Topic: rumors of 3.5Th antminer S7 "lite" - page 7. (Read 8641 times)
I would prefer a smaller perhaps 1/3 version of the S7 see Should Bitmain make a Q7? . But a 2/3rds version S7 is progress.
still too big for my taste, it's only 1/4 less bigger, they should aim for 1/2 at least
for those that can't run a big miner in their home, and want to do only occasional mining activity
Doubt there's much profit available for them in a single hash board miner. Still gotta have a frame, controller, and fan - and that would make it very difficult to impossible for them to keep the price/TH competative.
It would actually be 1/3'd less presuming the "2 hash board of current design instead of 3" concept is what they pick.
"integrate the PSU" is not a good idea, way too many variations on size for PS used to power miners.
I suspect folks wanting a "small" miner are going to have to wait for whatever comes out of the Gekkoscience projects - and hopefully SOMEONE will start being willing to sell them chips on a reasonable basis soon in current-gen chips.
Depends on price of it I think. Some hated it but I always had good luck with S4's. I loved that it was all in one. Just the ease of use to be sit it up with no plugging in or doing really anything is nice.
But I agree like on my S5+ where it was a heck of a lot of PCIe cords it's nice that I could pick out the PSU's. Makes it pretty nice on that miner.
still too big for my taste, it's only 1/4 less bigger, they should aim for 1/2 at least
for those that can't run a big miner in their home, and want to do only occasional mining activity
Doubt there's much profit available for them in a single hash board miner. Still gotta have a frame, controller, and fan - and that would make it very difficult to impossible for them to keep the price/TH competative.
It would actually be 1/3'd less presuming the "2 hash board of current design instead of 3" concept is what they pick.
"integrate the PSU" is not a good idea, way too many variations on size for PS used to power miners.
I suspect folks wanting a "small" miner are going to have to wait for whatever comes out of the Gekkoscience projects - and hopefully SOMEONE will start being willing to sell them chips on a reasonable basis soon in current-gen chips.
If the price was right I think it's a great size. If you compare with 2 x S5 that I run at the moment, you can go from 2.3GH/s for 1180W to 3.5GH/s for 850W in half the space. You can't ask better than that. 
Although I guess they could do a single board one, perhaps integrate the PSU where the second board is and have 1.6GH/s for 425W?
Rich
Although I guess they could do a single board one, perhaps integrate the PSU where the second board is and have 1.6GH/s for 425W?
Rich
I will stick with my original and the most obvious thought that they go from 3 boards to 2. No new design or testing needed, parts are common across all S7 variants and a significant cost reduction.
The only "problem" with this was that the maths did not add up to 3.5TH unless you clocked at 650MHz which given the problems they seem to have had at 600MHz seemed unlikely.
So my best current guess is that the 3.5TH rumour is slightly off and that it will be a 2 Board 3.1TH - 3.2TH S7 lite.
Rich
I would be totally fine with that, especially at a good price!The only "problem" with this was that the maths did not add up to 3.5TH unless you clocked at 650MHz which given the problems they seem to have had at 600MHz seemed unlikely.
So my best current guess is that the 3.5TH rumour is slightly off and that it will be a 2 Board 3.1TH - 3.2TH S7 lite.
Rich
Me too. the S7 is just out of reach for me so something a touch smaller would be perfect
still too big for my taste, it's only 1/4 less bigger, they should aim for 1/2 at least
for those that can't run a big miner in their home, and want to do only occasional mining activity
I will stick with my original and the most obvious thought that they go from 3 boards to 2. No new design or testing needed, parts are common across all S7 variants and a significant cost reduction.
The only "problem" with this was that the maths did not add up to 3.5TH unless you clocked at 650MHz which given the problems they seem to have had at 600MHz seemed unlikely.
So my best current guess is that the 3.5TH rumour is slightly off and that it will be a 2 Board 3.1TH - 3.2TH S7 lite.
Rich
I would be totally fine with that, especially at a good price!The only "problem" with this was that the maths did not add up to 3.5TH unless you clocked at 650MHz which given the problems they seem to have had at 600MHz seemed unlikely.
So my best current guess is that the 3.5TH rumour is slightly off and that it will be a 2 Board 3.1TH - 3.2TH S7 lite.
Rich
Me too. the S7 is just out of reach for me so something a touch smaller would be perfect
I'm not worried about the size. But if they can make one out of 2 blades for cheaper and make it better price per hash... i will be a fan.
I want as much hash as I can get for my money. Guess we see what we really get. It is interesting all this happened after kinda leak of avalon 6.
I will stick with my original and the most obvious thought that they go from 3 boards to 2. No new design or testing needed, parts are common across all S7 variants and a significant cost reduction.
The only "problem" with this was that the maths did not add up to 3.5TH unless you clocked at 650MHz which given the problems they seem to have had at 600MHz seemed unlikely.
So my best current guess is that the 3.5TH rumour is slightly off and that it will be a 2 Board 3.1TH - 3.2TH S7 lite.
Rich
I would be totally fine with that, especially at a good price!The only "problem" with this was that the maths did not add up to 3.5TH unless you clocked at 650MHz which given the problems they seem to have had at 600MHz seemed unlikely.
So my best current guess is that the 3.5TH rumour is slightly off and that it will be a 2 Board 3.1TH - 3.2TH S7 lite.
Rich
Me too. the S7 is just out of reach for me so something a touch smaller would be perfect
I guess we will have to wait and see
I will stick with my original and the most obvious thought that they go from 3 boards to 2. No new design or testing needed, parts are common across all S7 variants and a significant cost reduction.
The only "problem" with this was that the maths did not add up to 3.5TH unless you clocked at 650MHz, at which point V Core would be a bit low and given the problems they seem to have had at 600MHz seemed unlikely.
