Topic: Swedish ASIC miner company kncminer.com - page 31. (Read 3049511 times)

Yeah, it makes complete sense that stuff works much more reliably when they are ran within spec. Running them reliably out of spec is always a gamble. Unfortunately, most mining hardware is made like shit. You get lucky when you can run shit hardware out of spec and reliably =) The even further problem is when manufacturers market them at a default speed which is beyond spec and miners think thats within spec. So, you have all this highly unreliable junk .... just to keep miners in competition with each other and further fuel the greed.
The end result: Mining hardware manufactures are the only ones who really make the money in this game.

but, as far as running out of spec .... as long as you keep things cool enough, 90% of the time you can get away with it just fine.

Even when I was GPU mining, I found that undervolting and running a bit slower/cooler was always a better option than hitting everything on head as hard as I could.  I'm not surprised that the Titans also run better this way. 

When I first got my Titan, it was totally unstable, but I wound up running it a little slower and it was better that way.  I should have taken the hint.  With everything that's transpired since then, it makes perfect sense.

And if people blow them up I can fix them :-) But really might be a better investment (not for me of course) to run them a bit cooler since these actually seem to have value over time. Which is kind of amazing.

250 would cause the supplies to go unstable on the re-wired one, remember this was with a destroyed power plug assembly (from running it too hot) and 200-225 is the top speed. That said I'm running another re-wire with one die running at 275, the other three running at 200, and it still pulls a good 60-65mh. Which is not bad when you consider the one running 275 on all dies only pulls 70mh.

Play with it, but keep in mind that when you run faster you pull more power for less mh and you stress the chip/supplies/board more. Ask yourself if a new supply every N months is worth it. Or winding up with a 0mh unit :-)

ive taco'd a few, toasted a few, hell im toasty right now!

in all seriousness though ive been able to push 4 titans on one board up to 250-270 range and keep it stable,
it took a lot of firmware hacking and liquid cooling, but it's totally possible!

And whatever you do don't crank the heatsink down on the chip or add washers on the top. Doing this will push down on the chip and result in a taco.

And don't run over 250w of power. 275 is FINE!

And always say your prayers before bedtime.

In the meantime I have been fiddling with speeds on re-wired Titans, I can get some of the dies up to 275mhz, but not more than two per board. Still, not bad.

Also figured out what causes the bridge boards to blow, it's the Pi boards on top shorting their little power supplies. When that happens the +5 (unfused of course) shorts dead and the board trace becomes the fuse. You can replace it with wire-wrap wire (or I can) but you really need to also look into buying a new Rpi. Make sure you get the right one, you can test the old one by trying to power it through the mini-usb port with 5v. If it don't come up it's toast.

Hi, I think I was bidding on that Titan you bought 

this is what I would do, -- replace the thermal paste, clean the heatsink & pcb with some 99.9% isopropyl alcohol & brush, blow it out real good, and replace the fans with noctura highspeed fans , then let them run for awhile. With the noctura fans & a clean heat sink you may not need any more cooling (dcdc heatsinks)

http://www.amazon.com/Arctic-Silver-High-Density-Polysynthetic-AS5-12GRFB/dp/B001JYVFLY?ie=UTF8&psc=1&redirect=true&ref_=oh_aui_search_detailpage
http://www.amazon.com/Bearing-Cooling-NF-A14-iPPC-3000-PWM/dp/B00KFCRF1A?ie=UTF8&psc=1&redirect=true&ref_=oh_aui_search_detailpage
http://www.amazon.com/MG-Chemicals-Isopropyl-Alcohol-Cleaner/dp/B005DNQX3C?ie=UTF8&psc=1&redirect=true&ref_=oh_aui_search_detailpage
http://www.amazon.com/MG-Chemicals-852-Non-Abrasive-Cleaning/dp/B004SPCXYO?ie=UTF8&psc=1&redirect=true&ref_=oh_aui_search_detailpage


good luck

Ok, I have been slowly reading back through this thread... It reads like a soap opera!  LOL

On the GenTarkin firmware I had a question.  For miners that have weak or non-hashing dies.  Does the firmware try to move the MHZ setting down on those dies to try and bring those dies back to life?  Or do we simply set the firmware at the starting point where we want to tune it and it mainly focuses on the leanest voltages at a given setting?   The rig I just bought I know has some weak or dead dies that I am hoping to bring back to life or get a few extra MH out of it.  I am waiting on this rig to arrive in the mail this week.  I have read what others have posted on how to get this tuned manually, but I was curious if the firmware is already doing some of this work. 

Also on the energy saver mode do you have a watts per hash calculation?  Maybe an option to choose a energy profile of "ultra conservative", "conservative", "passive-aggressive!" mode that calculates the best hash per watt consumed and tunes it down to that level based on the level of performance the user wants?   For example I saw the post earlier from Lightfoot regarding the heat and power consumption jump going from 225mhz up to 275mhz.  So I figured if the sweet spot for these dies for stable and cool operation is in the 225mhz range then it would seem that it would be possible to calculate energy profiles based on hash rates, temps, and power consumed.  If your already doing all of that please forgive my ignorance.  It is very appealing to me to be able to run this rig cool and stable... and save energy in the process while getting the best performance possible from the rig.  I have already ordered the parts and tools to do some of the mods to the heatsink, and the dcdc cooling.  I am hoping that by getting some fresh thermal paste, and getting better heat sinks on the dcdc converters that this will help with the rig I bought.  I wasn't able to tell from the thread if the investment in other fans would give me much of a benefit.  I am curious what others are doing in regards to the fans.  My rig is a batch 1. 

I think there is a copy of the GenTarkin firmware on this rig that I bought, but I have no idea if this is purchased copy or what the story is yet.  I will be reaching out once the rig is here and getting this license sorted out and get the latest copy of the GenTarkin masterpiece! :-)   

I appreciate all that you guys are doing and sharing on this thread.  It is very helpful!  Thanks!

This is a very interesting observation.  Have others had a similar experience?  Is there a good video out there showing how to do this?  I saw one video that wasn't bad, but I would love to have a close up view of how to do this.  If it really makes that big of a difference then it certainly sounds like something I would like to do also. 

Lightfoot, I like your thinking on this.  What happened at the 250mhz setting?  Is 225mhz the sweet spot?  What voltage can you turn it down to at that setting?  That is a huge difference and your right that grinding out those few MH for that much more electricity, heat, wear and tear doesn't make a lot of sense to me.  Would love to see more tuning in this direction... :-)  I finally bought a Titan and it will be here next week.  I am looking forward to getting it setup and tuned!  

my website: http://gentarkincustomtitan.pcriot.com

Where I can find the custom Titan firmware

Yeah, basically. And since it is much more likely you shorted the +.6 supply it's not that uncommon.

The problem though is getting flux under the die for a reflow, without flux shorted balls will never "re-ball" themselves under heat. You could pull the die and attempt to re-ball but this is really getting towards the edge of what I can do (not impossible, just that placing the chip with a .1mm error is one ball off).

If I can find a way into the chip without being destructive I can at least remove an errant die, so we'll see. This is all to fix one guy's cube, I just don't seem to want to give up on this. :-)

I watercooled both a Jupiter and a Nepturd.

With the Jupiter I fit
FOUR ASIC boards,
TWO controller cards, (so I can run it as 2 Saturns if I want to)
TWO 760Watt PSU,
TWO water pumps,
THREE fans,
and the 360x120x60mm thick radiator
ALL inside a standard Hoopiter case.

You have to click the links below, I only post links.
(to save people bandwidth if they are not interested)
http://imgur.com/GXvNz3g
http://imgur.com/IJzV8fA
http://imgur.com/xoHsaiJ

For the Nepturd...
First pic is during initial construction.
The water loop is running but only two ASICs are watercooled.
The three uninstalled waterblocks are sitting there waiting for assembly.
They are already part of the loop and have water running thru them.
http://imgur.com/r2Qjrig
Three unmodified air cooled ASIC's can be seen here on the shelf above during construction.
I used a tall Thermaltake water resivoir.
I also left an opening in the 220VAC outlet box for small children to stick their fingers in!
The two 2000 Watt PSU that make my 4000 Watt rail can also be seen here.
https://i.imgur.com/jeIRRcw.jpg
Next pic shows all five ASICs watercooled during initial testing.
http://imgur.com/9T8lRc2
I used two 360x120x90mm thick radiators for one nepturd.
http://imgur.com/5TktL32
I might redo the arrangement of my ASIC PCBs so the water and everything is in the center of the pentagon.
Then I can cool the DC/DC with a smaller fan and make a case with lower cost of materials.
I think I can use 1 fan for both PSU and the DC/DC converters.
I figured out how to easily watercool the DC/DC converters but never got to it.
(2 years later I doubt I will ever get to it)


For both of the designs I used two water pumps in series so I would still have flow even if one pump failed.
I also have a smoke detector that lives right on the miner rack.

A titan uses the same form factor so my mods would also work for it.


YMMV

So it's possible that "dead" dies are actually not dead at all, but the solder below them has just melted and shorted across the array?

Can't wait! Keep us updated!

https://translate.google.co.uk/translate?hl=en&sl=sv&u=http://img1.nsd.se/nyheter/boden/domstol-avgor-miljonkrav-mot-knc-9936377.aspx&prev=search

So it seems we will be getting an answer soon to the question that was mass-debated a million times in this thread - whether kfc's "we only sell to businesses, not consumers, hence you have no consumer rights" is lawful in Sweden or not.

If it's judged knc customers are consumers, knc are gonna be wholly fucked

I mean, if they go round saying they sell their shit to anyone....


We'll see indeed. Very soon.

Or maybe not. I just pulled a die and under it is what can be best described as an IGA (insane grid array). Let's just say it looks like a ball pitch of zero, but still bga. Maybe .1mm. But yes, the failure mode seems to be die overheats, melts solder under it, and either just shorts the +.6v lines or if very unlucky shorts the 3.3v supply to power the die's housekeeping.

Fascinating.

Scraping it off is not the issue: What we need here is a way to break that bond without putting up-pressure on the carrier since this thing is glued to what is basically a flexible interconnect soldered to the pins. Put up pressure on it and you will rip the interconnect.

Oh and you need to do it while it is on the board so we can have a running part at the end of the day. :-)

Can't use heat or you will cause the dies to explode their solder because *they* are glued to and when the solder heats up it has nowhere to go till it blows out (which is why the dies short).

Cheap nail remover (acetone+water) isn't doing anything. Will get some full strength acetone, question is what will that do to the board.

However if I can get the lids off pulling the bad die should be do-able. In fact re-seating the bad die should be do-able as well....

Did you try a single edge razor blade to scrape it off?  and/or carb cleaner, sprayed  on some cotton pads??