[ANN] Spondoolies-Tech - carrier grade, data center ready mining rigs

[Marvell1]

I'd love to know at what point it becomes uneconomical  for the corporates currently piling in and they pull the plug.
I imagine small miners have more endurance for making losses than they do, not that it helps right now...

when the price crashed to around 400s difficulty really flattened for a while we need another crash or two to put these corporates underwater and get their loans due lol

[BitCoin Operated Boy]

It looks like newest Bitfury's DC has got probably 30 PH not 20 PH as it was claimed.

[jimmothy]

It looks like newest Bitfury's DC has got probably 30 PH not 20 PH as it was claimed.

Source?

[Dans]

Collider, jtoomim, Roadsress or anyone with the early machines:

Please post your statistics and whether you achieved sustained ~4.5TH at 208-220V (please post your voltage) and what was the speed at 110/120V if applicable (I assume 4.1Th). Thanks.

I'm getting just under 4.4Th/s @ 205v/205v    
Input temp is 19C

Hoping to tweak a little more and get it up to 4.5, any suggestions (other than what ive already read in this thread, and tried)?

[Minor Miner]

I'd love to know at what point it becomes uneconomical  for the corporates currently piling in and they pull the plug.
I imagine small miners have more endurance for making losses than they do, not that it helps right now...

There is a VERY large difference between sunk costs and new capacity.
We are still running avalons from batch 1 and 2.   Would we buy another one of those?  No.
But as long as the money returned from mining is more than 1.25X the cost of mining, you would keep a machine operating.  But you would never buy a new machine.   Personally, I would not buy a new machine if I thought it would take more than 100 days to return the capital from the purchase (at bitcoin's vol, you would be crazy to -- buy the coins if you just want to speculate on bitcoin price, the optionality you gain by the liquidity in coins vs. pre-ordering a coin maker is worth a fortune).

So, let's focus on the question of when do you shut down an IN HAND SP10?   You can solve for this but there are two variables that affect the answer.   Difficulty and price (and they counteract each other).   Electric price is there too but that is known (it varies for many people but is known for the user).
Why do I not include 'rent'?  Because the cost is 'SUNK'.   Most people have leased the building and will have to pay it whether the machines are on or off.    Sunk costs are not used to determine what you would do on a variable basis, only the costs that can be eliminated are.

I will use 5 cent per kWh electric since that will allow people to extrapolate for residential and for the people that host in data centers and likely pay around 20 cents total.

Cost of operating 1 SP10 = 1.2kW X 12 days x 24 hours x $0.05 = $17.28 per difficulty adjustment.
So, you should turn this machine OFF when the risk of returning profit gets too high (because the $17.28 is FIXED and will have to be paid if it runs).   I would turn off when the machine produces only $17.28 X 1.25 = $21.60 of bitcoin produced.
To solve for difficulty, let's use $400 as the bitcoin exchange rate.
$400 btc = 21.60/400 = 0.054 bitcoin.
Bitcoin produced in the difficulty period = 43,200
Share of the network required to produce 0.054 bitcoin = 0.054 / 43,200 = 0.00000125
How large would the network need to be to turn the SP10 OFF?  1.4 TH/s / 0.00000125 = 1,120,000 TH/s

[Guy Corem]

Collider, jtoomim, Roadsress or anyone with the early machines:

Please post your statistics and whether you achieved sustained ~4.5TH at 208-220V (please post your voltage) and what was the speed at 110/120V if applicable (I assume 4.1Th). Thanks.

I'm getting just under 4.4Th/s @ 205v/205v    
Input temp is 19C

Hoping to tweak a little more and get it up to 4.5, any suggestions (other than what ive already read in this thread, and tried)?

Please contact zvi@ for a possible tweaking. Zvisha is still tuning the firmware.

[zvisha]

Collider, jtoomim, Roadsress or anyone with the early machines:

Please post your statistics and whether you achieved sustained ~4.5TH at 208-220V (please post your voltage) and what was the speed at 110/120V if applicable (I assume 4.1Th). Thanks.

I'm getting just under 4.4Th/s @ 205v/205v    
Input temp is 19C

Hoping to tweak a little more and get it up to 4.5, any suggestions (other than what ive already read in this thread, and tried)?

Please contact zvi@ for a possible tweaking. Zvisha is still tuning the firmware.

Under 210V we can not pull more then 1250watt from the Emerson because of the FCC limitations on current in wires.  

[Jesus Christ]

Do I get a discount?

