Author

Topic: [ANN] Spondoolies-Tech - carrier grade, data center ready mining rigs - page 299. (Read 1260290 times)

legendary
Activity: 3234
Merit: 1220
start counting!
approx 8*3*3*6 (plus another 8*3) = ~450-460 units. This is one building, apparently 3 were in the compound

Has there been any investigation as to the cause? I know the owners are keen to put it behind them and move on. Personally I don't see it being caused by a miner. But then the finished pictures paint a much more professional setup than it looked in earlier pics, or the aftermath pics.

What I haven't seen is any pictures of the other side of the miners before the fire, the power side of the racks. That is what most people seem to be interested to see, it does seem a bit strange that there isn't anything showing that part of the setup, because for a minimal cost mine the finished photos like a damn site better finished than most.

Yes they didn't have many things that a professional datacentre should have, but then thats not what its about, you can't achieve ROI if you build or hire a professional datacentre for mining anymore.

I'd still like to see the business end of the racks and detail of how they were wired up, but I suspect that won't happen.
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
start counting!
approx 8*3*3*6 (plus another 8*3) = ~450-460 units. This is one building, apparently 3 were in the compound
legendary
Activity: 3234
Merit: 1220
Between 475 and 500
hero member
Activity: 742
Merit: 500
legendary
Activity: 1904
Merit: 1007
After the fire in Thailand on the mining farm.

How can they make up this fictitious amount for 100 Spoondolies?

Spondoolies-Tech confirmed that much of the hardware used in the facility was sourced from them.

The manufacturer's CEO, Guy Corem, told CoinDesk that he had visited the site in late June, when it was primarily populated by about 1,000 Dragon miners and around 100 Spondoolies-Tech miners.

Corem said the facility was expanded shortly after his visit, with hundreds of Spondoolies-Tech SP30 miners. These had operated for more than two months without incident and technical data indicates they were cooled properly, he added.

'Kosher' operation

There has been no official word on the extent of damage suffered by the facility. It is estimated that the farm housed upwards of 2,000 miners when it was destroyed, including about $2m-worth from Spondoolies-Tech alone.

100 sp30 at 2 million is 20k each.  Someone is sharpening the pencil in accounting.

Pay attention to timing please. There were no SP30 in June. Those are SP10 miners...
hero member
Activity: 882
Merit: 1003
After the fire in Thailand on the mining farm.

How can they make up this fictitious amount for 100 Spoondolies?

Spondoolies-Tech confirmed that much of the hardware used in the facility was sourced from them.

The manufacturer's CEO, Guy Corem, told CoinDesk that he had visited the site in late June, when it was primarily populated by about 1,000 Dragon miners and around 100 Spondoolies-Tech miners.

Corem said the facility was expanded shortly after his visit, with hundreds of Spondoolies-Tech SP30 miners. These had operated for more than two months without incident and technical data indicates they were cooled properly, he added.

'Kosher' operation

There has been no official word on the extent of damage suffered by the facility. It is estimated that the farm housed upwards of 2,000 miners when it was destroyed, including about $2m-worth from Spondoolies-Tech alone.

100 sp30 at 2 million is 20k each.  Someone is sharpening the pencil in accounting.
hero member
Activity: 546
Merit: 500
Owner, Minersource.net
hero member
Activity: 818
Merit: 1006
You may want to change the fan speeds as well. I'm not sure yet what the optimum is...

I did a little experiment with SP30s. I took 10 machines, and gave the even-numbered ones a fan speed setting of 80 to start with, and the odd numbered ones a fan speed of 70. I then waited an hour, and recorded their temperatures and performance (from monitor.php). I then switched them up (odd 80, even 70), waited another hour, and recorded again. Here's the result:

