Seems BFL are responding with a Monarch prototype.
it does not take a lot to make a prototype board for demo purposes if you don't show the ASIC chip more smoke and mirrors.
That's right, exactly like the smoke and mirrors boards KNCminer were showing photos of until two days ago when we saw the first chips. If you look at the picture, there is a lot on that BFL board, a Cyclone FPGA, an Altmel 32uc3, it's no toy. |
|
|
|
Hey BarnTech
Is paypal not an option anymore for Batch 2? I have been trying to use the minersource link since lastnight, but it just comes up with a page not found error. The rest of that website seems to be working though. I'm sure it was working ok just 2-3 days ago, but not anymore.
Thanks
Paypal is no longer an option as mentioned already in this thread. Paypal wont take any more money until after hardware is actually shipping. So anymore Paypal pre-orders are out of the question. |
|
|
|
Seems BFL are responding with a Monarch prototype.  |
|
|
|
|
Pretty big chips, not as big as the KNCminer ones but still large, looks like 1,020 pins.
Interesting to see the Cyclone FPGA driving it, and the 32bit Atmel.
|
|
|
|
Any opinions on whether it is OK to place the miner on the side (standing up like a desktop PC) instead of lying down flat?
Would this put the fans under extra stress? Or maybe other bad things could happen?
I wouldn't it may reduce the heat pipe circulation. |
|
|
|
Erk - it depends on how much DYI do you want and are capable of doing. I'm not sure how familiar are you with bitfury's chip and electronics in general so let me know if I'm being too technical - I'll try to put it in a bit more general terms below:
The Bitfury chips (fortunately) are very easy to put in a chain. On the H-card that's exactly what you have - a power converter module (to get from 12V to 0.8-0.9V at high amperage) and a bunch of chained BF chips, and the card provides (through the slot connector) pins for connecting to the previous card and also to the next card (so that you can chain cards as well). So, if you have one working device that can control (at least one) BF chip you can chain another H-card after it (and another after it, etc). All you'll need to do is hook the 3 data signals (SCK/MOSI/MISO) and 1.8V reference voltage from the "control" board to your H-card, and provide 12V to the H-Card.
If you already have means to provide the 12V (e.g. not afraid to solder two thick wires on the H-Card) then I guess the rest of your question is a "control board". The M-board is one such example. It connects to the RasPI and provides the necessary voltage conversion so that it can talk to one or more H-cards. The voltage conversion is really not much of a big deal and if you have some DYI experience you could make your own (that converts from the RasPI 3.3V to the 1.8V needed for BF chips).
If you're looking for a non-RasPI solution then any other self-contained miner (like c-scape's) could possibly be used - that one has a processor, the voltage conversion part and a bunch of BF chips, so you'll have to find the last chip in the chain and if the 3 data signals are accessible wire them to your H-Card.
Another solution would be any of the USB miners. There are several such designs (like the BiFury, RedFury and mine - NanoFury). I don't know for sure the details of the other ones (although I guess it's a similar story across all of them) so I'll speak just about my design - the NanoFury NF1 device has those 3 signals available at 3 small pads (test points). So in order to use a NF1 as the "control board" you would connect 5 wires (GND/1.8V/SCK/MOSI/MISO) from the NF1 to your H-card and then cgminer (or whatever miner you use) would detect the multiple BF chips at that port and drive them all.
Yep that all makes sense, so it basically an SPI bus, what's the function of the 1.8v ref voltage? Good idea about using the NF1 as a controller, saves having to work on firmware which I am novice grade unless I am using something like Arduino dev environment, I am comfortable with C style languages. |
|
|
|
I'm still here and haven't abandoned the project. I've had some disruptions over the last week as the tile guys that I've been waiting for since March finally arrived and started putting in floors. I've been moving stuff around from room to room, and it's a mess, stuff piled up everywhere. Timing couldn't have been worse, and my time diverted. I'm setting up my workbench and tools again and back working again, though the room I'm in still has no floor and I'll have to move back again in 2 days. If it were just me here I could probably have delayed them but my wife (and other family) presses on me as well. Plus I really need a floor.
I appreciate the kind words above and want to finalize this whole thing so I can move on. I feel like I'm stumbling around not knowing what's actually going to happen regarding Avalon chips, much disheartened by everything, but trying to keep a more positive outlook.
Have you considered redoing the Klondike board to use Bitfury chips like Burnin did? I think that would be a viable project. Bitfury chips seems to be all over the place with their pricing, but the main thing is they are out there, unlike Avalon which fell into a blackhole. |
|
|
|
Running at ~20GH/s speed:1257 noncerate[GH/s]:53.386 (2.224/chip) hashrate[GH/s]:55.205 good:3729 errors:205 spi-err:2 miso-err:0 jobs:297 cores:97% good:23 bad:1 off:0 (best[GH/s]:54.360) Sun Sep 29 05:45:42 2013
0: 825 33.086 34.637 2311 138 2 0 15 1 0 (2.068/chip) 95%
4: 432 20.301 20.569 1418 67 0 0 8 0 0 (2.538/chip) 100%
Not bad for my first attempt at (re)drawing a PCB =P A few pages back I remember some folks complaining about hand placing 0402's...have a look at this cheap DIY pick and place: http://vpapanik.blogspot.com/2012/11/low-budget-manual-pick-place.html even though I made one, I still ended up hand placing some of the 0402s because it was actually faster. but the pick and place helped a lot when you're really frustrated and hands are shaking. It's running with R01F @ 4320 ohms (0.8492 V). It's using a TPS53353 (20A) regulator instead of the 30A (just a drop in replacement for the default TPS53355). Is this board designed to go into the M-board like a normal H-board or do you have some other way of driving it? The reason I ask is I am looking for some other way of driving a H-Board, I can't justify the cost of the M-board starter kit, it would kill the ROI on my 1 H-board setup I am currently looking for. I know some people were experimenting with driving a H-board directly and was wondering if you had conquered that? |
|
|
|
Looks like Burnin went to "the dark side". Please do not fund Cryptx hashing farm with this Bitfury burner price scam.
