Mainline kernel OLTP performance update

[Styner, Douglas W]

Summary: Measured the mainline kernel from kernel.org (2.6.29.2). 

The regression for 2.6.29.2 against the baseline, 2.6.24.2 is 2.07% (2.6.29.1 regression was 2.35%).  Oprofile reports 70.419% user, 29.5709% system. 2.6.29.1 -> 2.6.29.2 comparison is below. 

Linux OLTP Performance summary
Kernel#            Speedup(x)   Intr/s  CtxSw/s us%     sys%    idle%   iowait%
2.6.24.2                1.000   22106   43709   75      24      0       0
2.6.29.2                0.979   30509   43139   75      25      0       0

Server configurations:
Intel Xeon Quad-core 2.0GHz  2 cpus/8 cores/8 threads
64GB memory, 3 qle2462 FC HBA, 450 spindles (30 logical units)

======oprofile CPU_CLK_UNHALTED for top 30 functions
Cycles% 2.6.24.2                   Cycles% 2.6.29.2
74.8578 <database>                 67.7080 <database>
1.0500 qla24xx_start_scsi          0.9487 qla24xx_intr_handler
0.8089 schedule                    0.8117 schedule
0.5864 kmem_cache_alloc            0.6215 qla24xx_wrt_req_reg
0.4989 __blockdev_direct_IO        0.5439 kmem_cache_alloc
0.4357 __sigsetjmp                 0.4784 qla24xx_start_scsi
0.4152 copy_user_generic_string    0.4703 __blockdev_direct_IO
0.3953 qla24xx_intr_handler        0.4416 try_to_wake_up
0.3850 memcpy                      0.4253 __sigsetjmp
0.3596 scsi_request_fn             0.3803 scsi_request_fn
0.3188 __switch_to                 0.3701 __switch_to
0.2889 lock_timer_base             0.3619 copy_user_generic_string
0.2750 memmove                     0.3476 rb_get_reader_page
0.2519 task_rq_lock                0.3149 symbols)
0.2474 aio_complete                0.3006 aio_complete
0.2460 scsi_alloc_sgtable          0.2903 memset_c
0.2445 generic_make_request        0.2883 ring_buffer_consume
0.2263 qla2x00_process_completed_re0.2719 lock_timer_base
0.2118 blk_queue_end_tag           0.2699 __list_add
0.2085 dio_bio_complete            0.2474 blk_queue_end_tag
0.2021 e1000_xmit_frame            0.2453 memmove
0.2006 __end_that_request_first    0.2392 e1000_xmit_frame
0.1954 generic_file_aio_read       0.2290 ipc_lock
0.1949 kfree                       0.2290 task_rq_lock
0.1915 tcp_sendmsg                 0.2249 generic_make_request
0.1901 try_to_wake_up              0.2167 kref_get
0.1895 kref_get                    0.2147 tcp_sendmsg
0.1864 __mod_timer                 0.2045 qla2x00_process_completed_re
0.1863 thread_return               0.2045 pick_next_highest_task_rt

-- 2.6.29.1 vs. 2.6.29.2
Linux OLTP Performance summary
Kernel#            Speedup(x)   Intr/s  CtxSw/s us%     sys%    idle%   iowait%
2.6.29.1                1.000   30570   42818   74      25      0       0
2.6.29.2                1.003   30509   43139   75      25      0       0

Server configurations:
Intel Xeon Quad-core 2.0GHz  2 cpus/8 cores/8 threads
64GB memory, 3 qle2462 FC HBA, 450 spindles (30 logical units)

======oprofile CPU_CLK_UNHALTED for top 30 functions
Cycles% 2.6.29.1                   Cycles% 2.6.29.2
64.5424 <database>                 67.7080 <database>
1.1571 qla24xx_intr_handler        0.9487 qla24xx_intr_handler
0.9209 schedule                    0.8117 schedule
0.6533 kmem_cache_alloc            0.6215 qla24xx_wrt_req_reg
0.5456 qla24xx_wrt_req_reg         0.5439 kmem_cache_alloc
0.5247 try_to_wake_up              0.4784 qla24xx_start_scsi
0.4858 qla24xx_start_scsi          0.4703 __blockdev_direct_IO
0.4485 __sigsetjmp                 0.4416 try_to_wake_up
0.3976 __blockdev_direct_IO        0.4253 __sigsetjmp
0.3857 __switch_to                 0.3803 scsi_request_fn
0.3692 copy_user_generic_string    0.3701 __switch_to
0.3648 aio_complete                0.3619 copy_user_generic_string
0.3633 scsi_request_fn             0.3476 rb_get_reader_page
0.3259 rb_get_reader_page          0.3149 symbols)
0.3109 ring_buffer_consume         0.3006 aio_complete
0.3050 memset_c                    0.2903 memset_c
0.2900 pick_next_highest_task_rt   0.2883 ring_buffer_consume
0.2885 page_fault                  0.2719 lock_timer_base
0.2855 task_rq_lock                0.2699 __list_add
0.2691 mwait_idle                  0.2474 blk_queue_end_tag
0.2661 lock_timer_base             0.2453 memmove
0.2616 symbols)                    0.2392 e1000_xmit_frame
0.2616 __list_add                  0.2290 ipc_lock
0.2541 tcp_sendmsg                 0.2290 task_rq_lock
0.2302 blk_queue_end_tag           0.2249 generic_make_request
0.2242 e1000_xmit_frame            0.2167 kref_get
0.2198 scsi_softirq_done           0.2147 tcp_sendmsg
0.2183 qla2x00_process_completed_re0.2045 qla2x00_process_completed_re
0.2168 memmove                     0.2045 pick_next_highest_task_rt
0.2138 cpupri_set                  0.2024 __mod_timer
0.2078 qla24xx_process_response_que0.2004 kmem_cache_free
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/