ActM is probably going to be only third on the 28nm market Sorry for my lack of research, but who is #2? (assuming KnC is #1) fpga copy 28 nm: KnC ActiveMining fully-custom 28 nm: BFL (not sure if custom, but specs look like it is) HashFast Cointerra If it's not the second, it's going to be bad. Not entirely sure this is accruate Structured Design: ActiveMining Standard-Cell Design KnC Full-Custom Bitfury HashFast Cointerra BFL Falls in either Standard-Cell or Full-Custom |
|
|
|
But in the Bitcoin world eASIC is perfect because we only need a short lifespan of the Chips to pay for the NRE + more even if they are inefficient. Again, I agree and always did say its a sensible approach, although the advantage of fast time to market is basically lost by now, and I worry about the power consumption handicap. Time will tell. Time to Market isn't lost quite yet, I believe the real value comes in when they want more Wafers made after they have a working miner. eASIC has "Fab Time" allocated to them and again because their wafers are prepped somewhat this time should be significantly less than other companies for additional wafers. This is the advantage of eASIC, if this doesn't exist/isn't realistic then I agree that it would be unwise to use eASIC |
|
|
|
The NRE is lowered because it is an FPGA Copy.....
The NRE was ~1M
When I visited California and talked to many individuals that are much more familiar with ASIC costs than I am all stated that 1M NRE for 28nm eASIC (FPGA Copy) is a pretty high NRE.
eASIC Bulk buys these Wafers and has them ready for their customers, they go ahead and charge a couple thousand dollars more per wafer than normal retail and benefit from their bulk purchase and the markup per wafer.
This is my understanding and could be wrong.
First of all, $1M is way lower than you would pay for a full maskset on 28nm. Depending on the number of layers and who you believe, that could easily cost $3-$5M. Im just talking about the physical mask, not the design. Secondly, I dont think you understand how structured asics work. They cant be prefabricated, or at least no entirely. Easic may do wafers with some of the standard layers ready, but they will still need to be processed at the fab to implement the customized routing and extra layers. Its not like easic can take those from the shelve, its not an fpga. So whatever price you heard, is that for fully processed wafers, or preprocessed wafers? Either way, per wafer price is going to be (substantially) higher than for traditional asics, otherwise everyone would do it. I think you are overpricing 28nm maskset, Physical Design + Full MaskSet + First Batch Wafers would cost $3-$5M total on a Full Custom 28nmLike I said, this isn't full custom. This is an FPGA Port and is much easier to do the Physical Design, many less Layers, and the Wafers cost a little more because they are all primed and ready with the basic layers needed to speed up fabrication. (You are effectively paying a premium for speed) Many people don't use a FPGA Copy because it is power inefficient and you could get much more out of a Full-Custom. Read as Wasted Silicon. But in the Bitcoin world eASIC is perfect because we only need a short lifespan of the Chips to pay for the NRE + more even if they are inefficient. Either way even if the Wafers are a tad more expensive than I'm led to believe, wafer costs are the lowest cost of this whole ordeal. $15,000 Wafer (for example purposes) 2,500 Chips 16GH Chip $0.375/GH Edit: Either way none of this even matters because we have zero information and likely will continue to have zero information until there is a physical device ready. I don't understand this way of business but at this point there's not much else we can do besides waiting for that day to come (or not) In the ASIC industry there's also no standard pricing, everyone gets different deals for this and that so it's really hard to come up with what should cost what anyways. |
|
|
|
The wafer pricing isn't much more than normal wafers from what I hear (1K-2K more) The real cost of eASIC is in the NRE.
The whole point of nextreme is lowering the NRE for smaller volumes by eliminating the need for a full mask set. The NRE is lowered because it is an FPGA Copy..... The NRE was ~1M When I visited California and talked to many individuals that are much more familiar with ASIC costs than I am all stated that 1M NRE for 28nm eASIC (FPGA Copy) is a pretty high NRE. eASIC Bulk buys these Wafers and has them ready for their customers, they go ahead and charge a couple thousand dollars more per wafer than normal retail and benefit from their bulk purchase and the markup per wafer. This is my understanding and could be wrong. |
|
|
|
If 30,000 chips is only 6 wafers then not bad but if it ends up being 25 wafers then I believe it will be difficulty for ActM to keep up it's own mining operation while selling units.
