[WORLDWIDE]Bitmine - 950 GH/S 1500 USD - Next day delivery.

[Collider]

DEVS
ASC   Name   ID   Temp   MHSav   MHS5s   Accepted   Rejected   Utility   ShareTime   Total_MH   Diff1Work   LastValidWork   DHardware%   DReject%
0   BA1   0   0.00   240G   240G   227   0   6.50   21:02:59   503.82T   441   1398.43M   1.5625   0.0000
1   BA1   1   0.00   240G   240G   252   0   7.22   21:03:05   502.54T   485   1398.43M   2.6104   0.0000
2   BA1   2   0.00   244G   243G   238   0   6.81   21:02:56   510.29T   436   1398.43M   4.1758   0.0000
3   BA1   3   0.00   243G   244G   246   1   7.04   21:03:02   509.8T   475   1398.43M   1.0417   107.7895

But pool reports ~ 950, so thats okay.
Definately better than the dragonminer firmware.
(stats are at 950 frequency)

Reject metric per blade seems defective, as it cant possibly have 107% rejects... and it varies between the blades when i do a restart.

[tzortz]

Image definately works, as to how high the actual hashrate at the pool is, i dont know yet.

What is yours?
Also, how high did you set the chip frequency?

In my unit, every "blade" only has 6-pin pcie plugged in, but there is a 8-pin connector and counterpart available for each board.

Is it plugged in in your units?

Nice, everything same as mine.

Yes, it is only 6-pin connected.
 I could see one of the pins was NOT soldered of one of the blades. I soldered it and now has fewer errors.

[tzortz]

I wonder how could we see if there is some overheating issue.

NO TEMPS AT ALL!   

[Collider]

did you solder them on the pcb?

[tzortz]

did you solder them on the pcb?

Yes, I did.

I will post pictures.

[tzortz]

[gallery2000]

bought two units on Monday but no tracking.  Please send tracking.

[tzortz]

Email them, they won't see it here.

[Collider]

Temps should be absolutely fine on these, the heatsinks are massive and on both sides of the chips, with nice airflow through them.
Normal chips can handle 119°C 24/7 without damage. But these are probably at far lower temps.

I just dont get why they didn´t connect all 8 pins of the pcie to the board, as they clearly are running ~250W per board and have the 8pin pcie connectors on hand.
The production cost savings of this must be under 1$ per unit, and it makes the design more stable.

Anyway, the custom firmware works atleast 10% better and more stable than the dragonminer firmware.

[tzortz]

Also I believe there is some software issue not letting it run at 1th/s + speed.

It is not the PSU as mentioned to some article.

[Collider]

Yes, i think only the original firmware controls the right amps for the dc/dc, when there is no input like with the dragonminer firmware, there will be stability issues, hence lower hashing.

[tzortz]

I used a server psu of 2000W for an hour, but could not see any difference in the hash rate. Nor HW.

It is exactly as you said, the DCDC need to be calibrated somehow by its software.

[Collider]

Did you try above 950mAmp?
Or was it the clock frequency you could set at startup?

[tzortz]

No , I did not change frequency, that software does not allow to do that.

Right?

[Collider]

I thought you could edit it at startup but i didn´t try. Maybe it is just fixed and only looked like it could be edited.

[s1lverbox]

Did they come back in stock? .

[tzortz]

Not yet, we need to contact them through email.

They wont see anything here.

[finlof]

bought two units on Monday but no tracking.  Please send tracking.

same here.  i was expecting my 3 units no later than today, but still dont have tracking.

[tzortz]

Oops, thats bad news. 

[gallery2000]

Did they come back in stock? .

No, I bought them before they were sold out.