On Sun, Jul 17, 2011 at 9:05 AM, Xiaofan Chen <xiaof...@gmail.com> wrote: > https://lists.berlios.de/pipermail/openocd-development/2010-October/016725.html > Øyvind Harboe wrote on Tue Oct 19 2010, >> As a start for debugging delay issues, type "help jtag_flush_queue_sleep". >> >> Plot the performance of 0, 10 and 100ms of an operation. It should allow >> you to guestimate how much of the performance hit is due to long latencies." > > Here are the results. Does this mean that "jtag interface latency does play a > big part in performance"? 1ms latency does not play a big part here. But > from 2ms onwards, the performance dropped significantly. > > Delay Flash write speed > 0ms 17.544 KiB/s > 1ms 17.370 KiB/s > 2ms 10.304 KiB/s > 5ms 6.457 KiB/s > 10ms 3.916 KiB/s > 100ms 0.487KiB/s (seems to take forever) > > Tested under Ubuntu 11.04, EK-LM3S1968 demo board, JtagKey2 > (FT2232H based high speed USB interface). >
The other advice from Øyvind Harboe's is also quite good but I do not know how to modify the codes. https://lists.berlios.de/pipermail/openocd-development/2010-October/016727.html > One interesting modification to OpenOCD might be to add some profiling > metrics on inter-jtag_execute_queue() times. > > If there are large inter-jtag_execute_queue() times then I'd expect an > async interface to help." Yet the other suggestion is to reduce the number of JTAG_EXECUTE by issuing longer clock/in out sequences. https://lists.berlios.de/pipermail/openocd-development/2010-October/016711.html https://lists.berlios.de/pipermail/openocd-development/2010-October/016722.html -- Xiaofan _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
