Spondoolies-Tech vs Bitmain for a larger operation?

[Korbman]

EDIT: I noticed the equipment totals changed to 5 AntMiners and 3 SP20s. Though this changes the overall numbers, the concept is still the same.

This is actually kind of an interesting thread.

@OP - If you are legit, don't purchase any mining gear....yet.
With the 10 miners you've purchased for BTC19 (or the fiat equivalent), I imagine you're running at about 14.3 TH/s, generating BTC0.17394234 per day at the current difficulty (or $39.14 @ $225 per coin). Using roughly 9 kWh for the gear alone, you're daily electricity costs should be around $9.67 (since you hinted toward $0.045 per kWh), and that's fantastic! It means you're difficulty ceiling (the point where revenue is equal to expenses) is a bit over 167,042,141,322, or 4 times where we're at today.

Obviously you're in a good position given your datacenter setup (racks, cooling, power capabilities) and cheap electricity. The problem is the low price per Bitcoin. Recouping the initial test expense of BTC19 would take ~110 days...but only if the difficulty remains constant. There will be ~8-9 difficulty changes between now and your hypothetical breakeven point. So the fun, and challenging, part of researching a feasible setup is determining what the difficulty will look like after 8, 9, 10 adjustments and how that ultimately changes your breakeven analysis.

So! I noted "yet" above because it may be better to wait for the price per coin to increase (not guaranteed of course) or to wait for the next generation of more efficient hardware to come to market (which is, unfortunately, also an unknown). Waiting on the latter is usually my preference because most massive mining operations will have already implemented the gear by the time smaller operations can even begin purchasing it. That usually gives me an opportunity to see how the network reacts and how the difficulty rises, allowing me to readjust my predictions as necessary before making the purchase (if it's still feasible).

[CptTripps]

Korbman: What an incredibly informative and thoughtful post. I appreciate it! I'd already started factoring quite a bit of that, so another "set of eyes" on this is appreciated.

I'd come up with similar numbers built using an excel sheet that I've been hobbling together over the last week. I'm used to dealing with headcounts, salaries, project scopes and timelines....getting down and dirty with the most basic of economics is a refreshing change.

That's why I wanted to start with a small operation...made up of two different kinds of gear. See which is easier to setup and maintain and take actual temperature readings to see what we need to do about cooling.

My friend (referenced earlier, and in a video link that someone else posted) just finished a 500TH deployment with the S5, and the older 150TH they decommissioned was only a year old. So I need to factor in a yearly hardware swap. That being said, I'm likely to be in a better position with a particular manufacturer if I am already buying/using their gear.

Just checked the order status, and it looks like I'll have my S5s on Monday...no shipping info for the SP20s yet.

[kendog77]

If history is any guide, I would expect to see an announcement from Bitmain soon on the S6. The S6 should be more efficient than the S5, so I would not be surprised to see specs at around 3.5 TH / 1400 W (.4 W/GH).

Also, Spondoolies is a wild card since they have given no information about their next generation Pick-Axe chip. It could be available next month or months from now.

But, since your power costs and overhead are so low, I think a huge competitive advantage over 95% of the other miners out there, so loading up on current generation hardware would probably be very profitable for you unless the entire BTC mining market implodes due to low BTC value.

Mining hardware one generation old also has decent resale value, so reselling on Ebay is not too difficult when you want to upgrade.

[CptTripps]

Here's a breakdown of what I've spent so far. This will become useful in a week or two after I've let everything run for a bit and then I can compare apples to apples.

=============================================
Antminer S5 build:  

5 Antminer S5 @ $449.99ea = $2,249.95     (I bought these on Amazon. I know I can get it cheaper, but this was state-side and easy.)
5 Corsair CX750M power supplies @ $79.99ea = $399.95
Total: $2,649.90

Expected Specs:

6,500 GH/s @ 590W (Edited)
=============================================

=============================================
SDTech SP20 build:

3 Spondoolies Tech SP20 = $1,195.00
3 EVGA SuperNOVA 1300G2 @ $169ea = $507.00
Total: $1,702

Expected Specs:

4,800 GH/s @ 1,200W (Edited)
=============================================

For arguments sake, let's assume power costs $0.05 kwh.

I'm going to factor all this together and see where it lands. (I'm still trying to get my excel sheet to work correctly.) Right now it looks kinda like a toss-up.

[Syke]

I'm going to factor all this together and see where it lands. (I'm still trying to get my excel sheet to work correctly.) Right now it looks kinda like a toss-up.

Put the S5 at 1.3 TH/s, and the SP20 at 1.6 TH/s. Those are more realistic numbers.

[Syke]

Looking at the Hashcoins Uranus is interesting...gotta run the numbers on that one too.

Ack, no!!! Don't waste your time. There are no options outside of Bitmain and Spondoolies. None.

[tntdgcr]

Looking at the Hashcoins Uranus is interesting...gotta run the numbers on that one too.

Ack, no!!! Don't waste your time. There are no options outside of Bitmain and Spondoolies. None.

That's 100% the truth ... for the OP , I have 120+ hosted SP20s that I've been mining if you want to try them out. Nothing like skipping DHL Customs

[padrino]

If you are serious about building an operation at the scale you mentioned earlier you could certainly improve your power choices, you are taking a capital and 10-15% efficiency hit with what you choose.. If you are going to scale it up there are much better options...

For example I pulled together a build in a US data center for a customer last fall, about 150Th/s.. Coincidentally I was deploying within a week of one of Bitmain's US deployments (they have a few) and deployed the power for about 10K less in capital, and used 12% less power.. So much of a difference the data center execs came to me to help invalidate Bitmain's claims power wasn't being measured correctly.. Sure enough the PSUs Bitmain choose were poor..

[CptTripps]

If you are serious about building an operation at the scale you mentioned earlier you could certainly improve your power choices, you are taking a capital and 10-15% efficiency hit with what you choose.. If you are going to scale it up there are much better options...

That's an excellent point.

I'm going to look at the PSUs that I already have, but more efficient choices could make a big difference at this scale. A 100w difference per PSU would add up to 12,000W diff at the end.

[sidehack]

If you're looking at 120 PSUs outputting S5-range power (say, 600W) 100W per PSU is the difference between a 93% efficient quality unit and an 80% piece of junk. Hopefully the range of options you look at don't have that much range.
The difference between 93% and 90% at a 2KW output from a big fat PSU is about 72W per PSU.

[Korbman]

Korbman: What an incredibly informative and thoughtful post. I appreciate it! I'd already started factoring quite a bit of that, so another "set of eyes" on this is appreciated.

I'm always happy to help where I can, and I'm glad you found my post informative. Mining on a larger scale tends to be vastly more complex than most people realize, and while I'm sure you know that, I still try to point out some "things to think about" for anyone else reading the forums.

I'd come up with similar numbers built using an excel sheet that I've been hobbling together over the last week. I'm used to dealing with headcounts, salaries, project scopes and timelines....getting down and dirty with the most basic of economics is a refreshing change.

Not sure if you've started doing this yet, but what I've always tried to do is to simplify *everything* down to what it costs per kWh (employee hourly rates, space rental, cooling...really any reoccurring expense) and add it to your electricity costs. The idea being to create an all encompassing expense, an example being $0.08 per kWh ($0.03 to cover monthly rent, monitoring, and maintenance + $0.05 for the raw electrical rate). I've found this usually helps when making difficulty and revenue predictions (among other things).

Other than that, my biggest question comes down to depreciation...and this is something I'm also asking the other experienced miners / veterans (both to help with my models and with CptTripps' setup). Is there a standard depreciation method you use for your hardware, or do you typically just run the equipment until it's no longer valued at anything (ending its useful life)?

[Biodom]

Korbman: What an incredibly informative and thoughtful post. I appreciate it! I'd already started factoring quite a bit of that, so another "set of eyes" on this is appreciated.

