Topic: KnCMiner Jupiter Miner First Impressions - page 15. (Read 65993 times)

yes, the one miner I didn't touch which has one lucky board with 8 VRMs runs 70 watts less and is still on .93 and ~520 Gh

The other is a mess now and but I have it running .94 and back to ~410Gh
I have some more theories to prove out so I will try a few things later tonight with this one

The power supply can't supply more power than the DC modules can convert.

4 DC modules @ 40A each = 160A.  Chip nominal voltage is 0.750 VDC so that is 120W output per board.  By spec the DC converters are about 90% efficient that means 133W input.  If you assume they are running hot and thus less efficient maybe 150W input.  4x 150W = 600W of 12VDC.  Throw in another 6W per fan * 6 fans + 10W for the host you are looking at <650W.  On the firmware which overvolts the chip to 0.9V you are still talking <770W.

Now the 8 VRM version might be able to push the chips harder given enough power.

He means remaining on 0.93 firmware.  The VRMs are running very hot but they provide a higher throughput (even when derated for higher HW error% and reject%).

What do you mean by 'baking'?

How they come back the dead core?
What they do?

Maybe if we put more power psu like 1000w will have solution?

@ Sitarow:  let us know which front fans you get, I am shopping as well.   You have the good early miner so I would say no one will have better conditions than you for jupiters.
That is why I am planning on making two saturns from my one jupiter.  If I can't get another controller from KNC, I will atleast cut the case in half (front and back) then put the back part 90 degrees like |_ so the ASIC cables can still connect from the separated case.  I can make a makeshift grill to hold new case fans in front of it


@ jelin:  It is a 850psu that was running 770 watts stable.  Some say they got their dead dies to come back, we'll see if I play around with that as solo sometime

http://www.microcenter.com/product/410331/Hale90_Series_V2_850_Watt_Modular_Power_Supply


@ ecliptic: I would like KNC to explain the decision more.  The VRMs had their caps and everything around removed from the board so they sure saved a few nickles at our expense.  I am still baking my 4 VRM boards since they give better performance than using .95

Those regulators are speced for a max current of 40A each....

Has anyone pointed out that's a good suspect for the VRMs and caps failing?

Update 46 hours in.

Case cover on for last 6 hours ambient room temperature 21°C

HW Errors climbing however seems system is stable.



Next I will swap out the 2 front case fans with higher cfm fans.

DPoS

what psu have at dead core jupiter?
850 or more?
corsair or somthing else?

Your results are what I have noticed as well now that I have put the cover back on and using the stock fans for ventilation. However my ambient is 20 -21 ℃.

My Device temps are similar.

When I put the case cover back on. HW % went up from 4.2% to 4.48% and still going up.

The miner I have to test with has #1 stickers on them.

have you ever remove you jup cover? if yes did you check which sticker they have? if  my theory is correct all of your boards would have a #1 sitcker.

it's not a theory in my miners all boards with siticker #1 are almoste perfect. I had one with sticker #3 and it has one die with all 48 cores disabled.

and again to show the pre 95 state of a miner that had one board with 8 VRMs and lower wattage against the other boards that were left to run higher and hotter

Not sure as to they why.

However if I were to speculate as to the why, I would look at the possibility that the added components may help dissipate heat from the PCB and potentially alleviate the other components thermal stress.

Edit: I forgot to add the possibility that component resistance may have been another factor. With the reduction of components one can test if supplying the vrm's with increased power potentially making the default firmware settings a bit high.

Would be great to have an official statement. However that opens up liabilities in an already turbulent situation.

do not take your poor comprehension and place it in my words ok?   If you have trouble with deductive reasoning then leave that at your feet.

How many times do I have to post the same picture showing that one miner that has 3 boards with 4 VRMs and one board with 8 VRMs show a difference on their voltage readings?  You say they are 'off' on the 8 VRM boards but yet the 4 that are on, run cooler than the boards with just 4 pre .95 firmware.  

.95 firmware lowered the voltage on the 4 VRM boards which prior to its release, were left to overclock after 4 were removed without any testings of consequence.

If those 4 form the 8 were 'not doing anything' as you say, then why would the 4 that remain increase ~20C on each?   They produced more amps as well so yes that is overclocking.  So you and KNC tell me why and how ok?  Obviously KNC saw that as a problem and released .95 after all the reports on overheating and need for cooling.

The performance of .93 & .94 outshine the performance of .95 on boards with 4 VRMs.  I have posted those pictures as well and others have switched back too.  People rather run the boards with higher voltage/amps (overclocking but you hate that term) and cool them to prevent as much of them failing than to run a gimped .95 that does nothing to keep more cores running.

You get more WU with .93 & .94 than with .95.  The cooling helps reduce HW errors regardless if you ignore that reality.


You can still make up non arguments if you like. Perhaps you were the person that made the cavalier decision and can't face the music

2.5hours later hardware errors % goes up and speed goes down. Also power draw is at 563 watts @ 120v.

Here are the results. Firmware v0.9.5 first 24 hours ambient temp at 17°C with open case and house fan.

The last 2.5 hours case on and cooling/airflow stock from kncminer ambient room temp is 21°C

8 VRM Model.

The explanation is simple: the VRMs works just perfectly, and their status is OK, it is there for everyone to see.
What does not work is the chip communication so all units are disabled for some reason. No big deal.

But no amount of VRMs will ever change your hashrate. Your theory is: use backup VRM to provide more hashpower... such nonsense!

The VRMs work just perfectly.

P.S. apart of that thing with their programming, which is solely the fault of GE, who deserve their balls kicked. Please aim your ire at them. They deserve it. And they don't pay any taxes either.

If he told you. it wouldn't be a secret.   

Can you direct me to the settings of the clocking? I can not find it absolutely anywhere! I would like to change any clocks and timings on the boards, you claim to have done so, please show me how!
I'm not interested in better management software, I'm interested in overclocking, please give me the secret how to do that!

Can you explain to us, how on earth VRM modules which were NEVER turned on could drastically improve performance? The only difference between 4 VRM and 8 VRM modules is that 4 VRM module boards have less passive components (capacitors) mounted.

So I would love to hear your conspiracist expanation on how the non-functional 4 VRM worked. Some people with 4 VRMs have netter results than you have with 8 VRM, yet you believe it is not the inherent property of the hashing chip but a property of the VRM?
Interesting...