antminer u1 vs s1 | Bitcointalksearch.org

ineedit

sr. member

Activity: 252

Merit: 250

Quote from: philipma1957 on March 15, 2014, 03:48:28 PM

.....
yep it looks like they have figured out the minimum for watt/gh

It must be 640 chips running at 1.56gh each = 998.40 gh

since the chip can run 1.6 to 2.0 with the same voltage as set by the resistor on a u-1

the u-1 voltage is set far too high too run 1.6gh

I have done extensive u-1 runs with clocks from 1.5 to 2.1 and watts do not vary much. this means the volts are set too high on u-1 to run power efficient at 1.6 gh.

Since the new u-2 will run at 2.0-2.1 with same watts it is a confirmation of this in a backhanded way.

Now what I would like to see is space for more PSU's and a three pole switch, left for low power 1TH/1KW, middle for 1.5TH/2KW, right for 2TH/3KW.

EDIT: Or Summer, Spring/Autumn and Winter mode

philipma1957

legendary

Activity: 4102

Merit: 7765

'The right to privacy matters'

Quote from: elemental on March 15, 2014, 03:04:47 PM

Quote from: repairguy on March 08, 2014, 04:42:07 PM

I am not looking to purchase, just a wandering mind. I already have s1 units and love them. It would appear from your measurements and chip count that the u1's are only more efficient because they are clocked slower. One could simply underclock the s1 to get similar efficiency levels.

80watts might not sound like much but it is almost 30%. It could mean the difference between shutting them off or keeping them alive for a couple more difficulty changes. Like I said I am not looking to purchase anything just curious for verification of the actual w/ghs.

It is also (and more significantly) the core voltage of the chip itself - the U1 is configured to use a lower voltage than the S1, which is why you can't clock a U1 as fast as an S1. Theoretically, ignoring the other components, they should be identical with identical clocks and voltages.

On a related note, I was wondering what how the S2 is being set up. It seems to be using the same chip, but even more efficiently.

yep it looks like they have figured out the minimum for watt/gh

It must be 640 chips running at 1.56gh each = 998.40 gh

since the chip can run 1.6 to 2.0 with the same voltage as set by the resistor on a u-1

the u-1 voltage is set far too high too run 1.6gh

I have done extensive u-1 runs with clocks from 1.5 to 2.1 and watts do not vary much. this means the volts are set too high on u-1 to run power efficient at 1.6 gh.

Since the new u-2 will run at 2.0-2.1 with same watts it is a confirmation of this in a backhanded way.

elemental

newbie

Activity: 49

Merit: 0

Quote from: repairguy on March 08, 2014, 04:42:07 PM

I am not looking to purchase, just a wandering mind. I already have s1 units and love them. It would appear from your measurements and chip count that the u1's are only more efficient because they are clocked slower. One could simply underclock the s1 to get similar efficiency levels.

80watts might not sound like much but it is almost 30%. It could mean the difference between shutting them off or keeping them alive for a couple more difficulty changes. Like I said I am not looking to purchase anything just curious for verification of the actual w/ghs.

It is also (and more significantly) the core voltage of the chip itself - the U1 is configured to use a lower voltage than the S1, which is why you can't clock a U1 as fast as an S1. Theoretically, ignoring the other components, they should be identical with identical clocks and voltages.

On a related note, I was wondering what how the S2 is being set up. It seems to be using the same chip, but even more efficiently.

repairguy

sr. member

Activity: 252

Merit: 250

Quote from: philipma1957 on March 08, 2014, 11:35:10 AM

Quote from: repairguy on March 08, 2014, 07:33:37 AM

I notice the antminer u1 is significantly more efficient than the antminer s1. Does anyone have a solid explanation for this? I know the s1 controller uses power, but not that much!

Is it just the difference between the conversion of 12v (s1) vs 5v (u1)

Or perhaps the density of the chips to each other vs a single usb?

Does anyone have a slew of them on an externally powered usb hub that can verify readings at the wall?

yes and you are correct they use less power per gh. but they cost a lot more up front.

all power is due to the resistors used in the u-1 they send less power to the chip.

an s-1 has 64 chips = 180 hash

90 u-1 have 90 chips = 180 hash

if you buy an s-1 it is 1.11 btc and a good psu = 100 usd for a platinum total is close to 700 usd for 180gh burning 360 watts

90 sticks are 2.7 btc plus 2x 49 port hubs are .3 btc = 3 btc or 1860 usd lastly 2 good psu's run 200

total of 2060 usd for 180 gh burning about 280 watts.

I had 45 u-1's with a plat seasonic the used 3.1 watts

so to save 80 watts an hour for the same hash you spend about 2060 vs about 700 usd.

all testing done with k-watt-meters

u-1's have a place

but if you have money s-1's are far better cost wise

if you want a usa seller

https://www.minersource.net/

has a kit with 2 s-1's and
a good psu

https://www.minersource.net/products/bitmain-s1-dual-blade-180gh-400w-asic-miner-1-2-3

this is a better deal then u-1's will ever be.

Note you can buy direct from china and source your own psu and it would be a little cheaper. but may take longer to show up and longer for rma's

I purchased direct from china and got 1 dud out of 4 purchased.

I am not looking to purchase, just a wandering mind. I already have s1 units and love them. It would appear from your measurements and chip count that the u1's are only more efficient because they are clocked slower. One could simply underclock the s1 to get similar efficiency levels.

80watts might not sound like much but it is almost 30%. It could mean the difference between shutting them off or keeping them alive for a couple more difficulty changes. Like I said I am not looking to purchase anything just curious for verification of the actual w/ghs.

repairguy

sr. member

Activity: 252

Merit: 250

Quote from: Xer0 on March 08, 2014, 10:46:11 AM

is it?

S1 is 2W/GH at Wall, U1 is 2.5W/1.6GH at USB
so you have to add the Hub/PC PSU efficiency

The hub is going to be 99.9999% efficent. The only thing that is in it is a bridge IC (or a few depending on the size of the hub)

Yes the power supply efficiency would come into play but would be the same efficiency running an s1 or u1( I assume the power supply efficiency doesn't vary much if you are drawing the 5v or 12v.)

philipma1957

legendary

Activity: 4102

Merit: 7765

'The right to privacy matters'

Quote from: repairguy on March 08, 2014, 07:33:37 AM

I notice the antminer u1 is significantly more efficient than the antminer s1. Does anyone have a solid explanation for this? I know the s1 controller uses power, but not that much!

Is it just the difference between the conversion of 12v (s1) vs 5v (u1)

Or perhaps the density of the chips to each other vs a single usb?

Does anyone have a slew of them on an externally powered usb hub that can verify readings at the wall?

yes and you are correct they use less power per gh. but they cost a lot more up front.

all power is due to the resistors used in the u-1 they send less power to the chip.

an s-1 has 64 chips = 180 hash

90 u-1 have 90 chips = 180 hash

if you buy an s-1 it is 1.11 btc and a good psu = 100 usd for a platinum total is close to 700 usd for 180gh burning 360 watts

90 sticks are 2.7 btc plus 2x 49 port hubs are .3 btc = 3 btc or 1860 usd lastly 2 good psu's run 200

total of 2060 usd for 180 gh burning about 280 watts.

I had 45 u-1's with a plat seasonic the used 3.1 watts

so to save 80 watts an hour for the same hash you spend about 2060 vs about 700 usd.

all testing done with k-watt-meters

u-1's have a place

but if you have money s-1's are far better cost wise

if you want a usa seller

https://www.minersource.net/

has a kit with 2 s-1's and
a good psu

https://www.minersource.net/products/bitmain-s1-dual-blade-180gh-400w-asic-miner-1-2-3

this is a better deal then u-1's will ever be.

Note you can buy direct from china and source your own psu and it would be a little cheaper. but may take longer to show up and longer for rma's

I purchased direct from china and got 1 dud out of 4 purchased.

Xer0

hero member

Activity: 826

Merit: 1000

°^°

is it?

S1 is 2W/GH at Wall, U1 is 2.5W/1.6GH at USB
so you have to add the Hub/PC PSU efficiency

repairguy

sr. member

Activity: 252

Merit: 250

I notice the antminer u1 is significantly more efficient than the antminer s1. Does anyone have a solid explanation for this? I know the s1 controller uses power, but not that much!

Is it just the difference between the conversion of 12v (s1) vs 5v (u1)

Or perhaps the density of the chips to each other vs a single usb?

Does anyone have a slew of them on an externally powered usb hub that can verify readings at the wall?

Topic: antminer u1 vs s1 (Read 1235 times)