Just ask for a refund. The product is not worth the price, you will not even break even with this price - ROI
- forget about it. So do yourself a favor and let them use their own many to fund their hashing farm.
Why do all these projects seem to go pear shaped? One minute you have something that looks like it might make ROI @470EUR then the next minute a 48% price rise to 699EUR which has no chance. |
|
|
|
Surprised there hasn't been any updates on the progress of getting the initial thumb or board hashing. Issues?
Barntech isn't answering any specific questions regarding development, either through PM or in the thread. He seems more interested in chatting about postage and cases.  He is not the dev, therefore he has to forward questions and answers. |
|
|
|
User review?
Gracias.
None, nobody has received one yet. |
|
|
|
You switched to "selective replies" only? You missed my questions.
I did? Really? Of course, everyone here has free will, and if anything I think people's unwillingness to selectively read and reply contributes a lot to trolling around here. I certainly don't feel the need to respond point by point to every question, and none of you should either. If you feel I've missed something essential, please hit me up and PM and I'll reply more if I agree! In any case, I don't see any evidence on testnet of KNC mining there. Bitcoinorama made a falsifiable claim which appear to me to be false. I'm just asking him to clarify here to avoid any confusion. Cool, but what do you want to see? The miner that they showed yesterday and various iterations that have banked an entire 1BTC earlier. Aside from that there is the FPGA revisions that utilised testnet. Asking Sam he said it's not a problem, when back in Stockholm it means asking one of the software engineers for access. App thousands of coins were found on testnet as the diff is so low there, rewards were easy. Customers demand real world evidence. EDIT: idee2013 that's not KnC. No idea. So what you are saying is the topic of this thread is a lie. I thought as much. |
|
|
|
|
Thanks Kano,
this thread probably should be in the custom hardware forum where the board designers would see it.
|
|
|
|
|
I would have thought that nobody would buy from Yifu ever again after the chip fiasco. I guess people have short memories.
|
|
|
|
...
Thanks for sharing that.
I've said this phrase too often, but "Excuse my ignorance", shouldn't CGMiner display the actual mining rate without having to do calculations?
When I look at my erupters mining, it is displaying 333MHash/s each, plain as day. Why not here?
Thanks Kano,
IAS
It's up to the driver to be able to calculate it. Some devices you can't actually count how many hashes it has done. Others the calculations can be wrong or done badly. Most have to estimate something in there - the better the estimate the more accurate the result. The "nonce" calculation I've used in my post above (like is used in the Avalon and BitBurner drivers) is how you tell if the driver calculation is correct - but you need to do lots of hashes to consider it reliable. Bottom line is that I put a lot of effort into the Icarus driver (ICA, AMU, BLT, LLT, CMR) to make it more reliable reporting the MH/s. However, it itself is doing an estimate on a bit over 1/3 of the work it does also. My 3 AMUs do 335MH/s ... Working on the Klondike project, I recently changes the driver to use the "nonce" calculation and found that the original BKK code + the firmware together were not actually getting it right, they were getting it way UNDER what it was really doing for some configurations. So I switched that driver over to the "nonce" calculation and suddenly everyone was happy to see that indeed it was hashing as it should be  Been meaning to ask for ages, which of the ASIC chips do you find the best/easiest to develop your code for? |
|
|
|
I've said this phrase too often, but "Excuse my ignorance", shouldn't CGMiner display the actual mining rate without having to do calculations?
When I look at my erupters mining, it is displaying 333MHash/s each, plain as day. Why not here?
That's because the implementation of the hashrate meter is driver dependent, and in this case, we didn't write the driver. When we do write the driver, the hashrate displayed is the effective valid hashrate only (i.e. not hardware errors). What does a hardware error actually represent, is it a fault or something else? |
|
|
|
Bitfury is going to milk it until they can't and then r&d a new chip or enjoy the run while it lasts and move on to other things.
Bitfury probably just missed out on close to 2PH/s worth of KNCminer sales over the past few months. If they had their act together they could have had a fair chunk of those. I'm guessing (read hoping!) That Barntech et all are furiously working away assembling boards etc. You are guessing wrong, for starters the chips haven't arrived, and the boards are being outsourced. |
|
|
|
So anywhere up to 300 a day you think? At 469GH/s ( less than a BFL Minirig ) that's 140TH a day Current difficulty is 148,819,199.80509263 = 1065.289327 TH/s So if they ONLY spend a week making them, difficulty will go up ... 92% So the next 2 difficulty changes should together be over 184% ... Lets see who doesn't stop mining in 4 weeks  So lets say they only make them for 2 weeks - 184% - what will the return on them be? ... The 110nm miners and the BFL boys will be the first to switch off, well before KNCminer users. I wouldn't want to be holding ASICminer shares right now. |
|
|
|
the only way to win is to bet on all horses If all the options produce less Bitcoins than they cost to buy and run, then the only way to win is ... not to play. That's only if you follow the crowd. |
|
|
|
KnC Cole on CoinDesk:
"Our engineers have also been working through the night to tune the power consumption and the latest test indicates that we will have a much better result than 1.4 watts at the wall."
this makes me more confident to use a 850W PSU for a Jupiter. Why would you not be confident? 850watt is what they recommended, do you think they don't know what they are talking about? |
|
|
|
|
Show Posts
