Post
Topic
Board Off-topic
Re: 896 mh/s firmware release - Butterfly Labs
by
kano
on 30/05/2012, 07:01:23 UTC
Just uploaded the 896mh firmware, and I'm getting almost exactly half (ie ~470) mh. No throttles, just goes up to that and stays there. Any ideas on what would cause this? In cgminer btw.

What is your U?

I am running BAMT with cgminer 2.3.1f. When I uploaded the 872 firmware the hash rate is showing 433 but the U after 12+ hours is 12.09.

Not sure, just fired it up and thought that it was really weird.
Yes, a long term U value is way more reliable to determine if it is hashing at a given rate
(for FPGA in older versions of cgminer)

Even a short term (a few minutes) U will pretty much always be within 50% and given maybe an hour it should usually be within 10%
Anyone with a bit of stats knowledge should be able to come up with the actual % values/SD over time - but my figures are based on observation (and yes I have seem some very unexpected U values)

Earlier versions of the cgminer BFL (and Icarus) code had some issues displaying the correct MH/s
That may be exacerbated by the faster firmware
That 2.3.1f version is obviously from me (any versions floating around with extra letters on the end are usually from my git)
Though I've had pretty much nothing to do with the BFL code in releases (just testing with people)
I've spent most of my FPGA code effort on Icarus (ngzhang is great with technical details)
- though some of those Icarus changes may rub off on the BFL code eventually.
Though the next major FPGA change, if I get around to it (or someone else does it) will affect both of them.