Bitcoin Forum

Summer is approaching and to many people, the high temperature that comes with summer causes a lot of issues, in this topic I have compared the temperature of the following gears from Bitmain.

S17pro in Low power mode
S17pro in Normal mode
S17+ 67th
T17 42th
T17e 54th

https://i.ibb.co/MGLPwFG/compare.jpg (https://i.ibb.co/MGLPwFG/compare.jpg)

To make this comparison somehow accurate I set all fans to 90%, by the way, many people don't know to do this for the new models, you can follow this guide (https://bitcointalk.org/index.php?topic=5202112.msg53096836#msg53096836).

I believe these results are obvious for the most part, the only unexpected result is the T17e vs T17, I was expecting the T17e to run a lot hotter than the normal T17 given the power consumption, but that was not the case.

I have yet to test S17 and T17+.

Excellent work! Thanks. How much is room temperature? There is any extraction fan?

To add a bit of info would help.

I know the watts used per th on my s17 pros is around 36 or 37 a th.

and using  the same pro on normal it gets about 39-40 watts a th.

So watts used would be nice.

it's about 20 degrees Celsius at the time of taking that screenshot.


Yes there are, I use a split-room design to separate hot and cold air, the ventilation is not perfect but it's pretty good, you can refer to this topic  (https://bitcointalk.org/index.php?topic=5087770.0).


Unfortunately, I don't really measure power consumption per miner (i guess you know why :D) but once/if I have the chance to do it, I will be more than willing to do so.

I do know why.  when i go back to clifton I will try to do it.

Hi,

We did some reviews in January in our facility comparing  S17PRO_50TH/s vs S17PRO_53TH/s.
Power consumption ( average ) for S17PRO_50TH/s is 2081W and for S17PRO_53TH/s is 2217W.
Power consumption ( average ) for model T17e_53T is 3130W.

 Maybe it is useful info.

My t17e's were far worse  3400

but my s17pro 50th
and my s17pro 53th

were close to an exact match.

3400-3130 = 270w difference, there must be something wrong here, it's pretty strange that you two got the same wattage on the S17 pro but quite a large difference in the T17e, I trust phill, I also know TeslaWatt has no reason to actually give false data, I hope someone else can chime in with their results, teslawatt's results are pretty close to the specs indicated by Bitmain, maybe phill got so unlucky and received a miner with a bad PSU?

2 t17e were terrible   3330  and 3400.   So I said fuck it and did not buy any more of them

1 t17+ drops a  board over  and over again.

these were the worst 3 units bitmain sold to me  in 2019

say
 4 s9k's
 2  t17
 1  s17
14 s17pros maybe 15

and the 3 above.

I am not complaining or bitter  it is just that those 3 are the worst.

every other piece of gear mentioned above was good  or better then good.

21 good or better then good
2 17e  pull 6730 watts vs 6230

1 t17+ has a bad board.

I have 2 units of T17 that won’t mine due to overheating on the hottest time of the day. I think both units has a bad chip that is running hotter than the rest which resulted the fans to spin at max rpm  just to cool down that particular chip. Even at night the fan is spinning close to 6000rpm whereas others are doing 3000-4000rpm.

You could disconnect the bad board and get 2 of 3  boards to work.

Yeah, I supposed I could do that and connect it back again when the hot season is over. Thanks for the suggestion!

Sometimes something may stick on the heatsinks of one or a few of the chips, you might solve this issue by dusting the miner off, another way around would be using a known and trusted custom firmware that gives you the ability to undervolt/underclock the miner, I believe using Asic.to or Awoesomeminer's firmware you will be able to adjust the voltage of each chip, you can lower the voltage on those bad chips so that the rest of the chips can mine at full speed.

mikeywith has a point so if you open the t17 to drop a board.

you could look for dirt or a bug 🐜 clogging a spot and overheating 🥵 the unit.

if it is all clean pull the board,  or use aftermarket firmware.

2 more shots at full hash

and 1 shot at 2/3 hash.