
Amean Elapsd-8 33.34 ( 0.00%) 23.63 ( 29.12%)
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
---
drivers/block/brd.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/block/brd.c b/drivers/block/brd.c
index 0c76d4016eeb..83a76a74e027 100644
--- a/drivers/block/brd.c
+++ b/drivers/block/brd.c
@@ -504,6 +504,7 @@ static struct brd_device *brd_alloc(int i)
blk_queue_max_discard_sectors(brd->brd_queue, UINT_MAX);
brd->brd_queue->limits.discard_zeroes_data = 1;
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, brd->brd_queue);
+ queue_flag_set_unlocked(QUEUE_FLAG_NONROT, brd->brd_queue);
#ifdef CONFIG_BLK_DEV_RAM_DAX
queue_flag_set_unlocked(QUEUE_FLAG_DAX, brd->brd_queue);
#endif
--
2.6.4
Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
[RFC PATCH 0/4] Reduce tree_lock contention during swap and reclaim of a single file v1 Mel Gorman <mgorman@techsingularity.net> - 2016-09-09 12:00 +0200
[PATCH 2/4] block, brd: Treat storage as non-rotational Mel Gorman <mgorman@techsingularity.net> - 2016-09-09 12:10 +0200
[PATCH 3/4] mm, vmscan: Stall kswapd if contending on tree_lock Mel Gorman <mgorman@techsingularity.net> - 2016-09-09 12:10 +0200
[PATCH 1/4] mm, vmscan: Batch removal of mappings under a single lock during reclaim Mel Gorman <mgorman@techsingularity.net> - 2016-09-09 12:10 +0200
Re: [PATCH 1/4] mm, vmscan: Batch removal of mappings under a single lock during reclaim Peter Zijlstra <peterz@infradead.org> - 2016-09-16 15:30 +0200
Re: [PATCH 1/4] mm, vmscan: Batch removal of mappings under a single lock during reclaim Peter Zijlstra <peterz@infradead.org> - 2016-09-16 16:10 +0200
Re: [PATCH 1/4] mm, vmscan: Batch removal of mappings under a single lock during reclaim Linus Torvalds <torvalds@linux-foundation.org> - 2016-09-16 20:40 +0200
Re: [PATCH 1/4] mm, vmscan: Batch removal of mappings under a single lock during reclaim Peter Zijlstra <peterz@infradead.org> - 2016-09-17 03:40 +0200
[PATCH 4/4] mm, vmscan: Potentially stall direct reclaimers on tree_lock contention Mel Gorman <mgorman@techsingularity.net> - 2016-09-09 12:10 +0200
Re: [RFC PATCH 0/4] Reduce tree_lock contention during swap and reclaim of a single file v1 Linus Torvalds <torvalds@linux-foundation.org> - 2016-09-09 17:40 +0200
Re: [RFC PATCH 0/4] Reduce tree_lock contention during swap and reclaim of a single file v1 Mel Gorman <mgorman@techsingularity.net> - 2016-09-09 18:20 +0200
Re: [RFC PATCH 0/4] Reduce tree_lock contention during swap and reclaim of a single file v1 "Huang\, Ying" <ying.huang@intel.com> - 2016-09-09 20:20 +0200
csiph-web