[zvisha]

Collider, jtoomim, Roadsress or anyone with the early machines:

Please post your statistics and whether you achieved sustained ~4.5TH at 208-220V (please post your voltage) and what was the speed at 110/120V if applicable (I assume 4.1Th). Thanks.

I'm getting just under 4.4Th/s @ 205v/205v    
Input temp is 19C

Hoping to tweak a little more and get it up to 4.5, any suggestions (other than what ive already read in this thread, and tried)?

Please contact zvi@ for a possible tweaking. Zvisha is still tuning the firmware.

Up to 210V we can not pull more then 1250watt from the Emerson because of the FCC limitations on current in wires. 

BTW, some people here successfully hacked the FW to ignore the FCC limits, and get slightly better performance from the miner, but that voids the warranty and very dangerous if you don't know what you are doing   I am not sure it is worth the 100GH.

[2112]

Up to 210V we can not pull more then 1250watt from the Emerson because of the FCC limitations on current in wires. 

Whose limitation? FCC? It stands for Federal Communication Commission and deals with radio-communication interference.

Could you please look up the actual source authority for the limitation you're mentioning?

[zvisha]

Up to 210V we can not pull more then 1250watt from the Emerson because of the FCC limitations on current in wires. 

Whose limitation? FCC? It stands for Federal Communication Commission and deals with radio-communication interference.

Could you please look up the actual source authority for the limitation you're mentioning?

We work with FCC lab that provide us the FCC compliance. I get the requirements in form of document from our HW team, which is called FCC compliance document, so I call all those limitations FCC compliance limitations.

I for sure did not read all the documents that we need for safety compliance, and have no idea where they come from. I will try and check with HW team where this limitation is from.

[2112]

We work with FCC lab that provide us the FCC compliance. I get the requirements in form of document from our HW team, which is called FCC compliance document, so I call all those limitations FCC compliance limitations.

I for sure did not read all the documents that we need for safety compliance, and have no idea where they come from. I will try and check with HW team where this limitation is from.

Please do look it up. FCC regulations would deal with the things like harmonic emissions from the cables, which could be easily solved by a ferrite choke clamped on them.

In the past I've dealt with the fake/scummy/incompetent "electromagnetic compatibility" labs/inspectors, but never heard of one in Israel.

FCC is a fairly weak organization, as far as "the Feds" go. In the past some enterprising scammers even set up fake "RF emission labs" in an acoustic anechoic chamber that was sufficient to confuse inexperienced people and scam them out of payments for "expert evaluations" that were nearly worthless. In USA this is called "private enterprise" and "public/private partnership".

Edit: Are you by chance dealing with a remote "mail delivery only" lab in the USA, where nobody knowledgeable from your company ever visits the actual lab?

[BitCoin Operated Boy]

It looks like newest Bitfury's DC has got probably 30 PH not 20 PH as it was claimed.

Source?

Google.com

[jimmothy]

It looks like newest Bitfury's DC has got probably 30 PH not 20 PH as it was claimed.

Source?

Google.com

So no source?

[zvisha]

We work with FCC lab that provide us the FCC compliance. I get the requirements in form of document from our HW team, which is called FCC compliance document, so I call all those limitations FCC compliance limitations.

I for sure did not read all the documents that we need for safety compliance, and have no idea where they come from. I will try and check with HW team where this limitation is from.

Please do look it up. FCC regulations would deal with the things like harmonic emissions from the cables, which could be easily solved by a ferrite choke clamped on them.

In the past I've dealt with the fake/scummy/incompetent "electromagnetic compatibility" labs/inspectors, but never heard of one in Israel.

FCC is a fairly weak organization, as far as "the Feds" go. In the past some enterprising scammers even set up fake "RF emission labs" in an acoustic anechoic chamber that was sufficient to confuse inexperienced people and scam them out of payments for "expert evaluations" that were nearly worthless. In USA this is called "private enterprise" and "public/private partnership".

Edit: Are you by chance dealing with a remote "mail delivery only" lab in the USA, where nobody knowledgeable from your company ever visits the actual lab?

No, the lab is in Israel. I am actually flying on a 3 day vacation tonight, but we will check this. It's in our interest to provide as much hashing as possible to every voltage.

[TKeenan]

No, the lab is in Israel. I am actually flying on a 3 day vacation tonight, but we will check this. It's in our interest to provide as much hashing as possible to every voltage.

The equipment hashes 25% less than expected, consumes 30% more than expected, promised units not yet delivered - what to do next?  ...go on a 3 day vacation! 

I wouldn't advertise that.  People waiting are going crazy while the lead guy sits on the beach blowing dope and drinking vodka. 

