Bitcoin Forum

I am looking to expand my farm....

From a daily expense factor and/or mining result which is better?

Two s3's
Or
One c1

And why did you choose that answer!

If you have very cheap power the s3's are pretty good.

2 of them can do 1000gh and 800 watts

quiet? c1
better wattage? s3
more hash? c1
overclock-ablility? c1
faster webguy? c1 also less beeps than s3
....

also if u are handy you can use c1 to heat-up water for different usage!

It depends on the C1's personality.  If it does not need rebooted much and runs good I would go C1.  If they still have the syscooling pump I would replace it.

If it has problems then two S3's.   S3's are built like a tank.

Adding to others longevity, S3s seem to last longer than C1s.

I attribute a lot of this to original pump.  I have had no coolant issues after switching from syscooling pump.

Syscooling was a horrible company.  I'm glad I don't see a relationship continue with them and Bitmain on other models.  I still have not heard of any other water cooling company that dry tested pumps.  Batch one pumps were on last leg when we got them.

As an aside from the pumps even, the high clock speeds of the C1s leads to dropped chips / dying blades quicker. Not necessarily problematically quick, but quicker than the S3.

All nice advice so far.

Which solution gives off less heat to the surrounding environment?

It all just comes down to watts. If their power usage is the same, then their heat production is the same. Liquid cooling or air, it all just ends up as warmer air (after having passed through the radiator on the C1).

2 * s3+ should give less heat! also 900gh/s vs 1000gh/s
....

I would say it depends on what freq you S3+ is on.  I know I had a few that hit 500 with freq up.   I overclocked them during winter.   

The C1's I don't notice heat as much just because of a big fan pushing the heat outside currently.  But both get hot if you are running 1T with 2 S3 + or 1 C1.

stock vs stock!

don't want to think about ovk because it gets more complicated because each miner is different, even with the same model miners....

a proportion between the two tells to me that

1008gh/s : 806,4w = 880gh/s : 732w

so we are looking at 1.25 versus 1.20

c1 is 1/24 better, based on this i'll go with c1, i like also to have 1 powerful miner than many less powerful one, just because of density...

Aren't the C1 and the S3 based on the same BM1382 chips?  The power consumption is going to be the same from a watts per GH perspective.  As far as how much space things take, with each C1 you've got a PSU, miner and water cooling equipment.  With the S3, you can drive 3 of them from a single EVGA 1300 G2, compared to the single C1 the same PSU could drive.

Honestly, it's a tough choice.  My opinion is to go with the S3s just because you don't have the added headache of the water cooling.

Yes exact same chip's.  They take up a decent amount of space by time you set up them up.  They run at a much higher freq then the S3's. Below is a old picture of my 2 C1's.


This is within the first 2 day's of running as they still had syscooling pumps.

S3's are pretty much bulletproof same as older S1's.

C1's seem to be problematic.

I'd go for S3's all day long ;).

I have to agree with the recommendation regarding the S3's. They have a reputation for serious reliability, way better than the C1's cooling system. Also with two miners, you have the option of splitting your hash as you might see fit. Say across two pools, or maybe renting one on Nicehash/Westhash. If you really want to fiddle with a liquid cooling setup, then C1 is your choice. Otherwise I think it's S3 all the way.

C1's are fine after some tuing.  But I would agree S3's are tanks.  They take a lot less to setup and mine.

C1's are fun for a bit to play with watercooling.  But it's something I did for fun more then anything.  The only thing I can say for C1 that is absolutely better is the sd card being used on controller.  Makes it much harder to brick.

If you could find some way to run them in series... I don't know too much about water cooling and I never really got into the thermodynamics and fluid mechanics stuff... but if there was some way to hook up like 10 boards to a single pump/radiator then I could see the density argument.  Has anyone tried doing something like that?  Throwing 10 boards, a pump/radiator/fans/etc into a 4U enclosure?  I'm not sure if something like that is even possible...

They made a kit for 4 C1's to run on.   I don't think I ever saw a sell of it though.

I know I have seen two C1's in series, the nice thing was if one pump died it the other pump kept going.  So it's been done but not to a huge extent that i've seen.

I have had the original pump which broke down and let coolant out.  I would not put these into a nice enclosure, if something goes bad you will have a very mad data center.

I would think a series arrangement would be terrible for ten blades. As the water flows through the loop, it will increase in temperature, until the 10th blade where it's pretty warm and doesn't have as much capacity to absorb more. Wouldn't it make more sense to have the cool water exiting the radiator go into a manifold which essentially distributes the water down multiple paths to each blade?

You would have to pretty much have a radiator between each unit.   Or some way of keeping the coolant really really cold.   You will not be able to just hook up pump and run through 10 without some serious planning.

So you could run in series but would be kinda a cluster with all of it hooked up together.

Remember the fun days of extreme overclocking?  People would drop their Pentiums into liquid nitrogen to try and get the most possible speed out of the chips.

I agree the cooling performance of the liquid would certainly degrade the more boards it had to deal with, and perhaps the optimal limit is indeed what the C1 offers.  I know there were some folks doing total immersion tests with the Spondoolies boards to get some crazy density.  It would be interesting to see how far these things can be pushed.

During GPU day's people pushed cards hard.  But you had a very long warranty and it was not hard to unplug one card and send a RMA in.

With today's asis if you push to hard chances are you will end up with a dead chip.  Some gear does OC better then other Bitmain has always done good in OC area.

Yeah, the GPU/CPU were pretty easy to RMA.  It's considerably tougher to do the same with the ASIC mining gear.  I have had mixed luck with the Bitmain gear.  My S1s all overclocked and ran solidly at 200GH/s.  My S3s... well, I've got a few that stay steady at 478GH/s.  A few that won't go above 440GH/s.  One that just will not run consistently above 400GH/s and one that has been running at 504GH/s pretty much non-stop since I got it last year.  Oh yeah, and 1 that just flat out died on me at the end of January.  Oh well... it had earned its keep by then, so I just wrote it off :)

The GPU's RMA on most companies was truly amazing.  I think most of my cards were like 3 and 5 year warranty.  Which is insane for a video card.  It's out of date by that time.

I didnt have to do that many RMA's but I did a few.   It was very easy unplug the broken GPU and within 2 weeks normally you would get a working card like it.  Only thing was a lot of RMA cards were beaten up a little and did not look near as nice.

On asics manufactures they obviously can't offer this for to long.  In most cases 90 days.