Indeed too blunt .
I do believe they have the resources to upgrade SW
Topic: [ANN] Bitfury ASIC sales in EU and Europe - page 3. (Read 250506 times)
Quote
Chip delivers a minimum of 100 gigahash per second of computing power. On average, the BitFury 16nm ASIC can compute in the range of as high as 140 gigahash per second using air cooling, and up to 184 gigahash per second using immersion cooling
so one h-card will be 10/14/16 x 100 ghs = 1/1.4/1.6THs and 16 of these on a m-board will be 16/22.4/25,6THs?
@ 0.06 J/GH 25.6TH on an MBoard would be 1.5kW. Although possible, I think it's unlikely.
m-boards originally had issues when pushed past 600W, but that was largely due to there only being 2xPCIe connections (and 300W/each is too much for most wirings). There is/was an extra pair of screw terminals though, which could likely handle 600W+ themselves.
If you used all the connections available, you could likely achieve up to 900-1200W, which require active cooling and likely heatsinks.
Look at the number of orphans they pump out at the moment - and how on occasion they put out blocks that are well over 10s late on the network.
I'd not be surprised if they put them out much later and no one sees them.
Had one on my pool the other week (older bitfury with their software) that sent a stale block to the pool 69 seconds after the block change ...
So using their setup could be anywhere as high as 10% wasted ... i.e. could be as low as 90% of any specs quoted.
If I found any new miners on my pool with their hardware, I'd close the account - it's pretty much the equivalent of block withholding using a miner that is known to do this.
It is not about racism .
I do not like the way Chinese make business. They just eat everything.
I did not say I do not like Chinese people .
I do not like the way Chinese make business. They just eat everything.
I did not say I do not like Chinese people .
Actualy the problem for centralization are large mines in China
They do not look longtime . Mining for them is just USD income , same would be with manufacturing condoms
They do not look longtime . Mining for them is just USD income , same would be with manufacturing condoms
More unfounded racism. The American companies all abandoned us, the European companies all abandoned us and who is left? China, 2x. We could have been in the situation where there was no access to ASICs for the average Joe what so ever but instead they continued to sell to us.
Actualy the problem for centralization are large mines in China
They do not look longtime . Mining for them is just USD income , same would be with manufacturing condoms
They do not look longtime . Mining for them is just USD income , same would be with manufacturing condoms
Impressive work! Good job Bitfury!
*image removed*
*image removed*
good job for who ? for sure not for us small miners ...
Good job for network security!
Yep,the network is secure!!! By 3 or so corporate entities
Bout time to find another ASIC resistant coin to use
Well I never said that they are decentralized. The other coins combined don't have this kind of security.
Impressive work! Good job Bitfury!
*image removed*
*image removed*
good job for who ? for sure not for us small miners ...
Good job for network security!
Yep,the network is secure!!! By 3 or so corporate entities
Bout time to find another ASIC resistant coin to use
Quote
Chip delivers a minimum of 100 gigahash per second of computing power. On average, the BitFury 16nm ASIC can compute in the range of as high as 140 gigahash per second using air cooling, and up to 184 gigahash per second using immersion cooling
so one h-card will be 10/14/16 x 100 ghs = 1/1.4/1.6THs and 16 of these on a m-board will be 16/22.4/25,6THs?
@ 0.06 J/GH 25.6TH on an MBoard would be 1.5kW. Although possible, I think it's unlikely.
m-boards originally had issues when pushed past 600W, but that was largely due to there only being 2xPCIe connections (and 300W/each is too much for most wirings). There is/was an extra pair of screw terminals though, which could likely handle 600W+ themselves.
If you used all the connections available, you could likely achieve up to 900-1200W, which require active cooling and likely heatsinks.
so a Bitfury 2015 16nm full kit could be around 20TH at a similar price as the 2013 October 400GHs kit: €7,500
And that's assuming it'll still be at 0.06J/GH at 100GH/s, which is unlikely.
Quote
Chip delivers a minimum of 100 gigahash per second of computing power. On average, the BitFury 16nm ASIC can compute in the range of as high as 140 gigahash per second using air cooling, and up to 184 gigahash per second using immersion cooling
so one h-card will be 10/14/16 x 100 ghs = 1/1.4/1.6THs and 16 of these on a m-board will be 16/22.4/25,6THs?
@ 0.06 J/GH 25.6TH on an MBoard would be 1.5kW. Although possible, I think it's unlikely.
and easy to implement
as far I remember A1 chip from inno had this option too
as far I remember A1 chip from inno had this option too
I think what he means is, since the chips are chained each is at a different local ground so chip-to-chip signalling might need to be shifted; is that handled internally?
The beauty of this design is that the coms are chained as well, so chip to chip is already "auto" shifted. You can see all they have is just an external resistor divider network on the coms line. Of course the end nodes still need to be shifted, but in terms of serial chain design this is as simple as it gets.
Quote
Chip delivers a minimum of 100 gigahash per second of computing power. On average, the BitFury 16nm ASIC can compute in the range of as high as 140 gigahash per second using air cooling, and up to 184 gigahash per second using immersion cooling
so one h-card will be 10/14/16 x 100 ghs = 1/1.4/1.6THs and 16 of these on a m-board will be 16/22.4/25,6THs?
Would be very nice to see bitfury gear and or chips for regular folks.
I hope so too, last time we bought Rev2 chips price was way too high sure hope they ask normal prices. 
Would be very nice to see bitfury gear and or chips for regular folks.
Well ... Yes and No
The board design itself takes care for the shifted grounds, not the chip, but the chip is helping thanks to the internal structure of the IO and IOREF
The board design itself takes care for the shifted grounds, not the chip, but the chip is helping thanks to the internal structure of the IO and IOREF
I think what he means is, since the chips are chained each is at a different local ground so chip-to-chip signalling might need to be shifted; is that handled internally?
Yes that was my question, just that the board looked to be so free of other components I wondered if they had integrated the level shifters for the chained signals into the chips as I think Bitmain has now done with the BM1385?
Rich
I think what he means is, since the chips are chained each is at a different local ground so chip-to-chip signalling might need to be shifted; is that handled internally?
I don't know why everyone keeps quoting "for air cooling", they'll design their systems to run with their ridiculous immersion cooling first and foremost. What is possible with air cooling is an afterthought and doesn't need to hit the same specs.
In the press release they have quoted numbers for both air-cooling and immersion cooling:Was level shifting built into the ASIC's or are there components on the other side of the board?
No, those are just hashing boards, while the level shifters are on the masterboard.I thing the BFSB miner was designed to chain few hashing boards using the same level-shifter. This is not possible with multiple string boards, but if the string board is the last one, you may still use non-string boards before it.
For strings should use a dedicated level-shifter for each string and RPi has enough pins for chip enable for up to 16 slots.
EDIT: 16 slots without using multiplexers as it was done in the Metabank's miners
Jump to: