I am thinking these Prisma miners will over clock well. Glad that FC found a fix so we do not need those crap controllers now. I think AM will sell quite a few of these once they are in hand and not a pre order. I know that I am planning on getting a few!
You cant overclock much, because one column as feeding with TPS53355 sbc chip. So this chip handle 30A , each column has a 3 be200 chip. 1 be200 chip can't feed bigger than 10A. for 11.52gh be200 current is 8.8A . 11.52x192be200=2.21Th, finally you dont have much 1.2A to overclock 2.21Th is enough That may have been at the voltage of the tube, but i beleive the prismatic is undervolt ed a bit, which means much higher amperage for the same wattage.
|
|
|
Nah you're forgetting the capital to build a proper 100kw warehouse. You cant just put AC in a room and call it a day.
Also you assumed avg mining hardware is 0.7w/GH. I would say its 1w/GH due to all the rush happened when 1TH/s miner came out.
Home mining has one advantage, once you setup a mining space ( dedicated room...etc), the mining expense is only the electricity fee. At 15cent/KwH, it would be on par with 100TH mining farm's expense per TH/s
It only make sense for MW mining farms from here. All the small mining farms will close shop b4 avg home miners. Take note noobs, dont jump into any group buy hash share deals.
Capital is capital, its not a running cost. But yeah, setting up a space for 100kW means about $5,000-15,000 in electrical work and cable runs, plus $2,000-20,000 for a cooling soluton (air cooling is the cheapest) assuming 0.7W is fair i think, since the 1w/GH gear wil very soon be balanced with 0.5w/GH, and most manufacturers are now at 0.7w/gh. I would say that at this point in the game a 100TH farm in china (lower costs and wages) could be started up and over the salaries of two people while being profitable. In North America, you might need closer to 200TH due to the extra $15,000 year of a base salary and typically higher $/sqft for industrial space. some basic math i did looks like 2 people with access to a suitable location costing $1200/month and needing $20,000 for setup could distribute the cost across 4 years and pay themselves a small (~30k each) salary for a 100kW/100TH farm at a cost around $30/month (a 2yr period is a cost around $50/month) Thats roughly equal to 0.05-0.08/kwh
|
|
|
Hello Everyone,
S4 B1 has been already shipped out on Sept. 29th, and it is no shipping in the following days from Sept. 30 to Oct. 6th due to the Chinese Golden Week. B2 will go on sale at 22:00pm Sept. 29th, shipping starts on Oct. 10th Beijing (UTC +8 Time Zone).
BITMAIN
price? Is S2 part of the larger assembly process and thus more reasonably priced? I can't honestly pay anything over 3BTC shipped if it uses 1400W
|
|
|
While I respect AM and your guides Dogie, I have to respectfully disagree -
The math: 800(speed of the tube) X 240/270(ratio of clocks) X 2(ratio of chips per miner) = 1422 and because you have underclocked you can now lower the voltage resulting in some power savings. Come on man we've all done this with our GPUs; this is nothing different or special and nothing they couldn't have done right off the hop to give AM an edge.
parallell powering layout my ASS. Who came up with this BS?
The guy with inside information maybe? But your math checks out so it's k, lets go with that. I'm sure friedcat purposely made the tubes consume an additional 50% power for no reason. Not commenting on the series string vs VRM debate since I don't see any reason why they couldn't have done so, but if you're burning an extra 50% in your VRM you've done it extremely poorly. maybe not 50%, but I do know they get very hot, must be losing up to 10% power over the regulator and another 5% on the inductor. It also looks like theres a few extra chips. Couple all these things together and theres definitely potential to trim some power draw in comparison to the previously high-volted, overclocked chips
|
|
|
Crazy can you please verify that you are shipping from the USA...not France..in the main Prisma thread. Somehow it got confused where you ship from...even after saying you prefer USPS domestically...
