As far as I know is Nvidia better in floating point calculations then AMD, so the coin should be based on that.
FLOPS often involve rounding (result truncation based on the mantissa of the data going into the operation), which makes it lossy and non-invertible. If you reduce the operations to reversible ones I think you essentially just get integer operations.
|
|
|
Cryptography is dependent on integer op throughtput
AMD pulls way more IOPS
I doubt nVidia can compete in anything crypto-related
|
|
|
My guess is that someone has written a nice new virus for GPU mining and wrapped it into several games or other programs available as torrents. You can imagine if you seed a copy of a new game that's popular and 20,000 gamers download it, achieving such a hash rate is quite easy.
|
|
|
It looks like Dave got his hardware and is shipping out the BitFury devices, maybe his 200 TH/s mine is going up simultaneously too?
|
|
|
I agree it will begin leveling off at some point. Thinking that this point is the 80-90M difficulty is where I thought you were bonkers.
There are a LOT of ASICs preordered that more or less will show up and begin hashing. The 200TH (formerly 100TH) project is coming online fully, a number of other mines are coming on, and I fully expect the difficulty will shoot past 1B, which is where the early Avalon miners stop being profitable. People will continue ordering hardware well past the point it's worth running, IMO.
The increase in efficiency with ASICs is somewhere between 100-1000 fold. I'd expect the difficulty to end up plateauing between 100 M and 5 B, since we were plateauing at 1 M - 5 M before with just GPUs. At the current price and 5 B as the difficulty, my block erupters are only slightly unprofitable, while my BitFury device is still profitable.
|
|
|
TL/DR Whatever the relative performance of this FPGA is to a CPU miner it would be WORSE if the p value was higher. LTC decision to use a low p value makes what otherwise would be a nearly impossible task into one which is merely challenging.
I doubt it. You should have a look at Solar Designer's TMTO data with N=2^14, r=2^3, p=1. As you can see, as memory exponentially decreases integer ops exponentially increase. He was easily able to get the memory usage into the kilobytes and still crank out hashes. I'd guess that exactly the same is true with N=2^10, r=1, p=1 too. It's the same balancing act you run into no matter what value you use for N or r; at higher N you may increase the difficulty by a smaller constant factor, but overall I doubt increasing N or r will make scrypt much more FPGA/ASIC unfriendly when they finally iron out the FPGA implementation.
|
|
|
I always like to develop ideas openly and encourage their theft -- probably have spent too much time in academia. But whoever I pass this to can develop as they see fit.
|
|
|
Sure, I'll take one. Your website says they aren't available at this time, though. Let me know how much shipping is to Canada.
|
|
|
Probably because the 55 nM chips from BitFury are circuitry works of art while the KnC chips are thought to be direct ports of the logic circuits derived from their own less efficient FPGAs. KnC also chose to work on a more difficult process, so they may experience piss poor initial yields that require huge amounts of voltage to hash. I would be wary of "predicted" efficiency numbers, as BFL's predictions were off by almost an order of magnitude. Even BitFury was unable to meet their target, but got pretty close.
|
|
|
@_@ almost to group buy 7
|
|
|
Considering the good news, I'm rather surprised to see 160 hashing boards still available for October. I thought there would be a run on the shop as soon as shipping day was confirmed. Maybe I should go snap up a few more before it's too late. Calculating it out and assuming that Birfury chips get out in a reasonable timeframe, between the European sales, the 200 TH/s mine with Dave, the private Bitfury pool (already >100 TH/s), and all the North American sales, unless another company brings something dramatic to the table (probably KnC) it's likely that by the beginning of November half the blocks on the network will be hashed with a Bitfury device
|
|
|
By October, we'll have worked out this issue. We are playing with different R01F (input) resistors now to see how well that works.
Cool, sounds good
|
|
|
So here's the status guys:
Niko's new guy in China is very fast. Niko's already got boards and has shipped a few. They have also got the v2 M-board working, but I haven't produced any for August shipment.
I've got tracking numbers in hand for the first shipments of M's and H's - about 15 full rigs worth. More will ship each day as they are made. Probably much larger quantities on the next shipments. We are also balancing production with Niko's guy, since we are very happy with the turnaround time. According to FedEx, I should be receiving first boards on 8/29 - we shall see about that.
Niko is also seeing some under-performing cards compared to what Tytus was seeing with his test rig. This could be due to the chips (different wafer, different packager, etc). It could be due to the v2 M-board, different PCB manufacturer, components, software changes, etc. The testing process is time consuming because we must run the boards long enough to get good data - they have to warm up and run for a few hours, then get tuned.
We are committed to shipping the promised hashrate, so you will receive a tested rig that can produce what we claim. If that means we add in extra cards then that's what we will do. For full rig owners, if your rig is coming up short we will send an extra M-board + H-boards until we reach the advertised 400GH/s rate.
If shipments arrive on time, the testing & shipping process will run straight through the holiday weekend until everything is out the door. We may lose a shipping day due to Labor Day (Monday, Sept. 2nd), but the team will be working.
Cheers, Dave
I ordered 250 GH/s worth of H-boards with my starter kit, if I'm lacking > 25 GH/s from that will I more H boards to compensate?
|
|
|
where's the important links omg 100TH
See this thread: https://bitcointalk.org/index.php?topic=251966.0Bitfury and crew have been very good at cranking out product without generating a lot of noise, unlike a lot of the competition. Apparently the assembly is operating on the volume of hundreds of H-boards (25 GH/s each) per day, so that 100 TH/s came online so quickly isn't a big surprise. This is the East Euro mine, the NA 200 TH/s mine that dave is operating should be up in a week or so too. You can buy stock here: https://picostocks.com/stocks/view/19
|
|
|
100 TH/s now.
As stated in the US Bitfury thread, dave was bringing all of his 100 TH/s online at the end of this month.
|
|
|
Okay, I've fixed this. It's extremely weird, though.
I had two cards, MSI 1 GHz and Sapphire 1 GHz. MSI card was getting 450 KH/s, saph 700 KH/s. Mucked around for 3 hours until I finally hit the solution below.
Catalyst 13.3, MSI AB 3.0.0 beta, APP SDK 2.8, cgminer 3.3.1
To get them both running at 700 KH/s: 1.) Create two instances of cgminer miner with -d 0 and the other with -d 1, on both: -g 2 --thread-concurrency 8192 -I 13 DO NOT RUN BOTH CARDS IN THE SAME CGMINER INSTANCE 2.) Go to MSI AB and hit reset for both cards to reset them to stock clocks. 3.) Start both cgminer instances. 4.) Apply your OC.
|
|
|
I'll take a preassembled case shipping to Canada, I'm not sure how to enter the queue but PM me and I can send you the quantity in BTC for it.
|
|
|
|