daserpent1
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 05:19:44 AM |
|
Cointerra, can't you just upload a video showing the miner hashing away at 1Ths or 2THs ? That would be enough to prove every hater, wrong.
|
|
|
|
aerobatic
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 08:21:54 AM |
|
Cointerra, can't you just upload a video showing the miner hashing away at 1Ths or 2THs ? That would be enough to prove every hater, wrong.
Daser, no one that has designed any silicon that is waiting for their silicon to come back from the fab can show it work til it arrives back, which is 2-3 months after tape-out. Thats the way it works, and so knc, hashfast and cointerra are in the same boat. Knc's silicon is due back end of this month, hshfasts is due end of october/ early nov, and cointerras is due in december. Until the time that they can show it working they can show you the design specs, or pictures what it might look like, their bios and proof that theyre experts... or some other evidence that they know what theyre doing, but this is the way silicon works. You design it first, then you send it off for physical design to a specialist company, then you tape out, and wait 60-100 days while the fab makes it, mask by mask - almost a day for each mask - then you get your chips back and if you did everything correctly, they should work! All cointerra can do til they get their chips back is try and convince us that they know what theyre doing. If they could show working silicon, today, they would be selling it. Its the same with everyone who has taken a pre-order... They just dont have their silicon back yet.
|
|
|
|
daserpent1
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 04:59:51 PM |
|
Cointerra, can't you just upload a video showing the miner hashing away at 1Ths or 2THs ? That would be enough to prove every hater, wrong.
Daser, no one that has designed any silicon that is waiting for their silicon to come back from the fab can show it work til it arrives back, which is 2-3 months after tape-out. Thats the way it works, and so knc, hashfast and cointerra are in the same boat. Knc's silicon is due back end of this month, hshfasts is due end of october/ early nov, and cointerras is due in december. Until the time that they can show it working they can show you the design specs, or pictures what it might look like, their bios and proof that theyre experts... or some other evidence that they know what theyre doing, but this is the way silicon works. You design it first, then you send it off for physical design to a specialist company, then you tape out, and wait 60-100 days while the fab makes it, mask by mask - almost a day for each mask - then you get your chips back and if you did everything correctly, they should work! All cointerra can do til they get their chips back is try and convince us that they know what theyre doing. If they could show working silicon, today, they would be selling it. Its the same with everyone who has taken a pre-order... They just dont have their silicon back yet. Thanks for clarifying it. Well i guess, we just have to wait it. And people are too suspicious about cointerra because the chip specs they released (500Ghs per chip at $3/Ghs) is too good to be true.
|
|
|
|
creativex
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 07:01:12 PM |
|
Thanks for clarifying it. Well i guess, we just have to wait it. And people are too suspicious about cointerra because the chip specs they released (500Ghs per chip at $3/Ghs) is too good to be true. People have been pointing out that mining equipment vendors' margins are enormous since forever. What looks too good to be true to some now, may well be perceived as overpriced within just a few months. Pre-order vaporware vendors have been riding this gravy train for months. Hashfast sierra and coincraft are now under $8/Gh allegedly for November delivery, making knc look tardy now. VMC...-60% recently. Frankly, I'm glad this price war is moving so rapidly. I was concerned it would take years to get to reasonable prices, but all the new vendors are pushing pricing down with the quickness.
|
|
|
|
Puppet
Legendary
Offline
Activity: 980
Merit: 1040
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 07:04:40 PM |
|
. I was concerned it would take years to get to reasonable prices,
Define reasonable price. No matter what price point per TH you pick, if miners overbuy it wont be profitable.
|
|
|
|
creativex
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 07:33:58 PM |
|
. I was concerned it would take years to get to reasonable prices, Define reasonable price. Reasonable price = potential for profit for more than just vendors. Reasonable expectation for a positive ROI. When people stop placing orders months in advance for products that are losing value at a rate of 2.5%/day(or go broke) this glut of orders will dry up. Of course many are still echoing the sentiment that only those early in company xyz's pre-order queue will make the big bux so it may take a while. Perhaps watching CoinTerra drop their prices by 62% since late August will help speed this process along. No matter what price point per TH you pick, if miners overbuy it wont be profitable. Agreed, but I won't be adding to the pile of stupid money being poured into vendors' wallets. There are too many entrants into the hardware peddling market now for vendors to maintain their ridiculous margins. Sure some misguided new to mining youngsters with more bux than sense will have to get burned, but the excesses will eventually be washed out. Much of this BTW should have been clear with the original context you excluded in your quote.
|
|
|
|
Puppet
Legendary
Offline
Activity: 980
Merit: 1040
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 07:41:03 PM |
|
Point is that its not vendors pricing thats to blame for the lack of profitability. LIke I said, they can put any price on it per TH they want, 5x lower than today, and it would still not be profitable because miners would just buy 5x more and still over invest. Once they stop doing that, asic prices will drop further (further increasing difficulty). There is no solution, its inherent to the nature of the mining market and asics.
|
|
|
|
creativex
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 07:48:09 PM |
|
Point is that its not vendors pricing thats to blame for the lack of profitability. LIke I said, they can put any price on it per TH they want, 5x lower than today, and it would still not be profitable because miners would just buy 5x more and still over invest. Once they stop doing that, asic prices will drop further (further increasing difficulty). There is no solution, its inherent to the nature of the mining market and asics. Great point! Wish I'd thought of it... ![Huh](https://bitcointalk.org/Smileys/default/huh.gif)
|
|
|
|
xyzzy099
Legendary
Online
Activity: 1065
Merit: 1077
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 07:56:22 PM |
|
Point is that its not vendors pricing thats to blame for the lack of profitability. LIke I said, they can put any price on it per TH they want, 5x lower than today, and it would still not be profitable because miners would just buy 5x more and still over invest. Once they stop doing that, asic prices will drop further (further increasing difficulty). There is no solution, its inherent to the nature of the mining market and asics. Great point! Wish I'd thought of it... ![Huh](https://bitcointalk.org/Smileys/default/huh.gif) I think the market has been severely distorted by pre-sales. Miners have been willing to pay more for gear because they have been paying for it in a time frame when it looked like a good investment - but by the time they actually get the gear, the ROI is nada. Miners will soon (out of necessity, if not common sense) come to realize that this preorder thing ain't working - and vendors will have to have gear in-hand to sell it. THEN, miners will be able to make realistic decisions about what prices make sense.
|
Libertarians: Diligently plotting to take over the world and leave you alone.
|
|
|
creativex
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 08:01:50 PM |
|
AWESOME SAUCE! NOW I GET IT! ![Grin](https://bitcointalk.org/Smileys/default/grin.gif) ...it's almost like I didn't say all of this half an hour ago in this very thread...
|
|
|
|
Puppet
Legendary
Offline
Activity: 980
Merit: 1040
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 08:07:24 PM |
|
I think the market has been severely distorted by pre-sales. Miners have been willing to pay more for gear because they have been paying for it in a time frame when it looked like a good investment - but by the time they actually get the gear, the ROI is nada. Miners will soon (out of necessity, if not common sense) come to realize that this preorder thing ain't working - and vendors will have to have gear in-hand to sell it. THEN, miners will be able to make realistic decisions about what prices make sense.
