Is anyone mining with a lowly 5850? My friend can't get higher than 28 KH/s.
I would have thought he could get at least 50 Kh/s. (It can do about 330kh/s on Scrypt)
Any ideas?
I know a couple are running 7850's in the 50-60Kh/s range, but I'm not sure of settings. You may PM "Thirtybird" as I know he has run one on UTC. In the meantime you could play more with TC and LG. Thirtybird gave some tips for TC as well:
If I set g=2 then my whole system freezes regardless of i / tc. On my 280X it has a significant impact on performance (more so than lookup-gap).
That's probably because the TC value is too high and the thread is being put into dynamic GPU memory (Aka system ram, which if there isn't enough gets paged out to your hard drive - every try and calculate hashes using a hard drive for storage?? BAD!) - I had this exact some issue when I was trying to get my miner up and running and that was what I discovered.
The easiest way I figured out how to know where to set TC or buffer size was using GPU-z (people on Windows 8 should use HWInfo). On the sensors tab, there are two graphs - Dedicated memory and dynamic memory. Have that open when you launch your miner - if dynamic memory goes up at all, kill the mining software ASAP. Then lower your TC until only dedicated memory increases when you launch the miner. MY 4GB cards would do this with any value for TC over 24576 with a lookup gap of 2 (using -g 2) - buffer-size 1,536 MB.
All I know is my P3 Kill-a-watt says my pc is pulling the exact same wattage from wall on UTC as any scrypt coin.. yet I run 7-8C cooler on UTC
A KAW is more for estimation and should not be used as completely accurate. They can vary 1-10% depending on the power draw; high inductive loads will net a larger error than pure resistive. So, the variation of temperature can occur with a drop in power and the KAW not correctly show this. We tend to run GPU's in the lower efficiency range as it is, so it's very possible you didn't drop a great deal of power though. Heat is the best comparator with tiny changes.
I've always said "the only thing humans have mastered in our entire existence is fire (heat)" Lol, that usually follows with; "do you know how a nulcear plant works?". People are surprised that we use all that technology to make steam with heat; to then run turbines, which in turn run generators. Yep, our best power system; is just a glorified steam engine.
From a test here:
http://forums.anandtech.com/showthread.php?t=137169PC, using a 750watt Ultra lsp model.
Core2, running at 3.3ghz, 7900gtx , pc running gelato (nvidia software that uses the gpu as a fpu, while also using the host cpu), maxes both.
Killawatt 316watts, fluke 329watts.another test here:
http://www.jonnyguru.com/forums/showthread.php?t=2695Reference:
Valhalla Scientific 2101 Digital Power Analyzer (which isn't perfect, but for this comparison, we'll assume it is. )
A. 173W 270VA, non PFC loads only(non PFC PC + LCD monitor)
B. 648W 676VA, non PFC + resistive mix (that PC + resistive dummy load)
C. 1065W add more resistive load
D. 1420W add more resistive load
KAW:
A. 173W 269VA +0.0%
B. 656W 686VA +1.2%
C. 1080-1100W jumpy +1.4 to 3.2%
D. 1435-1460W jumpy +1.0 to 2.8%
There will be variation between units, but I find +/- 10% adequate for PC power supply test and reviews. If you are loading up your PSU so much that +/- 10% makes a life and death difference, you're not giving it enough safety margin. It's only $20 and the measurements are far better than you can guesstimate.