Code:
jtoomim@server:~/wissamperf$ python ~/fan_comparison.py 
==== phase1 ====
machine-name timestamp hashrate front top bottom is_odd
wissamitter-10 1415393491 4601120 17 70 68 0
wissamitter-1 1415393491 4593400 17 78 75 1
wissamitter-2 1415393491 4585680 16 71 69 0
wissamitter-3 1415393491 4582785 17 79 74 1
wissamitter-4 1415393491 4560590 17 72 68 0
wissamitter-5 1415393491 4576995 17 75 73 1
wissamitter-6 1415393491 4596295 16 70 70 0
wissamitter-7 1415393491 4574100 17 73 71 1
wissamitter-8 1415393491 4562520 17 69 67 0
wissamitter-9 1415393491 4606910 17 78 74 1
==== phase2 ====
machine-name timestamp hashrate front top bottom is_odd
wissamitter-10 1415396747 4591470 17 73 72 0
wissamitter-1 1415396747 4585680 16 73 69 1
wissamitter-2 1415396747 4533570 16 74 73 0
wissamitter-3 1415396747 4576030 16 72 68 1
wissamitter-4 1415396747 4558660 16 74 71 0
wissamitter-5 1415396747 4571205 16 68 66 1
wissamitter-6 1415396747 4596295 16 74 73 0
wissamitter-7 1415396747 4562520 17 68 68 1
wissamitter-8 1415396747 4568310 17 72 71 0
wissamitter-9 1415396747 4581820 16 72 69 1
==== Summaries ====
 === Evens ===
  == phase1 ==
{'count': 5, 'front': 16.6, 'rate': 4581.241, 'top': 70.4, 'bottom': 68.4}
  == phase2 ==
{'count': 5, 'front': 16.4, 'rate': 4569.661, 'top': 73.4, 'bottom': 72.0}
 === Odds  ===
  == phase1 ==
{'count': 5, 'front': 17.0, 'rate': 4586.838, 'top': 76.6, 'bottom': 73.4}
  == phase2 ==
{'count': 5, 'front': 16.2, 'rate': 4575.451, 'top': 70.6, 'bottom': 68.0}
 == High speed ==
{'count': 10, 'front': 16.4, 'rate': 4578.346, 'top': 70.5, 'bottom': 68.2}
 == Low speed ==
{'count': 10, 'front': 16.7, 'rate': 4578.2495, 'top': 75.0, 'bottom': 72.7}

That's 4578.346 GH/s at 80, and 4578.2495 GH/s at 70.

Conclusion: if your intake temps are around 16C, and you're using September SP30s, fan speeds around 70 or 80 seem to be equally good.

Source code available upon request.
sr. member
Activity: 473
Merit: 250
If you are having weird behavior in SP3x : upgrade to 2.4.9. If all works well - don't.
Good day.

I take it you mean 2.5.9.

Confirmed: 2.5.9 is available
hero member
Activity: 818
Merit: 1006
If you are having weird behavior in SP3x : upgrade to 2.4.9. If all works well - don't.
Good day.

I take it you mean 2.5.9.
hero member
Activity: 560
Merit: 500
If you are having weird behavior in SP3x : upgrade to 2.4.9 2.5.9. If all works well - don't.
Good day.


Don't upgrade to 2.5.7 and beyond. Stay on 2.5.6 till the end of weekend at least - it is the last of the 2.5.x series.
Once I will release 2.6.X - upgrade to it only.
legendary
Activity: 1904
Merit: 1007
I've measured it at about 10 GH/s per 1C for SP30s, or around half what you indicate. I've measured this by taking a machine out of our racks (on around 14C air, IIRC) and sticking it into the coldest spot of our hot aisle (about 26C at the time, IIRC), and seeing about 120 GH/s drop in the hashrate. My memory is a little fuzzy on the exact numbers we got from that experiment, but it's accurate to +/- 20%. I've also done other measurements by looking at diurnal variation in performance on SP30s with similar results.

Man you spend too much time in your DC  Roll Eyes
hero member
Activity: 728
Merit: 500
I also have a SP31 that won't give more than 4.7Th/s. I am beginning to think this is the maximum, a far cry from the advertised 5.5Th/s.
My unit is running @ 240v and close to 3Kw.

While it was advertised for 5.5Th/s I think you missed Guy's update:

•    The hash rate of the SP31 will be 4.9TH/s +- 5%. Power consumption will remain the same 3kW at the wall as in the previous SP30.
That update means fuck all after the fact, Guy once again overstated and under delivered. Was it another slow corner batch this time? Such an arrogant bunch ...
hero member
Activity: 818
Merit: 1006
what are your temps like.  Even the SP30 looses close to 200TH it seems per 10deg temp rise
I've measured it at about 10 GH/s per 1C for SP30s, or around half what you indicate. I've measured this by taking a machine out of our racks (on around 14C air, IIRC) and sticking it into the coldest spot of our hot aisle (about 26C at the time, IIRC), and seeing about 120 GH/s drop in the hashrate. My memory is a little fuzzy on the exact numbers we got from that experiment, but it's accurate to +/- 20%. I've also done other measurements by looking at diurnal variation in performance on SP30s with similar results.

SP10s are much more sensitive to temperature. They also vary in performance by about 10 GH/s per 1C, despite having less than 1/3 the total hashrate.

I've heard that SP20s are more sensitive to temperature than SP30s, but I don't have any here to test.

