Tarkin, question: Does your source code fix the issue where a power supply literally disappears? I've had this happen twice on a Titan in, power cycling clears it and it is different supplies at different times. If your code fixes that then it is exceptionally "worth it" as trying to power a die with a single supply will result in a blown up component quickly...
If you have PSU's randomly disappearing from your Titan then you have some burglary you should check into =P
On a more serious note, are you referring to dies which eventually stop hashing and a power cycle is needed to bring them back?
If so, then yes, that was the first thing I addressed when I bought my Titan, just to see if I could issue power cycling of the dies as needed, automatically.
That was one of the reasons I started writing the firmware mod, once I found out this was possible a whole galaxy of cool ideas were born and thats what my firmware is today =)
So, my firmware will power cycle the DCDC's of a cube when a die on that cube is detected as being "sleeping, dropped off, stopped hashing ... whatever you wanna call it" ... so far, taking into account peoples success w/ my firmware. Id say about 99% of the time my firmware brings those dies back hashing =) .. it works flawlessly 100% of the time on my titan.
If theres a true hardware problem w/ ur titan tho then of course my firmware cant do much bout that =P
Check out all the features of my firmware here:
http://gentarkincustomtitan.pcriot.comI just read my site and seems I didnt have that clearly stated in my list of features of firmware. I stated it in paragraphs above it but I can see how that can easily be overlooked. So, I just tweaked the list and wording a bit =)
Oh wait wait, I just reread what u asked. Are you talking about one of the DCDC's disapearing for a respective die but the die continues to hash?
If so, to answer that, my firmware on its current release does not address that problem. Its rare and I didnt catch it for a long time.
Sometimes when it does that it will drive the single remaining DCDC as hard as it can and pulls near 50-60A through the poor thing =( ... eventually the DCDC will simply just shut itself down because these DCDC's do have overcurrent protection & overheat protectio(both extreme cases) built in. If the DCDC doesnt entirely shut itself down it will go into a current limiting mode where it will give the ASIC all the power it can but in a very spurty manner, so the ASIC can hash at roughly just over half the speed(half the output from the ASIC will be HW errors) while the DCDC can remain powered on.
The next release my firmware will catch that condition and it will power cycle the affected ASIC and bring both DCDC's back online(given that one of the DCDC's arent permanently damaged).