Topic: Bitmain Underclocking! (Read 1021 times)

maybe  this can be changed over the next 6 to 9 months. it would be nice if possible for it to happen.

I know you are looking to do 18 chip boards of the s-5 chips,

 but maybe just maybe it will be s-7 chips updating an s-5  case .

He's got a pretty good plan. One of the main reasons to buy the S2 was because it was modular and could be upgraded quite easily. Having a backplane with swappable boards and an easy-to-flash controller makes the thing very futureproof if you care to make upgraded cards for it. The S3 Upgrade Kit was another step in the right direction, with the obvious problem that it cost pretty much the same as buying an S3. I know the chips and other silicon on the PCB are a significant portion of the up-front cost of a miner (where most of the ASIC cost is actually dev) but being able to send a dozen PCBs in a shoe box instead of a 25lb case with heatsinks is such a good idea. I really don't like how few options there are for things like this from any manufacturer.

yes it is a good boost, especially because something like this will prolong the life of your miner and i always prefer a lower wattage and lower hash than the opposite, but the real boost will come only with 16nm or less

sadly i don't own any antminers, i can't help you on your needs

There is a massive difference between can and will. I think Bitmain has approximately zero incentive to take in old gear. What would they do with it? Bitcoin mining is rapidly changing to "Efficiency is King", if it's not already there.

Just my opinion.

Another alternative (I hate saying there name cause they can go to hell) BFL did when it did its first fgpa's

It gave customers a credit against there next miners if they sent them back. (Cause they could be reused for other projects)

Bitmain could issue a refund program after 6 months or as the next get of miner comes online, Send back your old miner & discount the next one.

This builds good customer relations by getting continued support of there own products, Virtually locks the customer into a buying & upgrading every time (U need money) Or create a new miner.

In turn helps the environment by keeping the parts out of landfill longer, Reduces the need to mine new minerals & gets them back nearly everyone of the chips to then downclock & resell at a higher hashrate in total at the same power usage essentially giving the miner another 3 to 6 months of life...

The scenario here i see uses something like a S2 box & setup with cards that can be swapped in & out.

Customer buys machine with however many cards u want to start with (Say 10) with i don't know cause i don't have a S2 (Say 20) chips as a standard machine running 1th at 1400w or better would be 1500w for headroom.

Upgrade for customer consists of user sending boards back (Include a prepaid return to sender padded box - Add it into the cost of the miner if the user wants this option so theres nothing to pay later for postage)

Then charge a half cost of the original miner to the user & send back 10 cards now with 60 chips on it (Off course this will be a first in first served basis or a good way to run out lower class excess chips or chips that are left in stock)

Anyone else care to chime in??

Hey if they can release an S4, Then an S5 & then go back and release an S4+ with an older chip nearly 12 months later, They can do anything..

Im no rocket scientist so correct me if im wrong, When they do chips big scale u get class 1 chips, class 2 chips & so on??

When they build current miners they would want all class 1 wouldn't they??

If so why not use the lower class chips that may have been binned for in-efficiency??

Does anyone but me think all bitcoin miners should be built at maximum underclock & volt & set them up to run a max 1500w.

This will give the best efficiency, With liquid cooling u could nearly kill the noise problem overnight

Even with fan cooling noise would be a half a current S4 is pumping out at 60db and more.

Starting a this point would then allow the creator to let the kids go harder & offer a 2nd solution to this miner by offering to sell it with a bigger power supply for an addition fee giving people to options say like a Max 2000w

Are u not happy with a 7% boost in efficiency? Cause at this point there is nothing new on the market, Any boost on what we have is still better than what we have...

This will make ur miner run quieter cause the fans don't need to be so fast to cool them...

And i notice u said its 7% better than the S5.... I was making the comparison using specs from S4 chips so make it more like 12 - 15% better than an S4...

I dont actually have a S5 so can u tell me what the best underclock using the standard bitmain interface

Or better still can u tell me what freq & volt u can set the miner up at

ie the S4 does

Freq 200
volt  0725

does 2th at 1400w

at

Freq 100
volt 0725

does 1th at just over 700w

It's sorta what they were trying to do with the S4+, set at about 700mV and gaining 10-15% because no VRMs. It's a good idea, if it weren't expensive - the number of chips required to get appreciable hashrate from low clocks goes up linearly, and so does the chip cost (which is the most significant factor in machine cost).

with 3024 and  consumption of 1443 the performance is a little above the that from a antminer s5, roughly 7%

because you have 3024:1443=1155:590 = 2,09=1,95 difference is 0.14 which is around 7%

On another note how about building the biggest miner u can but undervolted???

Think about this for a minute cause i dont want u to say fuck we will just boost the power & overclock it...

Think about this... (This is what bitmains S4 was based on so im using there data)

If a S4 could run 40 chips (As it does) and downclock to 0.60v or 0.48 w/g on wall



and achieve 1000gh (1th) 481w on the wall.

Then u could achieve a 3024gh (3th) miner running at 1443w at the wall....

----------------------------------------------------------------------------------------------------


Hey bitmain if u have so many BM1382 asics left over here's an idea!