BFL Singles w/ cgminer - What's a good Utility to shoot for?

[Phraust]

After some quick math, it looks like the target utility for a single using the 832 firmware would be around U: 11.62/m.  My thinking is that anything at or above this number would be ideal, right?

I've spent the last few days reflashing down from the 896 firmware to whichever firmware runs with a U =< 11.62/m.  I wanted to know what other people running the singles were up to, and if this was something even worth doing...

[1714001167]

1714001167

[1714001167]

1714001167

[1714001167]

1714001167

[Cablez]

11.6 sounds about right for U.  With the faster firmwares I think I had it up to almost 12 but don't rely on my memory.

[Fjordbit]

After some quick math, it looks like the target utility for a single using the 832 firmware would be around U: 11.62/m.  My thinking is that anything at or above this number would be ideal, right?

I'm topping out at 11.30/m with that flash and this difficulty. This is across 17 units. Almost all are above 11.22/m

[Phraust]

I'm topping out at 11.30/m with that flash and this difficulty. This is across 17 units. Almost all are above 11.22/m

Thanks for the input...  I'm trying to eke out whatever I can out of the singles with their current configuration.  So far (after 17hrs) they are looking like this:

PGA   ENABLED   STATUS   TEMPERATURE   MHS AV   MHS 5S   ACCEPTED   REJECTED   UTILITY      FIRMWARE

0   Y      Alive      52.40°C      856.07   865.42   12,463   28      11.98/m      872
1   Y      Alive      51.80°C      836.12   857.69   12,086   17      11.62/m      864
2   Y      Alive      55.50°C      856.14   865.43   12,448   18      11.97/m      872
3   Y      Alive      56.50°C      847.26   857.68   12,334   14      11.86/m      872
4   Y      Alive      54.50°C      878.98   890.71   12,574   24      12.09/m      896
5   Y      Alive      54.00°C      833.09   857.68   11,946   15      11.48/m      864
6   Y      Alive      53.00°C      855.80   865.46   12,445   21      11.96/m      872
7   Y      Alive      57.40°C      856.28   865.19   12,479   11      12.00/m      872
8   Y      Alive      53.60°C      856.97   874.05   12,450   31      11.97/m      880
9   Y      Alive      56.50°C      864.40   873.73   12,355   30      11.88/m      880

I've basically been waiting a day, then checking their utility.  If it falls below 11.62, I've dropped it down a notch and tried again.  Right now it's pretty good, haven't reflashed in a few days.  Until I can figure out a sure-fire way of keeping them cool without having to run the a/c 24/7, I'm probably going to leave them as is.

My plan is to eventually take them out of their cases, re-thermal paste all of them (so far only three have been redone) and flip the top fans.

I should note that I have switched to P_Shep's cgminer (https://bitcointalk.org/index.php?topic=88379.msg973745#msg973745), since it's optimized for FPGAs.  I was running into issues where over the course of a day or so, they'd slowly start throttling themselves down.  With P_Shep's update, this no longer occurs.

[Phraust]

Dammit, after posting I noticed that two had dropped below (1&5), time to reflash them to 832.

[crazyates]

Why are you down-flashing? A low U means they're throttling?

[Phraust]

That's my best guess so far.  My thinking is that if it performs at less than 11.62m (Estimated 832 mhps rate), and I'm using the same or higher firmware, then it's not optimal.

[Phraust]

Well, looks like BFL updated EasyMiner with some "Speed Tuning" software.  http://www.butterflylabs.com/drivers/ Time to give that a shot.

[cuz0882]

If your going to have them for a while, upgrade to a delta fan. The one I'm using still fits in the case. I'm getting around 38-41c with the 896 flash. It should drop the temp by a good 5 degrees, I would think more if its a hot room. I used a fan controller to be safe and just set the unit on top of the old fan to save the original wiring. The case can still be closed up with a couple fan cables hanging out.
http://www.newegg.com/Product/Product.aspx?Item=N82E16835213009

[crazyates]

That's my best guess so far.  My thinking is that if it performs at less than 11.62m (Estimated 832 mhps rate), and I'm using the same or higher firmware, then it's not optimal.

Wouldn't a Single's U rate be determined by MHps AND the current difficulty? So you have to recalculate every time the difficulty changes?

[Cablez]

I don't think so as most pools that I know of are accepting diff=1 shares at this time except for p2pool. So U is just shares per minute regardless of difficulty now. In the future that might change though.

[-ck]

That's my best guess so far.  My thinking is that if it performs at less than 11.62m (Estimated 832 mhps rate), and I'm using the same or higher firmware, then it's not optimal.

Wouldn't a Single's U rate be determined by MHps AND the current difficulty? So you have to recalculate every time the difficulty changes?

No, it's always a difficulty 1 share at the moment, so it's dependent on MH/s and LUCK. So over short periods luck matters and the utility can't be relied upon for at least an hour.

[GernMiester]

12