Need advice on S9 vs T17

[MSDA89]

Hello everyone

I need some advice. I have 100,000$ to spend on a small ASIC farm. I have read about different miners and contacted some sellers and friends overseas and bought s9i which is currently working well for testing purposes. I found the best two options for my plan but need some advice on which is best for my setup.

I will be using a cooling immersion method, it’s not included in the above budget. The electricity rate is between 0.02c to 0.05c per KW

The two miners are S9J with apw7 PSU and T17 42Th

The s9j current cost is about 500$ which means I can get 200 miners equal to 3400Th+-10%, 17Th per miner overclocked. About 330 kilowatt

As for the T17 costs about 1800$ and can get about 55 miners equal to 3025Th+-10%, 55Th per miner overclocked. About 165 kilowatt

I  did some calculations on https://btc.com/tools/mining-calculator  both miners perform almost the same on average. On 0.02c per KW S9j perform a little better but when  it’s 0.05c per KW seems the T17 getting better results

Some other notes

*S9J is older than T17  but I read a lot about heating issues in the 17 series and the high failure rate

*The miners will be used until they stop being profitable or broken beyond repair

*I have a specialist who will be fixing the broken miners if needed

*Cooling cost is not included  and it doesn't matter in this stage  even though s9j require more cooling and space
 
As for me personally, I’m leaning toward the T17 for the long term but not sure if the immersion cooling will reduce the risk of failing

So what do you think?

Did I miss something?

[mikeywith]

*I have a specialist who will be fixing the broken miners if needed

Will you be paying him a salary and he would fix all the units you have at no extra cost and in a timely manner? if not  (which is probably the case),  then this point is irrelevant, fixing those hash boards requires both time and money, with the high failure rate your guy will either ask you to pay him more or run away before his back needs some major surgery because he going to be doing a ton of fixing.
 
when you went shopping online didn't you ask yourself why do those T17 go for a lot cheaper than other miners of the same generation and efficiency? well, it's because these gears are nothing but crap.

The S9js are more robust than the T17 by a few orders of magnitude, but they are also pretty inefficient, and 2-5c isn't going to be spared when the bear market hits us and mining becomes barely profitable, with the power rate you have you need something in the middle of efficiency, 40-50w/th is good, this way, you utilize your below-average power rate by not having to pay 100$ per TH for 30w/th gear, and you also won't be forced to shut down soon when the 90w/th gears are no longer profitable at 2-5 cents.

Your best shot would be MicroBT M20s, with about 45w/th and it costs roughly 60-65$ per TH and it's as robust as any mining gear can be, I would not touch the 17 series (without any exception) from bitmain with barge-pole, unless if they are so cheap and you can ROI in like 1-2 months, then it's MAYBE worth taking the risk.

[MSDA89]

*I have a specialist who will be fixing the broken miners if needed

Will you be paying him a salary and he would fix all the units you have at no extra cost and in a timely manner? if not  (which is probably the case),  then this point is irrelevant, fixing those hash boards requires both time and money, with the high failure rate your guy will either ask you to pay him more or run away before his back needs some major surgery because he going to be doing a ton of fixing.

He will be working full time

when you went shopping online didn't you ask yourself why do those T17 go for a lot cheaper than other miners of the same generation and efficiency? well, it's because these gears are nothing but crap.

The S9js are more robust than the T17 by a few orders of magnitude, but they are also pretty inefficient, and 2-5c isn't going to be spared when the bear market hits us and mining becomes barely profitable, with the power rate you have you need something in the middle of efficiency, 40-50w/th is good, this way, you utilize your below-average power rate by not having to pay 100$ per TH for 30w/th gear, and you also won't be forced to shut down soon when the 90w/th gears are no longer profitable at 2-5 cents.

I know that bitmain miners are not the best but it's hard to find information and support for whatsminer and other brands. I live in the Middle East which is hard to find someone fixing miners for you. I met two and none of them know how to deal with miners other than antminer

Your best shot would be MicroBT M20s, with about 45w/th and it costs roughly 60-65$ per TH and it's as robust as any mining gear can be, I would not touch the 17 series (without any exception) from bitmain with barge-pole, unless if they are so cheap and you can ROI in like 1-2 months, then it's MAYBE worth taking the risk.

I will take this into consideration

Thank you

[wndsnb]

I'm skeptical that you'd be able to reliably get 55th on all those t17s. The stock T17 is already overclocked... it is exactly the same chip as an S17 except run at higher voltage and frequency per chip. They put fewer chips per hashboard and gave them larger heatsinks.

You can pretty much count on being able to get much better efficiency when underclocking though.

If you have someone local who can repair them, I would not eliminate the T17 as a possibility. They are easier to repair than the S17s since they have fewer chips that are more spaced out. And when underclocked you could get better than 40w/th on some of them, so when the market finally catches up you have a chance at still being profitable.

At least in the US, the listings I've seen for M20s are closer to $80/Th for 10 or more miners. Close to double the price of the T17. For that I'd take the risk of the T17, or I just would just keep buying btc and wait for better conditions to buy gear.

[MSDA89]

Code:

I'm skeptical that you'd be able to reliably get 55th on all those t17s. The stock T17 is already overclocked... it is exactly the same chip as an S17 except run at higher voltage and frequency per chip. They put fewer chips per hashboard and gave them larger heatsinks.

You can pretty much count on being able to get much better efficiency when underclocking though.

I didn't try it myself  but some firmware claims 55~Th by using immersion cooling

https://hiveos.farm/asic/t17/
https://vnish.farm/en/firmware/AntMiner_T17
https://asic.to/

Code:

If you have someone local who can repair them, I would not eliminate the T17 as a possibility. They are easier to repair than the S17s since they have fewer chips that are more spaced out. And when underclocked you could get better than 40w/th on some of them, so when the market finally catches up you have a chance at still being profitable. 

How often do you think I might run into a problem with this miner? I saw a video a few months back about someone who has around 64 ASIC from miners both models S17 and T17. He kept running into problems with both models, Once he change the miners from air to liquid cooling, most of the problems got solved.

Code:

At least in the US, the listings I've seen for M20s are closer to $80/Th for 10 or more miners. Close to double the price of the T17. For that I'd take the risk of the T17, or I just would just keep buying btc and wait for better conditions to buy gear. 

Do you think prices will go down anytime soon? T17 was 1260$ two months ago. I might wait a little more while doing some research on the M20s and M30s

[kano]

...
https://hiveos.farm/asic/t17/
https://vnish.farm/en/firmware/AntMiner_T17
https://asic.to/
...

Alas all those 3 firmware violate my cgminer license ...