Re: 2.5.70-mm1

Martin J. Bligh (mbligh@aracnet.com)
Wed, 28 May 2003 23:23:23 -0700


> . A number of fixes against the ext3 work which Alex and I have been doing.
> This code is stable now. I'm using it on my main SMP desktop machine.
>
> These are major changes to a major filesystem. I would ask that
> interested parties now subject these patches to stresstesting and to
> performance testing. The performance gains on SMP will be significant.

Sexy. SDET beats the hell out of this, and is much improved:

SDET 128 (see disclaimer)
Throughput Std. Dev
2.5.66-mm2 100.0% 0.6%
2.5.66-mm2-ext3 3.9% 0.4%

SDET 128 (see disclaimer)
Throughput Std. Dev
2.5.70-mm1-ext2 100.0% 0.1%
2.5.70-mm1-ext3 22.7% 2.0%

diffprofile 2.5.70-mm1-ext2 2.5.70-mm1-ext3
(for SDET 128: + worse with ext3, - better.)
1857406 245.2% total
1531720 431.5% default_idle
106589 0.0% .text.lock.transaction
40119 0.0% do_get_write_access
37170 0.0% journal_dirty_metadata
35031 6560.1% __down
24412 8030.3% .text.lock.attr
19535 2556.9% __wake_up
19201 907.0% schedule
11344 0.0% start_this_handle
10104 0.0% journal_add_journal_head
10007 0.0% .text.lock.sched
7352 0.0% journal_stop
5949 99150.0% prepare_to_wait_exclusive
5867 3008.7% __blk_queue_bounce
4724 335.3% __find_get_block
4618 0.0% ext3_get_inode_loc
4410 0.0% journal_dirty_data
3754 590.3% __find_get_block_slow
3079 738.4% .text.lock.base
2176 0.0% ext3_do_update_inode
2132 11844.4% unlock_buffer
1995 0.0% ext3_journal_start
1888 0.0% ext3_orphan_del
1783 145.2% __brelse
1642 4829.4% blk_run_queues
1555 0.0% ext3_orphan_add
1495 0.0% ext3_new_inode
1430 0.0% ext3_reserve_inode_write
1412 0.0% journal_destroy_handle_cache
1344 0.0% journal_cancel_revoke
1279 0.0% journal_unmap_buffer
1198 0.0% ext3_free_blocks
...
-1057 -88.4% .text.lock.highmem
-1064 -24.5% remove_shared_vm_struct
-1112 -52.8% .text.lock.dec_and_lock
-1126 -100.0% ext2_new_inode
-1516 -100.0% grab_block
-1594 -28.3% path_lookup
-1599 -28.0% atomic_dec_and_lock
-1660 -10.4% copy_page_range
-1695 -15.4% __d_lookup
-2585 -23.6% release_pages
-2614 -13.5% zap_pte_range
-9758 -20.7% page_add_rmap
-26185 -25.3% page_remove_rmap

-
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/