For some perspective, going by their specs, hashfast would get ~750 100GH dies (4 dies in one packaged chip) out of a 300mm wafer, so ~75TH per wafer. ActM uses a structured asic, there is a penalty for that, but I dont know how big. Not likely more than a factor 2x, so Im guessing they should get at least ~40TH per wafer. What can you deduce from that? Nothing really, because we dont have a clue of easic nextreme wafer pricing. The wafer pricing isn't much more than normal wafers from what I hear (1K-2K more) The real cost of eASIC is in the NRE. If each wafer is 40TH (2500 Chips/Wafer) Then they only need 10-12 wafers which isn't SO bad I guess but I think your penalty might be a little optimistic. |
|
|
|
the BFL forums used to be the primary place for LRM information... and more often than not info was posted on the BFL forum and not on BTCTalk.... can you please switch to BTCTalk permanently (as you appear to have done) and announce this as the place for all future news. Really frustrating and confusing trying to work out where to look.
+100! I demand my LRM information to be written out in the sky, in big letters, by at least 2-3 planes, while raining down confetti. Anything less is unacceptable! |
|
|
|
Here's a challenge for someone - can you work out how many of these machines ACtM will need to have running in 2months from now to capture 5% of the global market in BTC mining?
First answer gets a gold star.
Pick me Pick me, The answer is 17-20 or 1632 - 1920 256GH Cards or 26,112 - 30,720 16GH Chips YOU GET THE GOLD STAR!! A poxy 20 miners for 5% in 2 months. I like this. I have no opinion on this until we see chips and know the die size so we can see how many chips/wafer. If 30,000 chips is only 6 wafers then not bad but if it ends up being 25 wafers then I believe it will be difficulty for ActM to keep up it's own mining operation while selling units. |
|
|
|
Here's a challenge for someone - can you work out how many of these machines ACtM will need to have running in 2months from now to capture 5% of the global market in BTC mining?
First answer gets a gold star.
Pick me Pick me, The answer is 17-20 or 1632 - 1920 256GH Cards or 26,112 - 30,720 16GH Chips |
|
|
|
It appears to me that it's possible that third unit has a problem. If the problem is what I think it is, I can send you directions on how to fix it.
Please PM me with the serial number of that unit.
I have two units that are not operating correctly. One had the two LEDs closest to the Power Connection blinking back and forth like it is booting up (but after waiting 5 hours they were still blinking) the other one has the four LEDs closest to the Power Connection solid lit up (no blinking) and doesn't change from that state. Any help on these would be appreciated (I'll PM you serial numbers once I get to where they are located) |
|
|
|
I have 3 chilis, all set up the same way and attached to the same machine, and two work great but the third one isn't recognized. When I run Zadig to apply the driver it comes up as "USB Serial Converter" instead of BitFORCE SHA256 SC like the other 2, and subsequently does not appear in cgminer. What am I doing wrong?  Are they plugged into a hub or into three separate USB ports on your computer? Edit: If you don't want to fiddle around with CGMiner just follow these steps, it literally should take you 3-5 minutes to get mining without issue This is a workaround version of BFGMiner until a new firmware update for the chilisHere's a modified version of BFGMiner, I also included the driver inside of it as well that should fix your problem. (I don't know much about CGMiner though) http://www.speedyshare.com/uKncp/BFGMiner-Chili.zipOpen up bfgminer.conf and change the Pool and Username to whatever you want (It currently uses mine so make sure you change it) Run CDM v2.08.30 WHQL Certified.exe to install the correct driver for windows Once you do both of those, just open up bfgminer.exe and it should read from the config automatically and start hashing for you. If it doesn't see the Chili, all you have to do is keep BFGMiner open, unplug the USB, Plug it back in, press M then + then type all or auto and enter. It should find it. Please note if it still isn't working then you need to exit BFGMiner, unplug you chili power, plug it back in and wait for all the lights to stop blinking. If you try to find the miner while it's "Booting up" it will have issues. "Booting Up" is the two most right LEDs (Closer to the Power) blinking back and forth. Again once all the LEDs are not blinking anymore it should be ready to start hashing, start your BFGMiner and follow the above steps. Let me know if this helps  |
|
|
|
I ended up ditching the conf file because I think it was setting some things that interfered with how p2pool worked. I just used the same command line settings (only pool ip and user/pass) that I was using on normal BFGMiner. Seems to read everything fine.
Also, I'm occasionally seeing a Chili go 300GH/s / 0 GH/s (high/low) with 100% hardware errors (immediately obvious on opening BFGMiner) until I power cycle it.