Preliminary observations with SP31s (as well as predictions based on what I know of semiconductor physics) suggest that they're more temperature sensitive than the SP30s, but I haven't done any systematic measurements yet.
hero member
Activity: 546
Merit: 500
I have a question. would it be possible to put say 3xSP31 inside of one of these:

http://www.acoustiproducts.com/en/ucoustic_active.asp

I would have the ducting setup aswell so that i can vent the hot air out of the building, i am also in a cold climate. I am sure that there must be many people here intrested in making it easier living with these machines. Grin

Yes, this has been done with 2-3 spondoolies

Thankyou  Smiley
hero member
Activity: 818
Merit: 1006
Anyone with SP31 getting more than 4.7Th/s?

We have 13 SP31s here at Toomim Bros Bitcoin Mining Concern right now. With 16C intake temps, we're currently seeing the lowest performer getting 4743 GH/s, highest 4915 GH/s, mean 4835 GH/s.

Do this from the ssh console:

Code:
echo "1' > /etc/mg_ignore_110_fcc

http://www.spondoolies-tech.com/blogs/technical-blog

If you have a lot of machines to configure, you can also scp it over. Example if your machines are 192.168.1.5 through 192.168.1.8:
Code:
echo "1" > mg_ignore_110_fcc
for ip in 5 6 7 8; do scp mg_ignore_110_fcc [email protected].$ip; echo "Finished 192.168.1.$ip"; done

You will then want to change the power limit on the PSUs. In my experience, Murata PSUs can handle between 1330W and 1380W when run off of 235-245V in ambient temps no greater than 22C. Keep in mind that when Muratas hit their power limit, they sometimes lock up and need a power cycle to recover them, so you shouldn't increase the limits unless you're going to be nearby for several hours.

I also recommend using starting voltages of 650 mV or 645 mV with SP31s, not the default 655 mV, since 655 is too close to the settled voltage for some SP31s. You may want to change the fan speeds as well. I'm not sure yet what the optimum is, but I suspect it's the setting that will eliminate thermal throttling on the loop 4 and 9 ASICs, but no higher.
legendary
Activity: 1904
Merit: 1007
I also have a SP31 that won't give more than 4.7Th/s. I am beginning to think this is the maximum, a far cry from the advertised 5.5Th/s.
My unit is running @ 240v and close to 3Kw.

While it was advertised for 5.5Th/s I think you missed Guy's update:

•    The hash rate of the SP31 will be 4.9TH/s +- 5%. Power consumption will remain the same 3kW at the wall as in the previous SP30.
sr. member
Activity: 586
Merit: 251
Anybody with a SP31 found a 'sweet spot', can't get more than 4.7THs out of mine.

Anyone with SP31 getting more than 4.7Th/s?

are you on 240 V ? what are your temps like.  Even the SP30 looses close to 200TH it seems per 10deg temp rise

I also have a SP31 that won't give more than 4.7Th/s. I am beginning to think this is the maximum, a far cry from the advertised 5.5Th/s.
My unit is running @ 240v and close to 3Kw.
legendary
Activity: 2408
Merit: 1102
Leading Crypto Sports Betting & Casino Platform
Anybody with a SP31 found a 'sweet spot', can't get more than 4.7THs out of mine.

Anyone with SP31 getting more than 4.7Th/s?

are you on 240 V ? what are your temps like.  Even the SP30 looses close to 200TH it seems per 10deg temp rise
legendary
Activity: 2408
Merit: 1102
Leading Crypto Sports Betting & Casino Platform
wow so I got my SP 30 and the boards fell on top of each other ! don't know if it was in shipping or whenever I have a bad feeling this thing is DOA

https://www.dropbox.com/s/e02gb3gjxowvaek/WP_20141106_18_23_47_Pro%5B1%5D.jpg?dl=0


https://www.dropbox.com/s/uod47xkjbfqri91/WP_20141106_18_23_56_Pro%5B1%5D.jpg?dl=0


I guess now I have to take it apart and restack the boards I'll post more pics in a few


I had the same problem with one of mine as well and after disassembly, I found a chunk had been taken out of one of the VRMs.

I'll post pic's as well.

ouch !
I was a bit luckier , no actual damage to the boards despite the cracked corner.  The SP20 is a cute and sexy little machine but it definitly not worth
$1300 (with shipping)  Best hash rate I think at 110V (assuming you have great cooling)  

Looks like maybe 1.8 TH max at close to 300 watts per power slot.  

right now at 24c front and 79c and 80c  back it hashes around 1650 TH max while pulling over 1200 watts

(it actually spiked at one point and blew out my power meeter and maybe one of my PSU's , good thing i have spares)

maybe it does a bit more at 240V i have some ic14 to ic13 cables comming early next week and ill try them on a dedicated PDU and see what happens
Jump to: