Author

Topic: The Profitability of Scrypt Mining on Amazon EC2 (Dec 2013) (Read 1255 times)

newbie
Activity: 28
Merit: 0
Just finished to try this and it's really a pleasure, we all want to thank you for this interesting explanation.
full member
Activity: 122
Merit: 100
Nice experiment, but CPU mining Scrypt coins cannot be profitable if you pay for electricity, so why should it be profitable at Amazon EC2 ?
newbie
Activity: 2
Merit: 0
Yeah, a CPU only coin makes sense in this case since you cant access a GPU on there.

Agreed.
newbie
Activity: 6
Merit: 0
I've tried on Azure. Not worth it
full member
Activity: 168
Merit: 100
Yeah, a CPU only coin makes sense in this case since you cant access a GPU on there.
full member
Activity: 238
Merit: 100
Quote
The moral of the story is that, at the current difficulty and bitcoin price, rented server instances are not profitable to mine any coin. But if the price were to continue to skyrocket, or there were to be a drop in difficulty, it is an easy thing to set up and try.

Maybe you should try to mine a CPU only coin rather than scrypt coins if you're going to try an amazon ec2 instance.  There are plenty of CPU minable coins to choose from quark, protoshares, xpm etc.  In my experience cloud hashing on these coins (though not at amazon) has been slightly profitable, and a small profit per instance could be scaled up to a large profit with many instances.  Saying that it is not profitable to mine any coin this way is not correct, it is only not profitable to mine scrypt coins this way.

legendary
Activity: 924
Merit: 1000
very interesting experiement. Thank you for the information was debating to give this a try but will wait now.
newbie
Activity: 5
Merit: 0
So I ran a little experiment yesterday to see if mining scrypt-type coins could in any way be profitable using Amazon EC2.

I started by firing up a m1.small instance (Generic setup), which could be reserved using their spot instance system for $0.01 per hour. This is about as cheap as you're going to be able to get an Amazon server. m1.small is a 1 CPU instance.

I set up pooler's cpuminer on that m1.small, and let it run. I averaged 2.42 kH/s during the 3 hours that I was mining it. I mined it against a scrypt pool that picks the most profitable coin to mine (usually WDC), and at 2.42 kH/s, the profits were minuscule. Less than the 3 cents that I paid for renting the server.

So I switched to a m1.xlarge instance, which has 4 CPUS. Set up the same pooler cpuminer, this time with 4 threads. I averaged 4.85 kH/s per thread (or around 19.4 kH/s total) on that xlarge server. I was able to rent this machine for $0.07 per hour, or $0.21 for the 3 hours. I did not manage to extract 21 cents of value from the altcoins either.

I did not pick a CPU intensive instance, nor try the GPU miner. These are a lot more expensive, and there is no indication they are profitable.

The moral of the story is that, at the current difficulty and bitcoin price, rented server instances are not profitable to mine any coin. But if the price were to continue to skyrocket, or there were to be a drop in difficulty, it is an easy thing to set up and try.

I was able to get this experiment going, from idea to installation and running cpu miner, in about 30 minutes.
Jump to: