Bitmain introduces the Antminer S17 Pro, Antminer S17, and the Antminer T17

[BitMiner247]

could you test how much power each cord is pulling on normal and turbo mode?

I’ve heard from someone who’s contacted Bitmain about this that the PSU splits the load equally between the two cords. So it should be, based on his reported numbers, 1025 watts per cord in normal mode and 1250 watts in Turbo.

I have the 53 TH/s model and my metered PDU’s, which only measure in amps, are showing ~11 amps total in normal mode on 208v power at 53-54TH. They also only show whole numbers and sometimes round up, so it’s not super accurate. That would put me at around 2288 watts or 1144 watts per cord.

[1715668722]

1715668722

[1715668722]

1715668722

[1715668722]

1715668722

[evanescence]

It is supposed to be based on the BM1397, correct?

https://chainbulletin.com/bitmain-announces-its-new-and-improved-7nm-mining-chip/

55J/TH seems low compared to what was initially advertised (30J/TH). Still incredibly impressive, and blowing everything else out of the water. But I wonder where this huge discrepancy is coming from.

[cmanmatt]

it’s not definitive until someone tests it, depending on the amount of hash boards and the controller board it’s possible one plug pulls more than the other, bitmain isn’t exactly the best designers.

I have metered smart switches for each circuit-PDU so the only way I could measure each cord would be to connect to two separate circuits which I'd rather not risk. Otherwise I'd need to measure with my amp-meter with a split cord which I don't have. If I can figure out how to safely do it I'll let you know.

[rz66]

Okay i found my answer by searching out new videos uploaded in the last few days

S17 Pro in Normal mode: power meter shows 1109 watts pulled from 1 plug, if it is in fact 50/50 that’s 2218ish watts so let’s say 42watt/T

Video below :



Antminer S17 Pro instruction and review

[philipma1957]

It is supposed to be based on the BM1397, correct?

https://chainbulletin.com/bitmain-announces-its-new-and-improved-7nm-mining-chip/

55J/TH seems low compared to what was initially advertised (30J/TH). Still incredibly impressive, and blowing everything else out of the water. But I wonder where this huge discrepancy is coming from.

you don't understand how they measure.

a single chip with test leads  can do 30J/TH

this gear is doing 50th at 2050 watts on = normal speed  that is 41J/TH

you are doing turbo numbers  to mine on turbo speeds gives you  a little more hash and burns power like mad.

so this gear does 41J/TH using the  fans and controller  on normal speed.  I suspect fans and controller are at least 100 watts

so it is  doing about 1950/50th = 39J/TH on normal speed

In what seems like a million years ago bit fury shows its chip in a single chip setup  it did 60J/TH  which when built into a  big machine did about 100J/TH

Turbo should never be used at the moment with current coin prices.

Normal speed should be used at the moment with current coin prices.

But I am sure some can create an example that shows turbo is smart to do at the moment.

[cmanmatt]

I have metered smart switches for each circuit-PDU so the only way I could measure each cord would be to connect to two separate circuits which I'd rather not risk. Otherwise I'd need to measure with my amp-meter with a split cord which I don't have. If I can figure out how to safely do it I'll let you know.

UPDATE: I remembered I had a dual metered PDU that reports amps (in whole numbers) that I don't use that readout anymore.  So I tried one of the S17s on that so I could measure each cord separately.  It measured exactly 5 amps each on Turbo mode.  Exactly being that it only reports round numbers.  At my voltage that is around 1150 watts per cord.  But I'm sure it is 5.x not just 5 amps.  But either way, it seems to balance the load very close between the two cords.

... Turbo should never be used at the moment with current coin prices.

Normal speed should be used at the moment with current coin prices.

But I am sure some can create an example that shows turbo is smart to do at the moment.

I'd like to see your calculation/thinking. 

By my profit model turbo mode makes sense currently.  But I am working with 5.4 cent electricity so that obviously will make a difference.  Simple Cryptocompare shows $8.50/day for Turbo, $7.84/day for Normal, $6.24/day for Low Power mode.  So Turbo currently makes the most profit.  I have wondered if the extra wear and tear is worth it and for that extra 8% profit and maybe not. 

As energy price goes up then it would flip to where normal would win out and then low power mode.

[kano]

... As energy price goes up then it would flip to where normal would win out and then low power mode.

You have to pay attention to Difficulty and BTC price long before you have to worry about energy price. You wont find your power bill changing every day (and no one pays it every day).

You will find Difficulty changing every 2 weeks and BTC price changing every day.

[philipma1957]

I'd like to see your calculation/thinking.  

By my profit model turbo mode makes sense currently.  But I am working with 5.4 cent electricity so that obviously will make a difference.  Simple Cryptocompare shows $8.50/day for Turbo, $7.84/day for Normal, $6.24/day for Low Power mode.  So Turbo currently makes the most profit.  I have wondered if the extra wear and tear is worth it and for that extra 8% profit and maybe not.  

As energy price goes up then it would flip to where normal would win out and then low power mode.

So your example clearly shows 66 cent benefit on turbo  which means  it needs to run for 3300 days (3300 x .66 = 2178.00)  without breaking to reach a clear benefit for 1 unit.

You would risk running new unknown gear at this moment with zero knowledge of durability hoping that the 66 cent day profit will make it worth your while.