I'm always happy to help where I can, and I'm glad you found my post informative. Mining on a larger scale tends to be vastly more complex than most people realize, and while I'm sure you know that, I still try to point out some "things to think about" for anyone else reading the forums.

I'd come up with similar numbers built using an excel sheet that I've been hobbling together over the last week. I'm used to dealing with headcounts, salaries, project scopes and timelines....getting down and dirty with the most basic of economics is a refreshing change.

Not sure if you've started doing this yet, but what I've always tried to do is to simplify *everything* down to what it costs per kWh (employee hourly rates, space rental, cooling...really any reoccurring expense) and add it to your electricity costs. The idea being to create an all encompassing expense, an example being $0.08 per kWh ($0.03 to cover monthly rent, monitoring, and maintenance + $0.05 for the raw electrical rate). I've found this usually helps when making difficulty and revenue predictions (among other things).

Other than that, my biggest question comes down to depreciation...and this is something I'm also asking the other experienced miners / veterans (both to help with my models and with CptTripps' setup). Is there a standard depreciation method you use for your hardware, or do you typically just run the equipment until it's no longer valued at anything (ending its useful life)?

it's very difficult to figure out depreciation because it depends on the BTC price in $$, euro, etc..
Example: even the smallest piece of gear increased in price 3-10 fold in December of 2013 because BTC priced moved from $130 to $1200 in a space of 2-3 months.

[TheRealSteve]

I'd simplify it even further and take the gamble and headache of difficulty/revenue out of the equation altogether by planning for a total Thash/s output - a lot of the other numbers will fall into place, and the toss-up depends more on how long you plan to be running the operation (operational costs will make one overtake the other) and what you think you can get for it on the market once you upgrade or shut down.

[sidehack]

Make sure to consider underclockability into total machine life. If the S5 and SP20 are about neck-and-neck when the S5 is at stock and SP20 is already underclocked, consider where the practical volt/clock floor is for each machine. When something reaches breakeven returns, it might be made to run at a more efficient setpoint and hold on longer at reduced returns (but still above operating costs) so the machine with more room to go down in W/GH would be viable longer. The S1s built 14 months ago can be made to operate as efficiently as a stock Prisma.

[Medow]

I am in the exact situation here.

We have a cheap power here with a  rate of 2 cent per kw, and we are start to build our own farm. When me and my team decide on which gear to go with then at the last step we make a new  decision to wait for more 2 months to see what will go on with the new gears that we expect to come out next 2 our 3 months from now.

I advice you to wait a little bit, because it better to have a better deficiency than to be sorry.

[CptTripps]

That's 100% the truth ... for the OP , I have 120+ hosted SP20s that I've been mining if you want to try them out. Nothing like skipping DHL Customs

Already have my waybills, but I appreciate the offer.

I'd simplify it even further and take the gamble and headache of difficulty/revenue out of the equation altogether by planning for a total Thash/s output

I do like simplifying things. But I want a good handle on all the unit economics before I start making things easy on myself. That's a great point though.

With my company, we used to break everything down to the cost per transaction. After a few years we realized that logic was flawed as our volume went up exponentially but other costs went up as well. Once we started concentrating on the unit economics of particular business lines, that's when we started making more money.

This is MUCH simpler math. There are only a few factors involved, which is wonderful. There also appears to be much more volatility. That part is also very different.

I've been researching the hell out of this. There seem to be a lot of opinions, based on the costs. Speculation on the market is an interesting factor to add in. This means I'm going to have to form my own opinion from experience ultimately. That's why we're starting with a 11 TH/s setup.

[sidehack]

Another good way to save money - don't pay employees.

[alh]

Another good way to save money - don't pay employees.

Just to be flippant, does the above mean:

1) Don't hire employees in the first place?

2) Hire them and just don't actually pay them?

[sidehack]

I tend to lean more on option number 1, but a lot of people prefer option number 2. We call those people "bastards".

[CptTripps]

Ha. We already have about 30 people in the office, so that's already covered.

Another observation: Most people seem to just be using wire racks. I'm used to rack-mounted gear, so this is a bit of a change.