So my best current guess is that the 3.5TH rumour is slightly off and that it will be a 2 Board 3.1TH - 3.2TH S7 lite.
Rich
The only "problem" with this was that the maths did not add up to 3.5TH unless you clocked at 650MHz, at which point V Core would be a bit low and given the problems they seem to have had at 600MHz seemed unlikely.
So my best current guess is that the 3.5TH rumour is slightly off and that it will be a 2 Board 3.1TH - 3.2TH S7 lite.
Rich
BM1385 relay comms from one chip to the next so leaving out a single chip would disable every chip downstream from it - which in a string configuration means current ceases to flow in the circuit so even the first chips which are getting work data are starved for power and do not work.
Should be a way for Bitmain to bypass 1 or 2 chips on the circuit board per string, might need a small amount of glue logic in addition to required jumpers at worst, so they can run a couple chips less per string, run 2 boards instead of 3, and get 3.5TH out of the existing board design.
Or they could just aim at 3.2 TH with 2 of the existing boards.
I suspect the November 10 date is related to when Avalon plans to release the 6 - perhaps Bitmain feels the need to preempt upcomming competiton?
Honestly I believe it as it would push a lot of chips and units and it is something the home miners want, would love to see this and the avalon release (and possibly the bw miner), going to be a great generation!
This assertion seems reasonable.
Honestly I believe it as it would push a lot of chips and units and it is something the home miners want, would love to see this and the avalon release (and possibly the bw miner), going to be a great generation!
I like the comparison.
the lost generation: s1-s5, Sp10-30, av up to 4.1, asicminer (RIP), cointerra (RIP), knc (withdrew from selling)
the greatest generation-S7, bw?, av6 plus gekkoscience's sticks/pods
the silent generation-SP50?
baby boomers-will come after the halving, will be vast in numbers if price cooperates and employ improved 14/16nm for up to 0.05-0.07J/Gh. Network>1exahash.
bitfury is omnipresent, but invisible. I would call it "dark matter" or "dark energy"
My biggest fear is to see something like S5 with two blade in a plastic sided no full case device.
I will LOL so badly!
Any point for this miner will be to be quiet! other wise.... i don't really care
Do we have any sources on this rumor. I have people that also would like this to be confirmed or denied.
-D
-D
https://bitcointalksearch.org/topic/m.12769055
Honestly, nobody knows how accurate this is, hence it is 'rumors" in "speculation" rubric, NOT in hardware.
As a source it should be pretty accurate. If he truly was in a town purchasing 3PH of power I'm sure he would have talked to bitmain and avalon as it sounds he did.
The date of Nov 10th was thrown around so guess we just have to wait to see if Bitmain really did make such a big move. I suspect from reading thread he has avalon cloud he talks about for sell before this Nov 10th date. But guess we will have to wait and see.
Do we have any sources on this rumor. I have people that also would like this to be confirmed or denied.
-D
-D
https://bitcointalksearch.org/topic/m.12769055
Honestly, nobody knows how accurate this is, hence it is 'rumors" in "speculation" rubric, NOT in hardware.
Do we have any sources on this rumor. I have people that also would like this to be confirmed or denied.
-D
-D
If it's nice and balanced speed vs noise, then ok!
Otherwise give me s7++
Otherwise give me s7++
I doubt noise is a factor on it. I'm hoping they keep a full case design even though it's a little higher.
My biggest fear is to see something like S5 with two blade in a plastic sided no full case device. Not sure what to expect on the new version.
If it's nice and balanced speed vs noise, then ok!
Otherwise give me s7++
Otherwise give me s7++
Two chips per node instead of three would require a new PCB, not just not populating those chips.
Agreed. The "obvious" way to get 3.5TH is as I said above.
Quote
The simplest way to get it would be to go for a 2 Board version?
However to get 3.5TH from 2 of the current Boards you would need to run at 650MHz to give 3510TH which given the underclocked releases we have seen from 600MHz to 575MHz & 550MHz could be a problem?
650MHz is within the capability of the chip however the core voltage would be a little low for that hash rate. So if 2 boards I think they might need to change the number of chips on the boards?
However to get 3.5TH from 2 of the current Boards you would need to run at 650MHz to give 3510TH which given the underclocked releases we have seen from 600MHz to 575MHz & 550MHz could be a problem?
650MHz is within the capability of the chip however the core voltage would be a little low for that hash rate. So if 2 boards I think they might need to change the number of chips on the boards?
My concern is as stated. The 2 chips / node did not solve the problem either. I was just think typing
Rich
Two chips per node instead of three would require a new PCB, not just not populating those chips. BM1385 relay comms from one chip to the next so leaving out a single chip would disable every chip downstream from it - which in a string configuration means current ceases to flow in the circuit so even the first chips which are getting work data are starved for power and do not work.
Regarding Avalon, if the 6 works like the 4.1, it would suck to part them out because the machines are actually daisy-chained on an SPI bus. About every dozen machines or so (I forget the exact numbers) will be on a common bus to a USB adapter to the Pi. Parting out individual machines would mean having to stock additional USB adapters and such, and having to verify with the customer what the installation would be and how many accessory parts would be needed to get them all talking. It's not as simple as just tossing in an additional Pi.
Regarding Avalon, if the 6 works like the 4.1, it would suck to part them out because the machines are actually daisy-chained on an SPI bus. About every dozen machines or so (I forget the exact numbers) will be on a common bus to a USB adapter to the Pi. Parting out individual machines would mean having to stock additional USB adapters and such, and having to verify with the customer what the installation would be and how many accessory parts would be needed to get them all talking. It's not as simple as just tossing in an additional Pi.
Jump to: