43THs RT s9??? WTF...

[NotFuzzyWarm]

This started about 15 min ago, around 6:55p Eastern
Note card-3 (chain 4)


After a soft boot STILL card 3!


WHAT in the HELL???

[philipma1957]

What does the pool read?

[NotFuzzyWarm]

That's the odd thing. Just like the avg speed is normal, Kano reported normal speed as well. Still rebuilding reported speed from the soft boot but seriously doubt that is 'real' paid speed. Damn I WISH...

Reject/stales/discards all low.
Still going on. Bitmain GUI shows 37THs with card 3 being more than 28 of it.
again, WTF???

Gonna let it ride for a bit before doing a hard boot.

[fvineyard]

I seem to recall someone else posted about the same thing happening. A single board registering high hash numbers. I do not recall if the error rate was low or not.

[NotFuzzyWarm]

Just did a hard power cycle. After 5 min running, chain-4 is still insane....


As a side note -- and I seriously doubt this has anything to do with it but -- this is the miner I gave fubly's crack a run on and had to eventually reflash with Bitmains first s9 update to get running.

[in2tactics]

You probably have a few rogue dust bunnies in there.

As a side note -- and I seriously doubt this has anything to do with it but -- this is the miner I gave fubly's crack a run on and had to eventually reflash with Bitmains first s9 update to get running.

How can you be sure?

[NotFuzzyWarm]

You probably have a few rogue dust bunnies in there.

As a side note -- and I seriously doubt this has anything to do with it but -- this is the miner I gave fubly's crack a run on and had to eventually reflash with Bitmains first s9 update to get running.

How can you be sure?

Can't be 100%.
But - considering the reflash wiped and over wrote everything including the Beagle Bone's flash memory, along with it running perfectly for >48hrs after I did the reflash, highly unlikely but still to be noted.

Just tried lowering to 550MHz, same thing. Now back up to 600MHz to just let it run.

[Finksy]

Quick, post it for sale in the hardware section w/ screenshots of the GUI!   

[NotFuzzyWarm]

Quick, post it for sale in the hardware section w/ screenshots of the GUI!  

No matter what, it has always been a fast one, even down clocked 606 from the stock 650 the batch-1's were sold at Kano has always reported it as >13.6 THs, even now after the reboots Kano still says 13.2  so can't complain at all  

Side note to the fun. Decided to get Bitmain's latest s9 650 sec patch.
Though it bricked it.......
Could not reach the miner at all. Looked at it, all lights merrily blinking away.
 
On a random thought, decided to check my routers connected devices.
The security patch kept the Host name but made DHCP assign a new connection... rolf.
Got a glass of wine, hit my MM, pointed Awesome Miner and my FF link at the new addy. All is well.

Aside from chain-4 still reporting >17THs and very high HW errors that is  
edit: now 10 min after a soft boot to check something, >28THs for the board.
edit of edit: 20 min in, RT says >29THs for the board, Kano pegs miner average at it's usual >13.pointsomething so....
Guess keep on running until Majik Smoke escapes

[NotFuzzyWarm]

Anyone here know what HW errors can be and their effect on reflected throughput?

My guess maybe that if >threshold HW errors occur be it from cores failing or whatever, maybe the ASIC chain reports the need for new data (to try again) and maybe bmminer reports that request as 'Real Time' speed for new hash data to the board?

Just a wild thought of course as I have ZERO knowledge of how the Wizards (-ck and Kano plus those they built upon) Behind the Curtain does the Voodoo we see nor of actual flow through the dies themselves (2112 watching?)... Plus of course whatever Bitmain did to the code...

[NotFuzzyWarm]

Quick, post it for sale in the hardware section w/ screenshots of the GUI!  

This is even better. Just did another soft boot and...


69THs.... <jaw drop>
Just WTF is the bmminer code looking at for a boards RT speed?

[NotFuzzyWarm]

Ok... The miner just rebooted itself.
Getting late so pulled power and control cable from card-3 (chain-4), set remaining 2 cards to 625 and calling it a night for now on it. Got an idea but need meter from work to poke around and try something. (rather like the cobbler's children have no shoes...)

[HerbPean]

I suspect that the HW just mess up with the hasrate calculation of the board #3 (43hw in 36 sec)

Just guessing (software issue)

[tbonetony]

Quick, post it for sale in the hardware section w/ screenshots of the GUI!  

This is even better. Just did another soft boot and...


69THs.... <jaw drop>
Just WTF is the bmminer code looking at for a boards RT speed?

So all of a sudden your board #3 evolved to a quantum hashing blade despite it only appears so within its enclosure....

[Finksy]

On this topic, I've had machines (S7-B5) with negative HW error shown on the GUI before.  Clearly this isn't possible, but I always wondered if it messed with the actual hashrate...

[Xircom]

Fuzzy, happend to me as well before the boards died for me. 2 boards did the same on 2 different miners. Both are back for repair 

[NotFuzzyWarm]

Fuzzy, happend to me as well before the boards died for me. 2 boards did the same on 2 different miners. Both are back for repair  

Figured that as >1HW error per-second is not good.... Tonight going to swap control ports to verify that this follows the board. Then time to contact the folks in Colorado at BitmainWarranty to get the repair/replace process rolling.

[tbonetony]

Fuzzy, happend to me as well before the boards died for me. 2 boards did the same on 2 different miners. Both are back for repair  

Figured that as >1HW error per-second is not good.... Tonight going to swap control ports to verify that this follows the board. Then time to contact the folks in Colorado at BitmainWarranty to get the repair/replace process rolling.

Is BitmainWarranty more trustworthy and reliable than Bitmain? Just wondering. I have 2 S9's so would like to have a backup plan as well.

[NotFuzzyWarm]

Fuzzy, happend to me as well before the boards died for me. 2 boards did the same on 2 different miners. Both are back for repair  

Figured that as >1HW error per-second is not good.... Tonight going to swap control ports to verify that this follows the board. Then time to contact the folks in Colorado at BitmainWarranty to get the repair/replace process rolling.

Is BitmainWarranty more trustworthy and reliable than Bitmain? Just wondering. I have 2 S9's so would like to have a backup plan as well.

Yes. I've dealt with them only once before for a DOA s7, good communications and fast turn around. You can also check philipma1957 thread here on service.
I already have a ticket open with BitmainWarranty and the board will be mailed tonight or tomorrow. Yes I am doing it as a paid repair with cost depending on what they find.

Using USPS Standard Shipping box so $14 for shipping plus whatever for insurance as $450 value
Beats the hell out of around $95 or more I paid to send a board to Hong Kong.

[tbonetony]

Fuzzy, happend to me as well before the boards died for me. 2 boards did the same on 2 different miners. Both are back for repair  

Figured that as >1HW error per-second is not good.... Tonight going to swap control ports to verify that this follows the board. Then time to contact the folks in Colorado at BitmainWarranty to get the repair/replace process rolling.

Is BitmainWarranty more trustworthy and reliable than Bitmain? Just wondering. I have 2 S9's so would like to have a backup plan as well.

Yes. I've dealt with them only once before for a DOA s7, good communications and fast turn around. You can also check philipma1957 thread here on service.
I already have a ticket open with BitmainWarranty and the board will be mailed tonight or tomorrow. Yes I am doing it as a paid repair with cost depending on what they find.

Using USPS Standard Shipping box so $14 for shipping plus whatever for insurance as $450 value
Beats the hell out of around $95 or more I paid to send a board to Hong Kong.

Thanks for your info. I will keep them on my list too.