seems obvious its from the USA. that guy posting about it being a zip code for paris was just an idiot no need to be disrespectful. https://www.google.fr/maps/place/13e+Arrondissement,+Paris/@48.830759,2.359204,7z/data=!4m2!3m1!1s0x47e672274c4e0583:0x50b82c368941b10?sa=X&ei=kGsoVImTM9GM7Ab2o4GgAw&ved=0CJABEPIBMA8 I mean, cmon. read the thread. USPS, zipcode, mentions that shipping to EU means overseas, gave prices for US shipping.
|
|
|
Hi, i wonder, what is the average maintenance cost of a mid or large scale of mining farm in the units of Wh/GHS? Can it be calculated?
maintenance cost = electricity usage of mining hardware + power supplies + electricity usage of cooling solutions(air conditioner etc) + rent of land + hosting internet + payments of staff + maintenance of hardware + etc..
could it be lower than 1W/GHS?
cost isnt represented in w/GH... 1000sqft facility with access to 100kW = $800-2000/month internet = $100/month 2 guys who can spend average of 100hrs/month installs and maintenance = $2000-5000/month using 0.7w/GH equipment, add 10% for PSU losses and 10% for air-cooling (no large farm would use closed AC) = <1w/GH = ~100TH capacity so 100TH will cost 100kW + $2900-6000 per month. ($3-$6/kw/month when distributed) power in some locations suited for this is about $0.05-0.10/kwh or, or $35-$70/kw/month. thus, a large location with well-priced power could have operation costs of $40-$75/TH/kW/month. In comparison mining at home with $0.15/kwh will cost about $105/month not including any rent/salary/internet/maintenance EDIT: same math but for a 25kW capacity only would be about $50-$95/month
|
|
|
Crazy can you please verify that you are shipping from the USA...not France..in the main Prisma thread. Somehow it got confused where you ship from...even after saying you prefer USPS domestically...
seems obvious its from the USA. that guy posting about it being a zip code for paris was just an idiot
|
|
|
i chose the third option because this sort of thread is dumb and not belonging in hardware.
tl;dr - this thread belongs in the subforum dedicated to bitching about other users (there is actually one)
|
|
|
I recently put 4 undervolted S1's on a single CX750 using decent looking 18ga pcie splitters. Each one running 225mhz, 115gh, ~137w for a total of 460gh/550w. The wires don't even get warm. The PSU is a lil warm. Thinking about bumping them up to 140gh/~166w when I get time. Just ordered some more splitters so I can do 4 more on another CX750. Here are the splitters I'm using: http://www.ebay.com/itm/131197541426I've noticed that the S1's that performed well at 200gh with low hwe are the ones hardest to undervolt. Some of them eventually show up X's,etc... And the S1's that didn't perform well overclocked are the ones easiest to undervolt. I have one S1 that would never go above 195gh without many hwe. Now its been running for 78hrs at 115gh and only 3 hwe. Yes just 3. Those splitters you ordered from ebay, it is not going to work. You needs to buy splitter with 8 PCI-E that split into 2x6 pins. I tried that in the past, it didn't powered the s1 he just said his splitters work fine... and they should. 6pin->6pin splitters are totally fine to use. please note that these 18awg ebay splitters are often mediocre quality - 100W per split shouldnt be an issue, but if you tried it on a un-modded S1 or S3 you would likely melt the wires and start a fire at >150w per side of the splitter, or the combined >250W on the primary PCIe cable
|
|
|
Antminer S3+ Batch 9, 8 chips showing as non-operational. Tried removing the thermal compound, cleaning and re-apply non-conductive thermal compound without any luck. Tried switching the cables around, etc.. Anyone else having similar issues? Should I request a RMA (Anyone have any experience with the RMA processes? Do i need to send the entire unit back?) looks like an entire chain is down. try checking the voltage if you have access to a multimeter - its likely the DC/DC regulator for the first chip in the series isnt fuctioning (should see 0.7V across the large grey cube-like inductor) and as a result the entire chain does not function. check if there are any damaged resistors or missing bits
|
|
|
Also, I'll be placing the order for Batch 2 tonight, regardless of if we meet the 10 unit limit. Remaining units will be sold on ASICPuppy.net at a non pre-sale price. Batch 3 will commence shortly after Batch 2.
