lkml.org 
[lkml]   [2013]   [Oct]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v2] can: c_can: Speed up rx_poll function
On Tue, Oct 29, 2013 at 01:34:48AM -0700, Joe Perches wrote:
> On Tue, 2013-10-29 at 09:27 +0100, Markus Pargmann wrote:
> > This patch speeds up the rx_poll function by reducing the number of
> > register reads.
> []
> > 125kbit:
> > Function Hit Time Avg s^2
> > -------- --- ---- --- ---
> > c_can_do_rx_poll 63960 10168178 us 158.977 us 1493056 us
> > With patch:
> > c_can_do_rx_poll 63939 4268457 us 66.758 us 818790.9 us
> >
> > 1Mbit:
> > Function Hit Time Avg s^2
> > -------- --- ---- --- ---
> > c_can_do_rx_poll 69489 30049498 us 432.435 us 9271851 us
> > With patch:
> > c_can_do_rx_poll 103034 24220362 us 235.071 us 6016656 us
>
> Also nicer if you updated the timings table
>
>

Yes I just measured the timings again:

./perf_can_test.sh 125000 30
Function Hit Time Avg s^2
-------- --- ---- --- ---
c_can_poll 127882 6178806 us 48.316 us 4393411 us
c_can_do_rx_poll 63941 3764057 us 58.867 us 776162.2 us
c_can_enable_all_interrupts 255764 2213697 us 8.655 us 1093934 us
c_can_plat_write_reg_aligned_t 807252 1607115 us 1.990 us 10053684 us
c_can_isr 127882 1220001 us 9.540 us 789.495 us
c_can_object_put 119887 1039222 us 8.668 us 1608.668 us
c_can_plat_read_reg_aligned_to 1015072 1033283 us 1.017 us 7021.465 us
c_can_read_msg_object 63943 1026159 us 16.048 us 718.464 us
c_can_activate_all_lower_rx_ms 7992 755782.4 us 94.567 us 55.270 us
c_can_mark_rx_msg_obj 55951 709072.1 us 12.673 us 39.974 us
c_can_object_get 63943 555669.2 us 8.690 us 96.211 us
c_can_msg_obj_is_busy 183830 527826.1 us 2.871 us 7289.221 us
alloc_can_skb 63943 170966.6 us 2.673 us 165.765 us
c_can_has_and_handle_berr 63941 47063.18 us 0.736 us 2.757 us

./perf_can_test.sh 1000000 30
Function Hit Time Avg s^2
-------- --- ---- --- ---
c_can_poll 270678 30290751 us 111.906 us 5809627 us
c_can_do_rx_poll 207109 24322185 us 117.436 us 171469047 us
c_can_object_put 841431 7278794 us 8.650 us 305841.0 us
c_can_plat_write_reg_aligned_t 4037180 6244636 us 1.546 us 4581066 us
c_can_read_msg_object 453988 6033464 us 13.289 us 128809.6 us
c_can_enable_all_interrupts 541342 5849826 us 10.806 us 22458900 us
c_can_activate_all_lower_rx_ms 55349 5237761 us 94.631 us 19004.79 us
c_can_mark_rx_msg_obj 387429 4865632 us 12.558 us 380606.4 us
c_can_plat_read_reg_aligned_to 4597629 4633247 us 1.007 us 315286.2 us
c_can_object_get 453988 3919692 us 8.633 us 59847.76 us
c_can_msg_obj_is_busy 1295419 3706862 us 2.861 us 316655.7 us
c_can_isr 270671 2496734 us 9.224 us 530.967 us
alloc_can_skb 453988 856917.4 us 1.887 us 18157.68 us
c_can_activate_rx_msg_obj 11210 141177.4 us 12.593 us 123.068 us
c_can_has_and_handle_berr 63569 44995.85 us 0.707 us 12.780 us

It is interesting that the number of hits for c_can_do_rx_poll is twice as much
as it was with find_next_bit. Unfortunately this reduces the overall benefit of
this patch. Any ideas how to increase the number of waiting message objects we
handle in one poll call?


Regards,

Markus

--
Pengutronix e.K. | |
Industrial Linux Solutions | http://www.pengutronix.de/ |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |


\
 
 \ /
  Last update: 2013-10-29 10:21    [W:0.069 / U:3.704 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site