I had a couple Chilis gives 100% HW Errors, it ended up being the USB wasn't registering correctly with BFGMiner (This is the best way I can describe it) You should be able to just unplug/replug in the USB to that miner, let Windows see it and then restart BFGMiner. The other issue with this is that you tried to start BFGMiner while the Chili wasn't fully booted up. It will see the Chili but not communicate properly. Make sure all of the LEDs are not blinking anymore before firing up BFGMiner. Hope this helps |
|
|
|
what is everyone using for thermal ?
pads or paste
what pads are you using if so
what is the the best option for a $50 or under heat sink
|
|
|
|
Bargraphics, is that a 32-bit or 64-bit build?
If it is 32-bit, is the 32-bit version of BFGminer safe to run in 64-bit Windows?
I'm about to build my Chilis and I want to make sure I'm on the right track.
64bit I believe and yes I think even 32bit can run on 64 without issue. |
|
|
|
Dave and Labrat had lunch the other day. What incentive would Dave have to fly to Labrat if Dave didn't have hardware to deliver?
He was in the area anyways, but Labrat is a large customer so it worked out. Dave has always delivered so I'm not worried about any of that, if anything Labrat setting it all up would be the part to worry about since the shear quantity of devices  |
|
|
|
Here I use newest cgminer with zadig.  "cgminer -o stratum+tcp://stratum.mining.eligius.st:3334 -u myaddress -p x -G" Not hashing, doing nothing. Try it on 2 PCs, same thing. My Bitburner Fury works fine. Change usb cable, same result. LED 2 blinking LED 5,6,7 on LED 1,3,4,8 off  Here's a modified version of BFGMiner, I also included the driver inside of it as well that should fix your problem. (I don't know much about CGMiner though) http://www.speedyshare.com/uKncp/BFGMiner-Chili.zipOpen up bfgminer.conf and change the Pool and Username to whatever you want (It currently uses mine so make sure you change it) Run CDM v2.08.30 WHQL Certified.exe to install the correct driver for windows Once you do both of those, just open up bfgminer.exe and it should read from the config automatically and start hashing for you. If it doesn't see the Chili, all you have to do is keep BFGMiner open, unplug the USB, Plug it back in, press M then + then type all or auto and enter. It should find it. Please note if it still isn't working then you need to exit BFGMiner, unplug you chili power, plug it back in and wait for all the lights to stop blinking. If you try to find the miner while it's "Booting up" it will have issues. "Booting Up" is the two most right LEDs (Closer to the Power) blinking back and forth. Again once all the LEDs are not blinking anymore it should be ready to start hashing, start your BFGMiner and follow the above steps. Let me know if this helps |
|
|
|
And yes lol, 50% is within the rules. 400% is the limit I believe, and has been reached before.
Wow. We're fucked. This is the most accurate description of our situation I've seen so far. Cypherdoc would disagree why would i necessarily disagree? do you disagree? back when i was being paid, the #'s looked a lot more compelling than they do now. let's not forget you were being bullish on KNC/Terrahash ordering back then too. the relentless, exponential growth has surprised all of us. HR is looking a little toppy right now; we'll have to see how much, if any, current lesser efficient asic mining gets turned off. I have no comment until we see how the MPP plays out. Hashfast's chips may overperform, they might offer something additional to not make it so bad. But if they don't offer anything additional and the MPP does exactly what it says it is going to do then yes I guess I would only agree that it's not looking that great. The outlook is much worse if you were not first in line and paid retail prices. Who knows what HF has up their sleeves. |
|
|
|
And yes lol, 50% is within the rules. 400% is the limit I believe, and has been reached before.
Wow. We're fucked. This is the most accurate description of our situation I've seen so far. Cypherdoc would disagree |
|
|
|
eligius under DDOS
I can't seem to pull up the website with stats but my miner shows it's getting shares accepted. New to Eligius, assuming this is par for the course? same here! Web server is down, pool is up. Pool is what matters however I'm sure they are aware of it and are working to remedy the web server. |
|
|
|
Butterfly Chili Rack   So after trying to fire up the ~60 chilis and feeling the heat coming off them I have got to ask you how your cabinet is handling those hah. We had to add a 25,000 BTU on top of the 22,000 BTU A/C unit we already had for the room. Definitely underestimated how much heat these chilis make, the name fits! |
|
|
|
As warm as it gets in there I'm pretty sure that is exactly what it is saying! |
|
|
|
|
Show Posts