looks like shipping to Canada is about 1/$105 or 2/$160, does that sound correct? any idea when this would actually ship? Im really interested in getting 1 or two units. would definitely need the BE controller but would also like to have the USB dongle since i have an RPi already (kept as backup for existing gear)
|
|
|
No point, the dogie guy will dodge this scenario as he insulted me while being delusional and pretending that I was incompetent for pointing out his failure. Don't forget those are manufactured specs. An example is gigabytes windforce cooler which they claim can dissipate up to 600watts. https://bitcointalk.org/index.php?topic=706302.msg8868719#msg8868719He mentioned high end air cooler or basic 120mm watercooler. I pointed out a dual 120mm closed loop cooler could do up to 300w but would love to see his unicorn air cooler.. Somehow he thinks quoting me as If I can't read and chose to not post this unicorn cooler and insult me instead, of course because we all know typing 3 words is easier than 30 I don't think dogie is totally wrong, you *can* find air cooling that handles 350W but as I showed above its often large, top-heavy monstrosities that cost as much as a liquid-cooling solution. That windforce one you mentioned is for GPU so doesnt make much sense in this regard. IMO, dogie is 90% right - cluster the chips and place mounting holes for common CPU heatsinks. Let the buyer go and get a decent liquid-cooling system from thier local market and most of those can easily take 300-400W TDP for a fair price and small footprint. Additionally, they can be salvaged and used on other equipment in 12 months from now when the HEX4RB is no longer profitable. (or if the HEX4RB underclocks to ~200W TDP in 6mos from now it can be swapped for a more tame air cooler)
|
|
|
Hello, needs the S4 a normal C13 cable (like a PC power supply) to power? Like this: I ask because Bitmain write 16A and C13 is normaly not rated for 16A. Yes, judging by the picture it's C13. You'll need a 14GA C13 if running on 120V, or standard 18GA if 240v. I'm in the UK and I can't seem to find a 16a c13. Could someone provide a couple of links to the correct cable needed for the S4? if you are in UK you should have 220V power source, so 18awg is just fine
|
|
|
Looks like a longer tube. How did they get the efficiency up?
Downclock and downvolt each chip. Nope. Its a new parallel powering layout which is simply more efficient. If I get a Prisma you'll be able to see it in the pictures. That may be true but it's also underclocked. The tubes were 8.3gh/s per chip and Prisma is 7.3gh/s. Where did you hear about this "parallel powering layout"? Downclocking the chips by 10% doesn't result in 25% power savings. The design was first done by XBTec on their Pacifics ( https://farm4.staticflickr.com/3874/14825302026_f8e55aca1f_o.jpg) which results in significantly lower power requirements by stringing the chips together. [Note their V1 didn't actually realise the power savings in the design due to other bottlenecks, but the V2 did]. This is the same 'technology' that friedcat has put on the Prisma to generate the power savings compared to the Tube. is that the stringing to run them off 12V without massive losses on the DC/DC converters?
|
|
|
Your post regarding an air cooler dissipating 350watt is nonsense. Dual 120mm radiators with decent fans of 1500-1800rpm would dissipate up to 300w. I am not sure where you're pulling your magical unicorn numbers from. Feel free to post a air cooler that can dissipate 350watts, make sure it doesn't have 5000rpm fans attached to it.
