From: Arnim L. <arn...@gm...> - 2008-09-06 05:18:02
|
Hi Steve, > 2) cable xpc_ext: slow - we know this is the case because the > necessary reverse-engineering of the protocol has yet to be done, and > we're just bit-banging the thing, AFAIK. No complaint there. The > one thing I would suggest we look into is that the Xilinx tools > always kick a "frequency" command into their svf's, which used to > blow the thing up, but now just hangs it. Short of making a script > or tweaking a makefile to parse the "freq" lines out of the svf, just > ignoring freq for the xpc would be preferable, I think. While working on the segfault for the RUNTEST commands, I found that the calibration loop was broken. Under certain conditions the calibrated delay value would underflow and you'd end up delaying for 2^32 x 100000 and even more. The calibration should now quickly find a solution for frequencies above 200k. Below of that some iterations are still required. Please let me know if this brings the xpc_ext driver back into the game. Cheers Arnim |