Quote from: Biontic on September 12, 2014, 03:01:24 PM

Hey just a quick one - I feel I am missing something here. Maybe someone can enlighten me.

I have an S3 B1, running the following...
antMiner_S320140826.bin
cgminer 4.6

Load is nice and low - but the hash rate seemed to be below what I am seeing on other S3's - so I was trying to push it a bit beyond to see if I could 'encourage' a few extra GH/s from it.

So SSH'd in and added the following -

option 'freq_value' '1406' #256.25M
option 'chip_freq' '256.25'
option 'timeout' '15

:wq -> reboot - but it is still showing as 250 in the gui - can anyone confirm if the new 'Advanced Settings' tab under 'Miner configuration' will over ride the manual entries I made in the asic-freq file?

Also - while I am at it...

I have such wide variance in the hashrate of the S3's anyone any suggestions on coaxing the slower ones towards 500. 2 of them are consistently at 505+ but some just seem determined to go slower - like sub 440 and even down as far as 400. Have tried playing around with freq's and FW and different PSU's etc. Has anybody had any solid results with checking/re-applying the thermal paste or even just re-seating the heat sinks, as I have not gone down this road yet?
Thanks

The problem S3 which I changed out the thermal paste was down about 430 this morning. I had moved both into the back room last night as after a warm day it got uncomfortable in my living room. So, I restarted the problem S3 this morning and it has been getting warm in there, 91° 31%rh. It has been running well since rebooting, the warmer room and low humidity. 6hr50min 440.91GH/s, 0 HW, but utility is 15, fewer shares tho the score seem equal to the other. It's temps read 47 & 46 while the #1 unit is reading 44 & 42, running at 441.60 HW 160 in 15 hours but its utility is 16.

Having learned the art of building heatsinks for high voltage FETS in the late 1970's, what you want for perfection is two perfect surfaces in contact but since perfection isn't possible you want as little thermal paste as possible filling any air gaps between the surfaces.

When I took the inner heatsink off a hashing board, a good board by the way, the thermal paste seemed applied by machine in 4 rows along the top ASICs and 4 more rows along the bottom ASICs. It was not well spread, just flattened. There were air gaps between the rows. Thermal paste conducts heat less well than the copper on the board or the aluminum. So, essentially the heat went from the copper plane to the rows of thermal paste to the aluminum on the other side of the thermal paste. I removed it all and applied an even thin layer of thermal paste to the surface and tightened from the center out. When doing the outer heatsink I suggest you carefully remove it after removing the 4 spring'd screws but not to slide the heatsink while removing. If you lift it up and off you won't build up thermal compound on the very fine ASIC pins as might happen if you slide the heatsink to the side. I'm sure the stuff doesn't conduct but dust might adhere to it at the pins over time. Big mistake is thinking more paste more conduction.

Also, as you remove the silver thermal paste from the inner heatsink, you may notice a couple or more small lumps of solid thermal compound stuck to the heatsink. At least that was the case of the bad hashing board that had new paste applied in addition to the original. No way that heatsink could have been flush, it stood away from the copper plane at least the depth of the hard lumps.