"at the moment with current coin prices" is the key  point of my statement.

so in a month if no one has broken gear on turbo you have a base line.  And you would lose about 20 bucks waiting for others to test the gear out.

or if prices spike hard and you make 32 a day vs 28 a day  so turbo gains 4 dollars a day .

but you could be correct so you earn the 66 cents and never break the gear you win the bet. you earn more your way.

my guess is run it on normal for a month and re assess what to do.

[cmanmatt]

So your example clearly shows 66 cent benefit on turbo  which means  it needs to run for 3300 days (3300 x .66 = 2178.00)  without breaking to reach a clear benefit for 1 unit.

You would risk running new unknown gear at this moment with zero knowledge of durability hoping that the 66 cent day profit will make it worth your while.

"at the moment with current coin prices" is the key  point of my statement.

so in a month if no one has broken gear on turbo you have a base line.  And you would lose about 20 bucks waiting for others to test the gear out.

or if prices spike hard and you make 32 a day vs 28 a day  so turbo gains 4 dollars a day .

but you could be correct so you earn the 66 cents and never break the gear you win the bet. you earn more your way.

my guess is run it on normal for a month and re assess what to do.

Compelling argument. Thanks for elaborating!

[philipma1957]

Compelling argument. Thanks for elaborating!

It gets more complex when you are at 100 units. That is 66 dollars a day.

You could risk running at turbo until one breaks.

35 x 66 is about 2200. So every 35 days that extra profit pays for 1 unit.

So if 1 or 0 units break in 35 days it is a win or a push.

If  0 or 1 units break in 70 days it is a win.

This part of the you have to have money to make money.

100 units is about 200k worth of miners.

I know of a few people with 20 of these. They do normal speed with 4 cent power.

Down the road they will bump to turbo.  Since 20 earn them about 160 a day.  In 30 days they will earn 4800 that means 2 of the 20 are paid off.  They will know of the risks of turbo and will have at least paid off 2 units.

[BitMiner247]

UPDATE: I remembered I had a dual metered PDU that reports amps (in whole numbers) that I don't use that readout anymore.  So I tried one of the S17s on that so I could measure each cord separately.  It measured exactly 5 amps each on Turbo mode.  Exactly being that it only reports round numbers.  At my voltage that is around 1150 watts per cord.  But I'm sure it is 5.x not just 5 amps.  But either way, it seems to balance the load very close between the two cords.

I tested my 53TH/s pro on two separate banks on my PDU. Like yours my PDU only shows whole numbers and rounds up or down sometimes depending on the threshold. Together on the same bank both cords pull 11 amps on my 208v electric. When I plug each plug into their own bank my PDU’s show 5 amp draw per plug. I’m estimating it’s probably more around 5.5 amps so on 208v that would mean 1144 watts per plug. This is on normal mode so it looks like the PSU’s split the load evenly across the two plugs.

[dimaze]

You should definitely be running them on Turbo.

As long as your environment is a mining conducive one and there is no failure within the first few weeks you'll be fine with these long term.

[jhartbarger]

I received a S17 56TH unit today and seems it will only run at about 50TH in Normal or Low Power mode, anyone else seem similar behaviour?

[MoparMiningLLC]

I received a S17 56TH unit today and seems it will only run at about 50TH in Normal or Low Power mode, anyone else seem similar behaviour?

i wonder if you got a miss labeled S17 pro, it runs at 50 TH for normal mode - does it have turbo mode option? If not, I would be in contact with Bitmain - 6 TH is a bit of a variance, too big to overlook in my opinion.

and if you did get a S17pro - I would be asking for some money back as its about 75 USD cheaper.

[madmartyk]

i wonder if you got a miss labeled S17 pro, it runs at 50 TH for normal mode - does it have turbo mode option? If not, I would be in contact with Bitmain - 6 TH is a bit of a variance, too big to overlook in my opinion.

and if you did get a S17pro - I would be asking for some money back as its about 75 USD cheaper.

I agree, my S17 is running at a constant 55194.09 in normal mode.

[jhartbarger]

i wonder if you got a miss labeled S17 pro, it runs at 50 TH for normal mode - does it have turbo mode option? If not, I would be in contact with Bitmain - 6 TH is a bit of a variance, too big to overlook in my opinion.

and if you did get a S17pro - I would be asking for some money back as its about 75 USD cheaper.

Nope standard S17 56th unit, it appears bitmains auto-tune function is broken because my my first 2 hashboards are avg 18000gh each and board #3 is averaging 12000gh and I only have Normal and Low Power, no turbo.

[MoparMiningLLC]

Nope standard S17 56th unit, it appears bitmains auto-tune function is broken because my my first 2 hashboards are avg 18000gh each and board #3 is averaging 12000gh and I only have Normal and Low Power, no turbo.

What sucks more is if it is defective, you will have to send it in and lose precious time mining.

[minefarmbuy]

Two reports of this?

[OgNasty]

Nope standard S17 56th unit, it appears bitmains auto-tune function is broken because my my first 2 hashboards are avg 18000gh each and board #3 is averaging 12000gh and I only have Normal and Low Power, no turbo.

Isn’t that how it’s supposed to be? I thought only the pro model had 3 settings. There is also a high amount of variance listed in the specs.

[MoparMiningLLC]

Isn’t that how it’s supposed to be? I thought only the pro model had 3 settings. There is also a high amount of variance listed in the specs.

we were asking if he might have gotten a pro because it was hashing at 50th - which is the advertised speed of the pro version in normal mode - the non pro versions are either 53th or 56th if I recall correctly.