Asic.to Firmware 17Series • Optimized efficiency/Performance •

[Scorpyy]

Some risks are not worth taking. I would personally either RMA the unit or buy new temp sensors.

Thank you for your suggestion. Do you know where can new temp sensors be bought?

[AniceInovation]

Hey there, sounds like an awesome firmware Taserz.
Do you have any numbers on underclock on S17+ and T17+ machines?

They are cheap right now and am considering using them with this.
Do they get better efficiencies than the non plus models?

[taserz]

https://www.digikey.com/product-detail/en/texas-instruments/TMP451AIDQFR/296-39298-1-ND/5142964

https://www.zeusbtc.com/Repair.asp

Hey there, sounds like an awesome firmware Taserz.
Do you have any numbers on underclock on S17+ and T17+ machines?

They are cheap right now and am considering using them with this.
Do they get better efficiencies than the non plus models?

Yes I don't have numbers or tune configured yet but yes

[AlecMe]

@taserz I understand the below has been answered, BUT I would like that answer to come from you. It has something to do with accountability if know what I mean.

I am only interested in the S17+ model for the below.

Do you want to answer these questions

[Q1]is the s17+ 105TH speed for immersion, or normal air cooling operating medium?
I need to make sure I am making a profit, after the increase in speed and dev fee calculation, hence asking.

Also, [Q2]what is the wattage increase at the OC speed? and,

[Q3]will the PSU need extra cooling, will it be ok in the long term?

what about the dev fee, as halving is soon. [Q4] will it be lowered?

I hope you will not avoid answering these questions, again...

[taserz]

@taserz I understand the below has been answered, BUT I would like that answer to come from you. It has something to do with accountability if know what I mean.

I am only interested in the S17+ model for the below.

Do you want to answer these questions

[Q1]is the s17+ 105TH speed for immersion, or normal air cooling operating medium?
I need to make sure I am making a profit, after the increase in speed and dev fee calculation, hence asking.

Also, [Q2]what is the wattage increase at the OC speed? and,

[Q3]will the PSU need extra cooling, will it be ok in the long term?

what about the dev fee, as halving is soon. [Q4] will it be lowered?

I hope you will not avoid answering these questions, again...

1. Your not pushing over 100T without liquid because it will draw to much power and psu will cut out. Colder = more efficient.

2. It depends it tunes itself sometimes it's less than factory w/th sometimes it is more it depends on the enviorment.

A client just sent me this

On the 91.5 profile when auto tuned it draws 3911 w.  On the 84th profile when auto tuned it draws 3527 w.  Out of the box it draws 2680w.

3.Psu is fine. Psu tapps out at around 4-4.1kw heating issues are rare but the psu is limited.

4. We are not sure yet but the underclocking on these machines is honestly crazy I am doing 966 watts at 38th/s on an S17

[favebook]

[...]

I could get you numbers for S17+ and S17 Pro.

[...]

Best achievements from my overclocking (S17+) are as follow (Wattage is read from software - I'll need to buy second power meter to measure it correctly):
95 TH/s - 3900W - 41.4 W/TH
92 TH/s - 3600W - 39.4 W/TH
85 TH/s - 3300W - 38.8 W/TH
50 TH/s - 1550W - 31.2 W/TH

Haven't gone lower than 50TH/s as I want to mine as much as possible before halving. But I will be aiming for less than 30W/TH on all of my machines after halving.

[fyoung55]

The software reporting of watts tends to be off by a fair bit (anywhere from 125 - 250 watts) on my s17 pro.  The more I under-clock, the greater the discrepancy, perhaps its fan wattage draw as the fan has to run faster on higher frequency/voltages to keep machine cool.  In all cases the chips stayed right around 75 as the software is programmed.  I would be interested to see how others are fairing in terms of at the wall watts.

Apologies as I couldn't figure out how to insert a table...

