Swedish ASIC miner company kncminer.com - page 16.

pandalion98

sr. member

Activity: 504

Merit: 250

Quote from: alh on July 13, 2016, 12:03:56 AM

Quote from: lightfoot on July 12, 2016, 04:27:53 PM

Quote from: GenTarkin on July 11, 2016, 09:28:21 PM

Interesting, Ive had many pi's over the years. Not one single issue.

I know, it's odd. And I've never seen a BB that did this either.

It is *possible* that the gpio pins are being overdriven running the various spi signals to the boards and the fpga, but that sounds like a bit of a stretch. Still it's possible, I've got a total of four dead Pi's from various people who sent things in for repairs. In all cases the Pis blew out the CPU and sink a lot of current.

Fortunately fix-able.

While I don't any KNC gear, my experience with Raspberry Pi has been quite favorable. I can also say that my sum total experience involves ZERO use of the GPIO pins. Everything I have done has been USB based one way or another. My recollection is that the GPIO pins are designed for 3.3V and NOT 5V. Might that explain anything, or do I have my figures wrong?

Your figures are correct. It can only output 3V3 and can NOT accept/output 5V input.

I repeat, the GPIO pins should NOT be exposed to anything above 3V3.

alh

legendary

Activity: 1846

Merit: 1052

Quote from: lightfoot on July 12, 2016, 04:27:53 PM

Quote from: GenTarkin on July 11, 2016, 09:28:21 PM

Interesting, Ive had many pi's over the years. Not one single issue.

I know, it's odd. And I've never seen a BB that did this either.

It is *possible* that the gpio pins are being overdriven running the various spi signals to the boards and the fpga, but that sounds like a bit of a stretch. Still it's possible, I've got a total of four dead Pi's from various people who sent things in for repairs. In all cases the Pis blew out the CPU and sink a lot of current.

Fortunately fix-able.

While I don't any KNC gear, my experience with Raspberry Pi has been quite favorable. I can also say that my sum total experience involves ZERO use of the GPIO pins. Everything I have done has been USB based one way or another. My recollection is that the GPIO pins are designed for 3.3V and NOT 5V. Might that explain anything, or do I have my figures wrong?

lightfoot

legendary

Activity: 3164

Merit: 2258

I fix broken miners. And make holes in teeth :-)

Quote from: GenTarkin on July 11, 2016, 09:28:21 PM

Interesting, Ive had many pi's over the years. Not one single issue.

I know, it's odd. And I've never seen a BB that did this either.

It is *possible* that the gpio pins are being overdriven running the various spi signals to the boards and the fpga, but that sounds like a bit of a stretch. Still it's possible, I've got a total of four dead Pi's from various people who sent things in for repairs. In all cases the Pis blew out the CPU and sink a lot of current.

Fortunately fix-able.

GenTarkin

legendary

Activity: 2450

Merit: 1002

Quote from: Tupsu on July 12, 2016, 01:18:21 PM

Quote from: GenTarkin on July 12, 2016, 01:04:00 PM

Quote from: Tupsu on July 12, 2016, 05:25:06 AM

Quote from: carlosmnk on July 11, 2016, 10:58:26 PM

Hi, i have a problem with some cubes; strangelly, they suddenly appear a message that sais "Dies flagged for bypass of health checking: 3:1 3:2 3:3 4:1 6:1" and theese dies are not hashing and i can do nothing to fix it (reboot, change the controller, change the psu, change config...). Here is summer, but i have them with a low configuration to mantain them below 70-75ºC, so i don't think it could be because it is too hot. Here is a pic:

PS: I'm using GenTarkin 1.0.3 and 1.0.4 (i tryed both)

http://i68.tinypic.com/2moceb4.jpg

I am worried about theese cubes...

I had exactly the same thing with GenTarkin software.
Better use Original KNC software with the same volts.

I did this and it is no problem any more. When i replace the memory card to GenTarkin software , the problem is back.

Interesting.... Im going to guess you have dies that eventually stop hashing over time w/ stock firmware?