[jtoomim]

Collider, jtoomim, Roadsress or anyone with the early machines:

Please post your statistics and whether you achieved sustained ~4.5TH at 208-220V (please post your voltage) and what was the speed at 110/120V if applicable (I assume 4.1Th). Thanks.

I now have 4 machines here running on 120V. One is personally mine. The other three belong to wanna4fun, our first customer. (He gave me permission to mention him.)

I have mine set up to exceed Zvi's recommendations for power for the PSUs, and it's getting about 4080 GH/s. I had it getting 4170 GH/s for a while, but that was with a different airflow configuration that allowed me to use more power. The PSUs are currently limited at 1125 W each. It uses the Emerson PSUs. This machine sustained about 4.5 TH/s for 24 hours while in the Spondoolies DC for testing, as I mentioned previously. It's using 2.2.28, because I found it easier to exceed Zvi's recommendations and controls with this version.

The three from my customer are capped at 1100W, as per Zvi's recommendations. They're each hashing stably at around 3950 GH/s. Currently, I'm seeing 3923.69 GH/s, 3951.675 GH/s, and 3982.555 GH/s. These three are all running 2.2.35.

For reference, I'm also getting 1377 GH/s on my April SP10 operating under similar conditions. It's showing much greater temperature sensitivity than the SP30s. In the SP10s performance graph, I can see a slow downward trend in performance as the room it's in heated up, then an upward trend from last night around sunset as the room cooled off again. I see no similar trends in my SP30 graphs. However, the SP30s are in a different room from the SP10, so this isn't a conclusive test.

Intake temperatures are currently pretty high, as I've got them all crowded in the break room of our office space while we finish up our DC. Temperature does seem to affect the propensity of my miner's PSUs to overheat and go into thermal shutdown, but it doesn't seem to affect hashrate directly, at least on 120V. As long as your PSUs are stable at that power output, your hashrate on 120V should be pretty consistent and independent of temperature. According to my observations so far, I think temperature and airflow might make up to 120W (60W per PSU) of a difference in the amount of power you can let your PSUs supply before going into thermal shutdown.

We energized our transformer for the first time today. It turns out we have the option of supplying about 258V, 249V, or a few other lesser voltages. I'm opting for 258V for now. That voltage will probably drop a little bit once we start to get close to our full load capacity. SP30s and most other machines are rated for a voltage up to 264V. We also got to test one of our evaporative coolers and exhaust fans today. It looks like they'll work very nicely.

I should be able to start migrating our machines over to 258V later today. I'll probably have some statistics on 258V to share tomorrow.

[TKeenan]

The three from my customer are capped at 1100W, as per Zvi's recommendations. They're each hashing stably at around 3950 GH/s. Currently, I'm seeing 3923.69 GH/s, 3951.675 GH/s, and 3982.555 GH/s. These three are all running 2.2.35.

This simply isn't the advertised (after adjustment) 4.5TH.  WTF?  Is spondoolies trying to lower the rate still further?  6.5Th -> 6.0TH -> 5.5TH -> 4.5Th -> 4.0TH.  Where does this stop?

4 is about 60% of 6.5.  That sucks tremendously.  That is not a simple engineering estimation error.  That is a major defect. 

[RoadStress]

The three from my customer are capped at 1100W, as per Zvi's recommendations. They're each hashing stably at around 3950 GH/s. Currently, I'm seeing 3923.69 GH/s, 3951.675 GH/s, and 3982.555 GH/s. These three are all running 2.2.35.

This simply isn't the advertised (after adjustment) 4.5TH.  WTF?  Is spondoolies trying to lower the rate still further?  6.5Th -> 6.0TH -> 5.5TH -> 4.5Th -> 4.0TH.  Where does this stop?

4 is about 60% of 6.5.  That sucks tremendously.  That is not a simple engineering estimation error.  That is a major defect. 

Product was advertised at 5.4Th/s. The 6Th/s was an added bonus and not available for long and for many customers. SP10 had a drop in hashrate on 110V so it was expected to be the same for SP30. The machine can do 4.5Th/s, but you just need to supply it the proper power.

[zvisha]

No, the lab is in Israel. I am actually flying on a 3 day vacation tonight, but we will check this. It's in our interest to provide as much hashing as possible to every voltage.

The equipment hashes 25% less than expected, consumes 30% more than expected, promised units not yet delivered - what to do next?  ...go on a 3 day vacation! 

I wouldn't advertise that.  People waiting are going crazy while the lead guy sits on the beach blowing dope and drinking vodka.