if they used actual thermal epoxy, it might indeed be a daunting task to remove the heatsink. we'd need to know if it's naked silicon under the heatsink, or if there is a metal cap that the heatsink attaches to. if its capped, often a razor blade and some patience go a long way, and you can scrape and sand any residue that can't be removed with a solvent. can't do that with naked silicon.
It has a heat spreader (no naked silicon). Still rather daunting. I guess BFL might be willing to sell boards with no heatsink or thermal epoxy but mounting a waterblock w/ thermal epoxy isn't something I would be comfortable with. Not the kind of wattage (thermal load) we are talking about here. Obviously the current cooling system is sub-optimal and they might be looking to change. If/when they do I hope they put some push pin holes in the board even if they aren't used. That would allow alternative cooling. Not just watercooling but miner's experimenting w/ larger cpu style heatpipe heatsink fan assemblies. the big question then is exactly how tenacious the adhesive they used is. if its just an adhesive TIM, no big job at all, if its thermal epoxy that cures strong like bull, well that's a much different situation, but most likely not a show stopper. as for mounting the water block, i'm pretty sure a couple of zip ties and an adhesive thermal pad would do the job well. i agree fully though, if they included mounting holes spaced like standard CPU holes, that would be the best option. |
|
|
|
No that will be when ASICs come out. (soon)
do you have information on ASICs currently being developed? on a side note, total cost of ownership as you scale needs to be looked at when comparing FPGAs to GPUs. with GPUs, your going to have to invest far more in support hardware (motherboards, cpu's, ram, etc) than you will with FPGAs. as an example, apparently you can run 100 of these singles off of one computer (1 CPU, 1 MOBO, 1 HDD, etc) where with GPU's, every few that you want to expand to, you have to buy another MOBO/CPU, etc. so comparing them 1 to 1 is only valid up to say 3 units (or whatever number of pci slots modern mobo's have generally), after which the cost of FPGAs seems to become much cheaper than expanding with GPUs. That is a good point but if you are smart you can get 6 GPU on a decent board like MSI 890FXA-GD70 (8 w/ dual GPU cards). Unless you intend of buying 8 of these you aren't getting any benefit from that theoretical 100 per CPU. Now someone like giga running a massive farm if he does purchase 100 boards will be able to benefit from lower host overhead. Still most people aren't going to drop $10,000+ on FPGA to get the benefit you are talking about. yes, the less invested, the more apt the 1 to 1 comparison is. the further along the MH/s scale you go, the more the cost advantage swings towards FPGAs. i'm kind of focusing on the Rig Boxes and that level. working out some details mentally on how to proceed with them, assuming the stats they offer are relevant, which even now seems to be a huge assumption. but even with another FPGA solution, at a 20k investment, FPGA's would seem to be the hands down more cost effective option. |
|
|
|
No that will be when ASICs come out. (soon)
do you have information on ASICs currently being developed? on a side note, total cost of ownership as you scale needs to be looked at when comparing FPGAs to GPUs. with GPUs, your going to have to invest far more in support hardware (motherboards, cpu's, ram, etc) than you will with FPGAs. as an example, apparently you can run 100 of these singles off of one computer (1 CPU, 1 MOBO, 1 HDD, etc) where with GPU's, every few that you want to expand to, you have to buy another MOBO/CPU, etc. so comparing them 1 to 1 is only valid up to say 3 units (or whatever number of pci slots modern mobo's have generally), after which the cost of FPGAs seems to become much cheaper than expanding with GPUs. |
|
|
|
Forgive me if someone asked already but I did not see it.
1. What platform is the included software for. Must I run Windows or can I plug these into my Linux systems and have them play happily?
2. Why is BTC not an option for payment?
weird, pretty sure it was. still is there on the rig box, but not the singles. |
|
|
|
it seems like water cooling might be a good option for these units, both from a performance and noise perspective.
That was my first thought. Since they appear to be the thermally limited I wonder if you could get 900 MH/s with watercooling. Two problems a) no mounting holes. If it had any kind of pushthrough mounting holes you could use a universal northbridge water waterblock. that would be ideal, but with such a small board size, i'm sure we could come up with some kind system, when coupled with an adhesive TIM, it would only be partially responsible for anchoring the water block. b) it looks like heatsink is epoxied on ? If so removing it is going to be tough and depending on the thermal epoxy used you might not be able to get enough of it off to get a smooth surface for mounting. if they used actual thermal epoxy, it might indeed be a daunting task to remove the heatsink. we'd need to know if it's naked silicon under the heatsink, or if there is a metal cap that the heatsink attaches to. if its capped, often a razor blade and some patience go a long way, and you can scrape and sand any residue that can't be removed with a solvent. can't do that with naked silicon. |
|
|
|
Consider this: Even if 100% of the network suddenly replaced their GPUs with FPGA, come December the block reward gets cut in half and the Single will be generating $1.50/day (assuming today's difficulty and BTC/USD exchange). That gives it a 400-day payback period. Is that good? Not particularly. yeah, but there are still ~300 days until the approximate time when the reward will be halved to 25btc. If you are planning to mine with a Single, there are only 300 days if you ordered it back in November and it arrives now. People ordering today will not see it for 2 months. Then there will be 250 days left.  yeah, i thought about the production delays as i was posting. hopefully many of the production kinks have been worked out, and they will be in a position to start delivering in a quicker more efficient manner. regardless though, as long as you can get a unit in the next few months, you should be able to take a huge chunk off the time to payback prior to the adjustment. |
|
|
|
|
it seems like water cooling might be a good option for these units, both from a performance and noise perspective.
|
|
|
|
Consider this: Even if 100% of the network suddenly replaced their GPUs with FPGA, come December the block reward gets cut in half and the Single will be generating $1.50/day (assuming today's difficulty and BTC/USD exchange). That gives it a 400-day payback period. Is that good? Not particularly. yeah, but there are still ~300 days until the approximate time when the reward will be halved to 25btc. |
|
|
|
hello,
everyone knows, it's Crisis Time everywhere in europa
so i decided to put all my savings into bitcoin
I bought them all at 6.8$ and now i pray for the charts to reach 6.81 again
a candle a day, hailelujah
-_-'
if you truely intend to have bitcoin be your main store of currency, i strongly recommend you look into dollar cost averaging. https://www.google.com/search?q=dollar+cost+averaging |
|
|
|
|
so nanominer will work on an FPGA hardware? lets say i stumble on a dev kit at a great price, but its not one used before for bitcoin, would there be any required optimization, or special steps needed?
|
|
|
|
Hi all newbies  ^ Doing that is easier. I hope you all don't mind! as the duly elected noob spokesman, we extend our thanks. |
|
|
|
|
i notice a lot of sigs with addresses in them. do people tip often?
|
|
|
|
|
definitely would have registered earlier if i knew i was on the clock!
|
|
|
|
Warning! Bitcoins are addictive.
unbelievably so. i have no idea why either, but damn i'm hooked. |
|
|
|
|
recently caught the bitcoin bug. went to reply on a thread and DOH!
|
|
|
|
|
Show Posts