No, they are selected cubes which has never been a problem with hashing at default settings.

Odd ... well, unless there is some unknown bug in my firmware which triggers erroneous die resets, those cubes most likely do have some sort of issue going on. Afterall, your cubes may be one of those that after many upon many software resets the dies come back just fine. There are a small number of Titans which act like this. I guess it could be due to faulty pool work / connection problems too.
My firmware uses the same cube/die problem detection code(die amp threshold), it just handles the situation way differently then stock firmware.
Anywho, log files would give an insight as to whats really going on.

Micky25

legendary

Activity: 974

Merit: 1000

using GenTarkin software is the most effective way to run the Titans, I spent weeks on testing.

Tupsu

legendary

Activity: 1218

Merit: 1003

Quote from: GenTarkin on July 12, 2016, 01:04:00 PM

Quote from: Tupsu on July 12, 2016, 05:25:06 AM

Quote from: carlosmnk on July 11, 2016, 10:58:26 PM

Hi, i have a problem with some cubes; strangelly, they suddenly appear a message that sais "Dies flagged for bypass of health checking: 3:1 3:2 3:3 4:1 6:1" and theese dies are not hashing and i can do nothing to fix it (reboot, change the controller, change the psu, change config...). Here is summer, but i have them with a low configuration to mantain them below 70-75ºC, so i don't think it could be because it is too hot. Here is a pic:

PS: I'm using GenTarkin 1.0.3 and 1.0.4 (i tryed both)

http://i68.tinypic.com/2moceb4.jpg

I am worried about theese cubes...

I had exactly the same thing with GenTarkin software.
Better use Original KNC software with the same volts.

I did this and it is no problem any more. When i replace the memory card to GenTarkin software , the problem is back.

Interesting.... Im going to guess you have dies that eventually stop hashing over time w/ stock firmware?

No, they are selected cubes which has never been a problem with hashing at default settings.

GenTarkin

legendary

Activity: 2450

Merit: 1002

Quote from: Tupsu on July 12, 2016, 05:25:06 AM

Quote from: carlosmnk on July 11, 2016, 10:58:26 PM

Hi, i have a problem with some cubes; strangelly, they suddenly appear a message that sais "Dies flagged for bypass of health checking: 3:1 3:2 3:3 4:1 6:1" and theese dies are not hashing and i can do nothing to fix it (reboot, change the controller, change the psu, change config...). Here is summer, but i have them with a low configuration to mantain them below 70-75ºC, so i don't think it could be because it is too hot. Here is a pic:

PS: I'm using GenTarkin 1.0.3 and 1.0.4 (i tryed both)

http://i68.tinypic.com/2moceb4.jpg

I am worried about theese cubes...

I had exactly the same thing with GenTarkin software.
Better use Original KNC software with the same volts.

I did this and it is no problem any more. When i replace the memory card to GenTarkin software , the problem is back.

Interesting.... Im going to guess you have dies that eventually stop hashing over time w/ stock firmware?

GenTarkin

legendary

Activity: 2450

Merit: 1002

Quote from: carlosmnk on July 11, 2016, 10:58:26 PM

Hi, i have a problem with some cubes; strangelly, they suddenly appear a message that sais "Dies flagged for bypass of health checking: 3:1 3:2 3:3 4:1 6:1" and theese dies are not hashing and i can do nothing to fix it (reboot, change the controller, change the psu, change config...). Here is summer, but i have them with a low configuration to mantain them below 70-75ºC, so i don't think it could be because it is too hot. Here is a pic:

PS: I'm using GenTarkin 1.0.3 and 1.0.4 (i tryed both)

I am worried about theese cubes...

This is happening because those particular dies are so unstable in any given point in time that multiple attempts were taken to get them hashing again through the form of soft / hard resets done by my firmware.
In short it means those dies are unstable and my firmware will no longer try to bring them back online to prevent bfgminer from constantly restarting.

The stock firmware will do the same thing repeatedly w/ no positive results - soft resets to infinity. While it doesnt cause bfgminer to restart constantly it can severely hamper mining performance.

You can attempt to lower the clocks on those dies and up the volts(which is stock already and I wouldnt recommend going any higher). Also, like AmDD stated, it may be an overheating issue of those dies and its causing them to become unstable, so reapplying therm paste may do the trick. Theres no way to tell the actual temps of the dies on these Titans.

Also, if you want you can save the log files and email them to me via my website.
Lastly, hitting "CLEAR" will unflag those dies and next time bfgminer runs it will retest them.

AmDD

legendary

Activity: 1027

Merit: 1005

Quote from: carlosmnk on July 11, 2016, 10:58:26 PM

Hi, i have a problem with some cubes; strangelly, they suddenly appear a message that sais "Dies flagged for bypass of health checking: 3:1 3:2 3:3 4:1 6:1" and theese dies are not hashing and i can do nothing to fix it (reboot, change the controller, change the psu, change config...). Here is summer, but i have them with a low configuration to mantain them below 70-75ºC, so i don't think it could be because it is too hot. Here is a pic:

PS: I'm using GenTarkin 1.0.3 and 1.0.4 (i tryed both)

[IMG ]http://i68.tinypic.com/2moceb4.jpg[/img]

I am worried about theese cubes...

Im no expert by any means but I get the same thing every few months on my titan. WHen this happens I normally re apply the thermal paste. So far this has solved my issue every time but I cant promise it will be the same for you.

Tupsu

legendary

Activity: 1218

Merit: 1003

Quote from: carlosmnk on July 11, 2016, 10:58:26 PM

Hi, i have a problem with some cubes; strangelly, they suddenly appear a message that sais "Dies flagged for bypass of health checking: 3:1 3:2 3:3 4:1 6:1" and theese dies are not hashing and i can do nothing to fix it (reboot, change the controller, change the psu, change config...). Here is summer, but i have them with a low configuration to mantain them below 70-75ºC, so i don't think it could be because it is too hot. Here is a pic:

PS: I'm using GenTarkin 1.0.3 and 1.0.4 (i tryed both)

http://i68.tinypic.com/2moceb4.jpg

I am worried about theese cubes...

I had exactly the same thing with GenTarkin software.
Better use Original KNC software with the same volts.

I did this and it is no problem any more. When i replace the memory card to GenTarkin software , the problem is back.

carlosmnk

hero member

Activity: 572

Merit: 506

Hi, i have a problem with some cubes; strangelly, they suddenly appear a message that sais "Dies flagged for bypass of health checking: 3:1 3:2 3:3 4:1 6:1" and theese dies are not hashing and i can do nothing to fix it (reboot, change the controller, change the psu, change config...). Here is summer, but i have them with a low configuration to mantain them below 70-75ºC, so i don't think it could be because it is too hot. Here is a pic:

PS: I'm using GenTarkin 1.0.3 and 1.0.4 (i tryed both)

I am worried about theese cubes...

GenTarkin

legendary

Activity: 2450

Merit: 1002

Quote from: lightfoot on July 11, 2016, 06:47:05 PM

I know. But every Bridge Board I have fixed came with a Rpi that was dead shorted on the CPU. Those things are garbage for 24*7, they should have stuck with the beaglebones. So what we really should do is just port the Titan code over to the BB. Which I would do but I keep being too lame to finish it.

I don't mind the work or bitcoin, but I would prefer if people never had the problem in the first place and could mine like hell. :-)

Interesting, Ive had many pi's over the years. Not one single issue.

GenTarkin

legendary

Activity: 2450

Merit: 1002

Quote from: adaseb on July 11, 2016, 06:49:41 PM

Original MSRP: $5000
Current Value: $0

This is how all crypto-mining ASICs ultimately end up.

adaseb

legendary

Activity: 3808

Merit: 1723

Well I finally junked my KNC Jupiter.

Had the miner for a little over a year, almost ran for a year. Bought it second hand really cheap. Made ROI many many times over.

Had the worse W/GHS out of any miner I had in the past (1.333W/GHS).

Parted out the controller for people who needed one for their Neptune/Titans
Parted out the fans and they are cooling my 280X which had dead fans.
Parted out the PCIe extension cables for many GPU mining projects.

Everything else was junked.

Original MSRP: $5000
Current Value: $0

So sad...

lightfoot

legendary

Activity: 3164

Merit: 2258

I fix broken miners. And make holes in teeth :-)

I know. But every Bridge Board I have fixed came with a Rpi that was dead shorted on the CPU. Those things are garbage for 24*7, they should have stuck with the beaglebones. So what we really should do is just port the Titan code over to the BB. Which I would do but I keep being too lame to finish it.

I don't mind the work or bitcoin, but I would prefer if people never had the problem in the first place and could mine like hell. :-)

GenTarkin

legendary

Activity: 2450

Merit: 1002

Quote from: lightfoot on July 11, 2016, 05:37:35 PM

Quote from: GenTarkin on July 11, 2016, 01:07:03 PM

Question...

Would anyone find it useful to have a checkbox in SYSTEM tab something to the effect of "boost rpi performance"
It would apply a mild OC to the pi that *should* work on all pi's the Titan uses.
Ive noticed w/ the rpi @ 1ghz vs stock and a few other items OC'd ... the CPU is not always pegged at 100% when Titan is mining, like stock is.

Given that blowing the RPI Cpu will result in the bridge board blowing I would say not a great idea. :-)

the CPU on the rpi can, in most cases, hit 950mhz w/o even a voltage bump. Its when people start crammin a ton of voltage at rpi's CPU that it dies.
But thats cool =) , less work for me. Ive been running pi @ 1ghz w/o issue for 3/4 a year now.

lightfoot

legendary

Activity: 3164

Merit: 2258

I fix broken miners. And make holes in teeth :-)

Quote from: GenTarkin on July 11, 2016, 01:07:03 PM

Question...

Would anyone find it useful to have a checkbox in SYSTEM tab something to the effect of "boost rpi performance"
It would apply a mild OC to the pi that *should* work on all pi's the Titan uses.
Ive noticed w/ the rpi @ 1ghz vs stock and a few other items OC'd ... the CPU is not always pegged at 100% when Titan is mining, like stock is.

Given that blowing the RPI Cpu will result in the bridge board blowing I would say not a great idea. :-)

GenTarkin

legendary

Activity: 2450

Merit: 1002

Question...

Would anyone find it useful to have a checkbox in SYSTEM tab something to the effect of "boost rpi performance"
It would apply a mild OC to the pi that *should* work on all pi's the Titan uses.
Ive noticed w/ the rpi @ 1ghz vs stock and a few other items OC'd ... the CPU is not always pegged at 100% when Titan is mining, like stock is.

Searing

copper member

Activity: 2898

Merit: 1465

Clueless!

Quote from: Micky25 on July 10, 2016, 01:04:50 PM

a little bit unrelated but, nice furniture! What kind of wood is this?

its an old walnut table I think....perhaps I should look into it getting refinished...heh..I myself thought it was kinda
beat up and junky...

man these have to be the most off topic posts EVER on the Swedish ASIC miner thread

Back on topic. Here are pics of my very dirty cubes and such along with the parts list etc needed to get them all up to speed.

The below link is on the Swedish Asic Hacking thread

https://bitcointalksearch.org/topic/m.15544737

Screen shots of temps before and after the Titan cube mods with fan improvement. Check it out. Hope it helps.

TXSteve

sr. member

Activity: 342

Merit: 250

Quote from: Micky25 on July 10, 2016, 01:04:50 PM

a little bit unrelated but, nice furniture! What kind of wood is this?

beautiful table, maybe oak, walnut stain?

nice shiny centerpiece too

Topic: Swedish ASIC miner company kncminer.com - page 16. (Read 3049520 times)