I just need to quote this to show you don't know what you're talking about and you're not worth arguing with. Get off your high horse. Hard to argue something you can't prove. You took the time to write horse manure instead of list a name of a product, so the only waste of time here is you. Dodge more when someone questions your opinion because you can't provide facts to prove it. You are claiming a "high end" air cooler can dissipate 350watts of heat. Do you have a name for one? Sorry you don't. The high end air coolers you mention cost the same as closed loop coolers. The closed loop coolers dissipate more watts of heat than the air coolers. When it comes to surface area, air and water, ambient temperature water flowing through a radiator with fans blowing through the radiator fins will dissipate more than the same surface area of heatsink with just air blowing on the fins. Again, feel free to prove yourself, because it would of took less words to type the model of this "high end air cooler". You could have saved everyone the time and just listed the name of this unicorn air cooler. 250W TDP at 26dB (huge, using 2 120mm fans) http://www.bequiet.com/en/cpucooler/482350W TDP http://www.techpowerup.com/141811/zalman-announces-cnps11x-extreme-cpu-cooler.htmlIMO the only fan/heatsink coolers that can handle such wattage are massive monstrosities that almost certainly cannot be shipped mounted or they will snap the PCB/mounts, and would require a big open case to offer the airflow. Closed-loop liquid cooling seems like the obvious winner
|
|
|
Too true ... Rule 1: NEVER buy the first incarnation of anything related to BTC mining with Bitmain. They will learn from any problems and refine.
Then the price may ... (should) ... drop, but so will the rewards with ongoing difficulty.
I got B1 (4.75) and B2 (4.25) S1 units and they were perfect. Mined about 6.5 and 6btc respectively, and resold for almost the electricity costs of the entire time running them.
|
|
|
THIS. 1) the thing has a cd-drive. its clearly not meant for bitcoins 2) that custom stand/case is probably worth $2000 and probably almost as much to ship. Its clearly a one-off and not viable in mass production Noone is (or should be) taking this seriously.
|
|
|
Really no need to fix those controllers if we can use a Pi!! So it's either buggy ethernet controller with uncertain fix or Raspberry PI with uncertain capability (possibly unreliable and/or underpowered) in an uncertain setup. If there's a problem with the Pi it's going to be the driver. There's no trouble running cgminer on a Pi and the BE200 ethernet controller is much weaker than a Pi. As in, 1/4000th of the RAM space. -- novak I believe the issue with the RPIs are USB and ethernet bus related. The BTCGarden units ran great on an RPI but they were using GPIO. I'm thinking the Prisma limit on RPI will probably be lower than the BE controller but we'll know soon enough. The good thing about the USB adapter is that it should be trivial to get these running on any controller you choose. I've been told by phasebird that the adapter should be compatible with tubes as well, waiting on confirmation. would this same USB dongle work for running S1 blades? (theres a thread dedicated to running them on an RPi)
|
|
|
Latest price update: $350 for 1 miner (4 boards) with 1 Rpi. 620GH/s. $0.56/GHs. 648W. $640 for 2 miners (8 boards) with 1 Rpi. 1.2TH/s+. $0.53/GHs. 1296W.
Any plans for a more efficient design, similar to the improvements implemented in the asicminer prisma? Doubling the chip density could get ~600gh/450w per miner, and improvements like this are necessary with the rising difficulty. >1w/gh is no longer a good buy for most people paying >$0.10/kwh
|
|
|
only one coupon can be used per S4. This is scam-ish at best. Not really - thats how coupons have worked ever since the start. $200 is a fair coupon, but imo the price is too high for the power draw. If this was drawing 1000-1100w I would buy it at the price, but 1400w is too much. With the s2 they basically doubled the number of chips to drive each at a more efficient voltage. But with the s4 it looks like 5xS3 worth of chips with only a marginal savings to power consumption. Hopefully they will bring out a revised design that uses 7xS3 worth of chips to achieve 0.5w/gh. I might buy 1 with a coupon just because I have been selling more hardware then I've been buying, but it's going to take a serious slowdown in the difficulty (7-10% jumps) for the next several months if this is going to bring a positive return.
|
|
|
|