I am running the new beta 1.3.1beta2 on a 7970 and bfl single. Just turned it up but so far so good. Is a couple mhash faster too; previously I was getting 8 mhash under the driver I ran (so 808mhash firmware would run at 800 on bitminter) but now it is 5-6 mhash less instead of 8. The 7970 seems faster too but could be my imagination as gpu tends to fluctuate quite a bit more than the fpga.
Windows 7 64 bit running latest java 7 update 11
Will let you know if I have any issues and post this on your software thread.
That's what I like to hear! Thanks for testing. Looks like 1.3.1 will soon be ready.
Each block (on average) you will lose half a nonce range ... if the code is optimal - i.e. if it aborts the nonce range when an LP arrives and throws away the work - without counting it since it can't be counted.
Yes, with every block change it aborts the current work. And the already processed hashes that are thrown away are not counted towards the hashrate. So at every block change you can see a dip in the hashrate. That can be a little confusing for some users, but it's a result of how the BFL FPGAs work.