Actual Data:

Auto tune setting          asic watts                           wall watts                        discrepency
38                               819 (21.6 w/TH)                 1065 (28.0 w/TH)             246w (27%)
43                               1206 (28.0 w/TH)               1445 (33.6 w/TH)             239w (20%)
48                               1484 (30.9 w/TH)               1710 (35.6 w/TH)             226w (15%)
53                               1752 (33.1 w/TH)               1980 (37.4 w/TH)             228w (13%)
58                               2080 (35.9 w/TH)               2245 (38.7 w/TH)             165w (8%)
62                               2361 (38. w/TH)                 2500 (40.3 w/TH)             139w (6%)
67                               2745 (41.0 w/TH)               2870 (42.8 w/TH)              25w (5%)

For fair comparison on the same  rig:

Standard firmware

mode                      wall watts     
low (40.5)               1755 (43.3 w/TH)
normal (51)             2330 (45.7 w/TH)
high (57)                 2655 (46.5 w/TH)

There is quite a bit of power savings at the wall as you can see using the modded firmware, especially if under clocking.

[AniceInovation]

Excellent, would love to see some numbers on a T17+ also.
Was sounding good in my head the numbers running, only to realize today halving is coming in ~20 days

lol

I really can't justify it based on ROI numbers, but i wish you all good luck

[taserz]

[...]

How hot is it where you are?

I just tuned an S17Pro 62th/s to compare to you numbers software reported 2532W

so 2532W - Sofware

I measured 10.76amp @240 = 2582.4 (small loss from pdu) your numbers seem WAY OFF. I checked with a few clients and the biggest range I am seeing in power reading being off from the watt meter is 4%.

[fyoung55]

It's 60 - 65 F pretty consistently.  Here's something I noticed, your software wattage reported (2532w) is way higher than what my software reported (2361w) at 62 TH auto tune, BUT it seems right in line with my draw at the wall measurement (2500w).  So that's very interesting to me...

Further background on my situation.:

The watt-meter I am using is hard wired right into the breaker panel, circuit specific for my miner.  I do have the 50T s17pro version if that matters at all.  I reset the firmware and re-tuned for 58 TH.  This time it came in a 2010W on software and 2200w on panel.  Maybe my watt meter is reading high, or maybe my software is reporting low but that seems unlikely.  Standard specs from bitmain claim 1975w for 50T and I'm registering 2330w for 51T... that's a pretty big jump right there as well, so maybe it's my meter?  I'd be interested to see other peoples numbers.

And again, I compared the stock firmware to the custom using the same at the wall watt meter, so even i it's off, it's off for both equally and should still give a reasonable comparison. As I said before, there is still a significant power savings using your firmware. I'm not arguing the efficiency improvement at all, just trying to figure out the discrepancy between software and "actual."

Here are the pics of the miner, currently in action!
https://imgur.com/a/GWnq0H6

[jacktman]

Hello greetings to all, currently my T17 after 5 months running smoothly has started to give me some problems;

Code:

2020-04-25 22:20:38 driver-btm-api.c:1443:dhash_chip_send_job: Verion num 4
2020-04-25 22:25:37 thread.c:384:is_zero_nonce_happened: chain_nonce[0] = 15645
2020-04-25 22:25:37 thread.c:384:is_zero_nonce_happened: chain_nonce[1] = 15748
2020-04-25 22:25:37 thread.c:384:is_zero_nonce_happened: chain_nonce[2] = 15840
2020-04-25 22:39:39 register.c:290:get_register: !!! REG_TYPE = 1. 939572226
2020-04-25 22:40:00 temperature.c:683:get_temp_info: read temp sensor failed: chain = 0, chip = 64, reg = 1
2020-04-25 22:40:19 temperature.c:683:get_temp_info: read temp sensor failed: chain = 0, chip = 48, reg = 1
2020-04-25 22:40:39 temperature.c:683:get_temp_info: read temp sensor failed: chain = 0, chip = 168, reg = 1
2020-04-25 22:40:59 temperature.c:683:get_temp_info: read temp sensor failed: chain = 0, chip = 184, reg = 1

and stop mining, this occurs between 12 noon and 6 or 7 p.m.,then it seems to work normally, the ambient temperature is about 31°c to 34°c
I'm not sure if it's a room temperature problem or a temperature sensor problem.
looking for some information I found your firmware, I have some questions;

1) Does your firmware solve this problem?
2) i have developed a script to keep control and check all my miners (hashrate, temperature, fans, ASIC status, etc) it shows the status of all the miners on one screen, it sends me email alerts if something is below the set values, i use socket connections and get the "json" response, with your firmware can i keep asking the miner for this information and get the same response by json?

here you can see a screenshot of my script: https://imgur.com/a/j5uKQOr

thanks.

