Software bug: changing cgminer configuration claims to work at runtime, but does not. Thus, changing pool settings requires a reboot.
Yes editing the configuration file has no effect on a running cgminer.
Your comment is entirely unrelated to the problem at hand.
The management software is fully capable of restarting cgminer.
Yes I know, I wrote the interface and the code that hooks into the internal functions to do all that without killing the software, but instead shut it down properly.
The interface is called the API (I wrote almost all of the API)
The commands to stop or restart it are 'quit' and 'restart'
I also wrote the code that deals with who can access the API and the cgminer options that set that.
None of that existed before November 2011 ...
As for what is in there in their version? No idea. But most likely all they've done is get Xiangfu to add hooks into getting extra work for the 24? work items required to be sent to the device each time, instead of 1 (no it's not a bASIC/BFL MCU queue - oh well) and write a driver-xyz.c that uses that and fpgautils.c
The screen output you show of the miner performance is directly from my API.
Pity they didn't use miner.php (that I wrote) that's in there also ...
Though since we don't have access to the source, they probably have a bug in there that ckolivas found today that doesn't release all work created ... and on a machine that creates work items 80x faster than a typical BFL FPGA Single, you can of course expect that memory issue to be way more obvious ...
