On Fri, May 8, 2009 at 9:53 AM, Magnus Lundin <lun...@mlu.mine.nu> wrote: > Ųyvind Harboe wrote: >> >> On Fri, May 8, 2009 at 9:40 AM, Zach Welch <z...@superlucidity.net> wrote: >> >>> >>> On Fri, 2009-05-08 at 09:33 +0200, Ųyvind Harboe wrote: >>> [snip] >>> >>>> >>>> I have to give a bit of thought on how to best to profile this, i.e. to >>>> find the precise location of the culprit. >>>> >>>> Any ideas? >>>> >>> >>> I have had success with oprofile. This would give you what you need. >>> >> >> We effectively need to know the *stack trace* for those >> jtag_execute_queue()'s >> that are just sitting there waiting for the driver to complete. >> >> > > You are wasting your time, I did some work on the STM32 flash performance > problems a few weeks ago, and improved write speed from 5kB/s to 15kB/s or > more, and the only real big difference was removing extra > jtag_execute_queue() calls, and thus avoiding the USB roundtrip delay. > > The numbers for ARM7 targets are worse for single stepping here the new code > is less than half the speed of the old one.
Thanks for the pointer, patience and constructive feedback. I *will* address these performance problems next. I have a few more things I have to iron out, then I'll start on performance. If someone has a bug/regression that comes first though. -- Øyvind Harboe Embedded software and hardware consulting services http://consulting.zylin.com _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