[taserz]

[...]

This is one of the meters that ties around the breaker right. I am using a clamp meter doing the same thing. This is odd. But these meters are generally pretty darn accurate. Does your breaker have 1 power lead or two (bigboi 30amp break).

[...]

That is a pretty hot room in general.... BUT

Temp sensors are STUPID FUCKING COMMON PROBLEM. Honestly it's more so your temp sensor shit the bed. Rebooting it might get it working or flashing the recovery firmware then flashing back.

you can call the bmminer-api stats to get the info. I am not sure if there is a webhook for it I will check.

[mikeywith]

Temp sensors are STUPID FUCKING COMMON PROBLEM.

Have you replaced any of those temp sensors? are they the same once used on the old models like Antminer S9? I tried googling and found no tutorial/guide on how to identify and replace these temp sensors, it seems like this is going to be a trend just like the "I lost a hash board" on the S9.

[fyoung55]

This is one of the meters that ties around the breaker right. I am using a clamp meter doing the same thing. This is odd. But these meters are generally pretty darn accurate. Does your breaker have 1 power lead or two (bigboi 30amp break).

Yep, it's 220v so 2 hots.  It's only a 20 amp breaker, but I only have the one miner so I am well below max current draw.  It's also a short run, miner is 10 feet from the panel.  I use the exact same model watt meter for my EV charger and it reports fairly accurately there as best I can tell.  I may pull the EV charger one and make a "home made watt meter with outlet" and plug miner into that and see what that reports.  If they report the same then I'm not sure where to go from there.  Just a very bizarre scenario!

[jacktman]

... or flashing the recovery firmware then flashing back.

taserz, thanks for your answer, yes it seems to be a common problem in this model  , where can I find the recovery firmware file?
I was looking in bitmain and I couldn't find it  

thanks

[mikeywith]

Bitmain doesn't publish the recovery files anymore and they want you to email them and request the exact recovery files you need, but you can get it here https://asic.to/download/s17-and-t17-sd-card-image-to-downgrade/ although that file contains more firmware than you need, it has S17+, T17+, S17, and T17, simply keep what you need and delete the rest.

[vr4dude316]

I just received new T17+.  Latest BM firmware.  I am trying to extract the Unlock SD card files from asic.to.  The archive is password protected?  Any suggestions?
https://asic.to/download/s17-and-t17-sd-card-image-to-downgrade/

Also, seeing files for S17 and T17, but which to use for T17+?

[taserz]

I just received new T17+.  Latest BM firmware.  I am trying to extract the Unlock SD card files from asic.to.  The archive is password protected?  Any suggestions?
https://asic.to/download/s17-and-t17-sd-card-image-to-downgrade/

Also, seeing files for S17 and T17, but which to use for T17+?

There is no password on this T17+ sd image?

[vr4dude316]

odd.  when i try and extract zip contents with peazip, it asks for password

so unzipped with 7zip.  no problem.  i do only see the S17+.  does this work for the T17+?

[philipma1957]

[...]

Auto tune setting          asic watts                           wall watts                        discrepency
38                               819 (21.6 w/TH)                 1065 (28.0 w/TH)             246w (27%)
43                               1206 (28.0 w/TH)               1445 (33.6 w/TH)             239w (20%)
48                               1484 (30.9 w/TH)               1710 (35.6 w/TH)             226w (15%)
53                               1752 (33.1 w/TH)               1980 (37.4 w/TH)             228w (13%)
58                               2080 (35.9 w/TH)               2245 (38.7 w/TH)             165w (8%)
62                               2361 (38. w/TH)                 2500 (40.3 w/TH)             139w (6%)
67                               2745 (41.0 w/TH)               2870 (42.8 w/TH)              25w (5%)

For fair comparison on the same  rig:

Standard firmware

mode                      wall watts    
low (40.5)               1755 (43.3 w/TH)
normal (51)             2330 (45.7 w/TH)
high (57)                 2655 (46.5 w/TH)

There is quite a bit of power savings at the wall as you can see using the modded firmware, especially if under clocking.

huge difference. in bold.

It will help with 1/2 ing.

Based on my s17pro numbers I suspect you have a hot meter. You seem to be much higher then my  s17pros stock readings.

I am
 1592 low
 2225 normal
 2555 high

old firmware from April 2019 on the one pro I tested.