Some benchmarks on 16-way NUMA-Q w/ 16 Gb RAM.
Kernbench-2: (make -j N vmlinux, where N = 2 x num_cpus)
Elapsed User System CPU
2.5.59 46.08 563.88 118.38 1480.00
2.5.59-mjb1 45.99 566.45 111.98 1474.67
2.5.59-mjb2 45.56 564.81 113.29 1487.83
2.5.59-mjb3 45.66 565.33 110.18 1479.00
2.5.59-mm6 45.95 563.16 115.85 1477.50
DISCLAIMER: SPEC(tm) and the benchmark name SDET(tm) are registered
trademarks of the Standard Performance Evaluation Corporation. This
benchmarking was performed for research purposes only, and the run results
are non-compliant and not-comparable with any published results.
Results are shown as percentages of the first set displayed
SDET 1 (see disclaimer)
Throughput Std. Dev
2.5.59 100.0% 0.8%
2.5.59-mjb1 94.4% 2.0%
2.5.59-mjb2 96.1% 2.8%
2.5.59-mjb3 96.7% 4.1%
2.5.59-mm6 100.4% 0.7%
SDET 64 (see disclaimer)
Throughput Std. Dev
2.5.59 100.0% 0.7%
2.5.59-mjb1 119.9% 0.8%
2.5.59-mjb2 122.3% 0.8%
2.5.59-mjb3 122.6% 0.4%
2.5.59-mm6 102.7% 0.8%
NUMA schedbench 64:
AvgUser Elapsed TotalUser TotalSys
2.5.59 0.00 235.18 3632.73 15.45
2.5.59-mjb1 0.00 232.06 3658.60 15.38
2.5.59-mjb2 0.00 233.38 3618.16 15.20
2.5.59-mjb3 0.00 236.59 3627.47 15.24
2.5.59-mm6 0.00 239.24 3634.88 15.89
----------------------------------------
Summary:
Kernbench is mainly boosted by dcache_rcu and wli's pfn_to_nid and
pgd_ctor changes. (note the system times).
Going from virgin to -mjb gives a small degredation on SDET 1,2,4.
but a good boost on SDET 8 up to 64. I presume this is a NUMA scheduler
effect, but haven't confirmed yet. I have Ingo's B0 patch apart from the
topo cleanup ... not really too worried about this - slight effect on
a huge machine at very low loads.
NUMA schedbench results are really very boring.
-----------------------------------------
kernbench-2
diffprofile 2.5.59 2.5.59-mjb3 (+ is worse in -mjb, - better).
2541 do_schedule
2112 .text.lock.file_table
422 file_move
265 prepare_to_wait
215 get_empty_filp
186 kmem_cache_alloc
121 find_get_page
116 eventpoll_init_file
95 do_anonymous_page
94 vm_enough_memory
91 file_ra_state_init
80 __copy_to_user_ll
72 vfs_read
58 mark_page_accessed
56 default_wake_function
52 __wake_up
51 pte_chain_alloc
...
-51 strnlen_user
-51 profile_exit_mmap
-61 __pte_chain_free
-68 buffered_rmqueue
-71 .text.lock.dcache
-84 fd_install
-101 kmem_cache_free
-115 eventpoll_release
-171 may_open
-173 path_lookup
-179 find_trylock_page
-183 kmalloc
-235 vfs_follow_link
-248 clear_page_tables
-259 prepare_to_wait_exclusive
-274 pgd_alloc
-441 path_release
-624 dput
-625 pfn_to_nid
-889 default_idle
-931 current_kernel_time
-933 __fput
-1135 dget_locked
-1382 link_path_walk
-2060 __d_lookup
-2398 schedule
-2881 .text.lock.dec_and_lock
-5163 .text.lock.namei
-8863 total
---------------------------------------------------------
SDET 64
diffprofile 2.5.59 2.5.59-mjb3 (+ is worse in -mjb, - better).
56544 default_idle
8837 __down
7385 do_schedule
3538 page_remove_rmap
2830 .text.lock.file_table
2665 __wake_up
1429 page_add_rmap
1209 d_lookup
873 find_get_page
799 follow_mount
458 release_pages
425 kmem_cache_alloc
418 file_move
360 get_empty_filp
272 ext2_new_block
263 ext2_free_blocks
261 vfs_read
...
-245 .text.lock.dir
-284 pte_alloc_one
-288 may_open
-307 set_fs_pwd
-309 proc_check_root
-358 find_trylock_page
-360 vfs_follow_link
-371 real_lookup
-425 open_namei
-430 sys_unlink
-436 pfn_to_nid
-489 d_delete
-524 d_rehash
-568 __fput
-570 prune_dcache
-598 kmalloc
-717 d_alloc
-865 pgd_alloc
-920 dput
-1053 d_instantiate
-1233 path_release
-1275 dget_locked
-1293 kmem_cache_free
-1354 .text.lock.attr
-1502 .text.lock.namespace
-1646 .text.lock.base
-1750 clear_page_tables
-2063 current_kernel_time
-2373 __d_lookup
-2774 link_path_walk
-3407 do_lookup
-4215 schedule
-5388 atomic_dec_and_lock
-7874 path_lookup
-25037 .text.lock.dcache
-33737 .text.lock.dec_and_lock
-55200 .text.lock.namei
-73012 total
---------------------------------------------------
SDET 1
diffprofile 2.5.59 2.5.59-mjb3 (+ is worse in -mjb, - better).
(though the tick count is small, this is the average of 5 runs).
988 total
930 default_idle
10 d_lookup
8 page_remove_rmap
8 do_schedule
7 page_add_rmap
6 __copy_to_user_ll
5 kmem_cache_alloc
5 do_wp_page
...
-5 schedule
-6 link_path_walk
-7 pfn_to_nid
-12 __d_lookup
-
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/
This archive was generated by hypermail 2b29 : Fri Feb 07 2003 - 22:00:10 EST