Not really. It may seem like presales is the problem, but its not. Let me explain: The problem with presales is that you dont know how many TH have been sold and will be sold by the time you receive your unit, let alone by the time your expect to ROI. Those TH will influence future difficulty and thus your revenue, so you cant accurately predict your ROI. However, even even without presales, if you buy available hardware you still cant predict how many TH those companies are selling and will sell in the near future before you reach ROI, so you cant really predict your ROI either. Well, in fact, you can predict your ROI in both cases. ITs quite simple; on average it will be negative. Because of the dynamics of the market. Whenever there is an opportunity for mining profit, it will be taken by usually over optimistic miners. If there doesnt appear to be one, asic vendors will drop their prices to make sure there appears one again. Which ensures the previous miners will see their profit evaporate. Because if they had predicted these price drops correctly, they wouldnt have bought. This will happen over and over again until we reach the bottom of asic pricing.
|
|
|
|
xyzzy099
Legendary
Online
Activity: 1065
Merit: 1077
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 08:10:56 PM |
|
AWESOME SAUCE! NOW I GET IT! ![Grin](https://bitcointalk.org/Smileys/default/grin.gif) ...it's almost like I didn't say all of this half an hour ago in this very thread... Sorry, I actually meant to reply to Puppet's post. Clicked the wrong one. I also had not read the whole, thread. My bad. We're right though ![Wink](https://bitcointalk.org/Smileys/default/wink.gif)
|
Libertarians: Diligently plotting to take over the world and leave you alone.
|
|
|
creativex
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 20, 2013, 08:21:52 PM |
|
We're right though ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) Naturally. ![Grin](https://bitcointalk.org/Smileys/default/grin.gif)
|
|
|
|
User705
Legendary
Offline
Activity: 896
Merit: 1006
First 100% Liquid Stablecoin Backed by Gold
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 21, 2013, 12:42:54 AM |
|
. I was concerned it would take years to get to reasonable prices,
Define reasonable price. No matter what price point per TH you pick, if miners overbuy it wont be profitable. Reasonable prices are prices very close to cost of production which means there is no more room to discount and therefore supply slows past a certain difficulty point. The faster we get close to production cost in 28nm format the better.
|
|
|
|
|
geofflosophy
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 24, 2013, 06:51:25 PM |
|
Can someone in the know explain the level of complexity of going from 2 FPGA to 10 FPGA for example? When KnC did their Mars FPGA demo, they used so many more chips; on it's face the fact that Cointerra only used two seems like a questionable move, but if this type of thing scales easily then I suppose it really doesn't matter.
|
|
|
|
Puppet
Legendary
Offline
Activity: 980
Merit: 1040
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 24, 2013, 07:02:19 PM |
|
They used the fpga to simulate 2 hashing cores. The actual asic will have 360 of those (im guessing three dies on a chip, each die with 120 cores), but obviously you cant cram as many cores on an FPGA as on an asic, otherwise we wouldnt need asics. Its just to demonstrate the RTL and software works.
|
|
|
|
geofflosophy
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 24, 2013, 08:20:26 PM |
|
They used the fpga to simulate 2 hashing cores. The actual asic will have 360 of those (im guessing three dies on a chip, each die with 120 cores), but obviously you cant cram as many cores on an FPGA as on an asic, otherwise we wouldnt need asics. Its just to demonstrate the RTL and software works.
That part I get. What I'm asking about is whether the number of hashing cores adds any level of complexity to the design process or if it scales easily or something in between.
|
|
|
|
creativex
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 24, 2013, 08:25:04 PM |
|
Heat density will likely be their largest obstacle going forward. We'll have to wait and see what they come up with. CoinTerra® also released that the system would be cooled by custom equipment created by CoolIT Systems of Canada that can handle up to 400 watts per chip. The cooling system is built for chips hashing as high as 700 GH/s. The system also uses multiple high speed 12 cm fans in a 4U case. http://cointerra.com/cointerra-demonstrates-working-fpga-releases-additional-chip-details/
|
|
|
|
Puppet
Legendary
Offline
Activity: 980
Merit: 1040
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
September 24, 2013, 08:38:01 PM |
|
That part I get. What I'm asking about is whether the number of hashing cores adds any level of complexity to the design process or if it scales easily or something in between.
It scales perfectly. Pretty much a copy/paste job. I dont think anyone seriously doubt cointerra can design a working hashing core RTL with that all star team. The only potential pitfalls are timing, manufacturing, system engineering (including cooling) and supply chain.
|
|
|
|
|