Free Public limited version will be available in the next few days.
Dear Sir,
I have just tested your public 718M and 825M version.
But the power cost is much different. I use 220v psu. 2500W.
1, The 718M version power cost is about 1850W, and the hashrate is about 150k. temp is about less than 90 degree. Seems good.
2, The 825M version power cost is about 2450W, and the hashrate is about 160k. temp is about 100degree.
i dont run for a long time, because of afraid of the power cost too much.
And i think you are very kind person, i like to TT btc to you. And would you please offer me the better version firmaware, but power cost is no more than 2200w when 220v?
And may i know your Z11 v1.7 version hashrate, power cost(220v, 110v), temperature? Very like to get it even pay the money. thanks a lot!
Thanks a lot!
The new 1.7 version is about 2000-2200 watts @825 high power and about 1900-2100 @825 low power both seem to get me between 158-162 k/sol with peaks in the 190's k/sol. There was a overhaul to both the public and private versions that increased performance while reducing power usage. I do not have a way to measure the 220v power draw. Both versions have user selectable frequency with a choice of high or low power for each frequency. The private version goes to the full 825m the public version is limited to 775m.
I have been able to keep my temps in the 70-100 @825 depending on the outside temp. I also keep them in a 12x12 building with air conditioning which also helps on the 100 degree days we have been having. I have ran mine as high as 108 for hours before the added cooling and the machine just kept chugging along. I believe the shutdown is about 110-115 then it thermal throttles.
I am finishing up the testing of the public version now and it will be available by tomorrow the private version is ready to go and has been fully tested. Both versions you may see some errors in the log or xxx on a board for the first 2-20 minutes while it does a behinds the scenes scan to determine the best freq. Once you see scan done in the log file the speeds should increase for the first hr or two then settle to what you will average.
If interested in the private version pm me and I will reply with the payment details or you can wait for the 100% free public version. Due to lack of donations this will be the last public version released with the exception of bugfixes for it but no additional features will be added. Private versions will continue to be upgraded as much as possible. Unless something changes this is where I stand on the public release.
(this is not a troll post.)
So.. you are saying that giving things away doesn't work and that charging is the best option? BTW, dev-fees exist on mine because of user requests; some wanted to pay $0, some wanted to pay something, some wanted to amortize the costs via dev-fees.
If you've decided to actually try to *use* the frequency scanning built in coupled with manipulating voltages, that's an interesting idea... but the actual scanning literally takes seconds, so I'm a little confused on the '2-20 minutes' part unless you are letting the automatic restart scanning portion run (i.e., starts at a given frequency, runs for a bit, if certain errors occur, will drop a calculated Mhz off of the current, and restart). If this is what you've come up with, then good job -- creative re-use of the existing functionality.The scanning actually underclocks sometimes and can leave "hash rate" on the table, which coupled with a fixed voltage(s) is likely to end up adding extra heat into ASICs when they could produce the same hashrate at a lower voltage, but that is a trade off. If this isn't what you've done, then you can have this idea as it should mostly work.
If that's the case, then I assume the difference in your 'free' and 'private' is simply the starting scanning frequency being hard coded plus a different voltage setting?
Hopefully that works well across a broader set of machines as a given voltage that works on one ASIC/machine is just as likely to work as it is to not work on another ASIC/machine.
In reality, you should expect to get something pretty darned close to the long term average in 5-10 minutes runtime. Are you sure on the "first hr or two" part?
Regarding "peaks in 190", are you measuring that in the web or the CLI? Running at 825 I tend to see peaks in the 225-230 range with my stuff, but I use the CLI most of the time.
-j