 Show Posts
|
|
Pages: « 1 [2]
|
Hello,
Are you talking about chip or board temperatures?
Chip temperatures. I believe the board temperatures are accurate. I don't know for sure what the cause is, but I believe it has to do with the calibration of the external sense input of the TMP451 used on these boards. The external sense input is used to read the temperature of a sensor integrated into the ASIC, and needs correct calibration settings to get accurate results. The internal sensor on the TMP451 that is used to get the "board" temperature is calibrated at the factory so it doesn't need any additional setup by the firmware. What mikeywith mentioned is also true though. You may need different flow rates because of the different power density. A s9 running at 1400W is using only ~7.5W per chip. A T17 at 1900W is using ~21W/chip. I'm doing 40T on my T17 with temps in mid 80s, since this reading is not accurate on Braiins, do you recommend I rise the dangerous temperature (92C) to avoid auto-shutdown? it's getting there. |
|
|
|
Hello,
Are you talking about chip or board temperatures?
Chip temperatures. I believe the board temperatures are accurate. I don't know for sure what the cause is, but I believe it has to do with the calibration of the external sense input of the TMP451 used on these boards. The external sense input is used to read the temperature of a sensor integrated into the ASIC, and needs correct calibration settings to get accurate results. The internal sensor on the TMP451 that is used to get the "board" temperature is calibrated at the factory so it doesn't need any additional setup by the firmware. What mikeywith mentioned is also true though. You may need different flow rates because of the different power density. A s9 running at 1400W is using only ~7.5W per chip. A T17 at 1900W is using ~21W/chip. Makes alot of sense. What Mikey said is also concerning to me. Before I even get a T17, S9 running with temp 65C is alot and is not what an immersion system gives. You said 80C is right on the target considering Braiins exaggeration, but the gear is immersed, I should not be on target but much lower. I'll look through my system and see what I can improve, I know for sure that everything I use is high-material, could be flow-related |
|
|
|
I have noticed in the past when comparing Brains, Vnish, and stock firmware all running fixed frequency and identical conditions that Brains reports temperatures close to 10degC higher than stock or Vnish on the 17 series miners. Also, stock firmware on the T17 has a temperature target of 80 deg C, so if Brains is telling you 90, you are probably pretty much right on target.
Hello, Are you talking about chip or board temperatures? |
|
|
|
I don't understand how a 1.4kw S9 is 65C and a 500W increase in a T17 shoots chip temperature to almost 90C!
It's not just the extra 500w, the number of chips on both gears are different, the chips themselves are different, and the way they attach the heatsinks on each of these models is totally different, you can only make this comparison between two S9s running at different hashrate. But to be honest with you, I think there is something wrong with your setup, I mean even the 68C on S9 seems too high for immersion cooling, maybe it's the 30c temp outside that is causing this, you are going to need a bigger radiator and a larger pump. Also, keep in mind that the T17 runs a bit hot by default, also the heatsinks are "loosely" attached and depending on what type of liquid you are using, the heatsinks might just come off at any second, with that in mind, 90C isn't the end of the world, your board temps show 51c, the max is 75c, this gear will die soon anyway, so I wouldn't even bother, just mine on. Hi Mike, Thanks for your valuable input. I too wasn't happy with S9 readings. I invested alot in this setup. The radiator is a 24x24" wood stove heat exchanger that many miners use to operate two S19s (rated for 6,000W), hooked up to a 3,000 CFM ventilation fan.I installed the radiator and the fan to an A/C hole to push hot air out and keep the room cool. The pump is also a March MDX that I've got after miners have recommended it. The tank is designed for S1X/T1X models with two chambers and a flowplate. This T17 is the upgraded version (reinforced), I got it from China for $1,000. They have sent me a confirmation video showing the machine running and it was giving 42T with max temp 80C on air. That's what's triggering me. It worked fine on Air but not on immersion ( I couldn't even get it to 35T by immersion) I don't know what's wrong with my setup, I've put serious money into this setup and spent 6 months building simple setups before I got to this stage, and unfortunately, the result is not exciting. Here's my setup https://imgur.com/a/LFZIk8E, if you're interested to give your two cents. |
|
|
|
Hello, I could not wrap my head around how weird this is. I have an immersion setup sized for 6,000Watts. I got an S9 to test things out and was able to hash at 16T drawing 1.4kW with chip temperature 65-68C. Never exceeding 70C. I felt confident enough with my setup to get a T17 42t, with Braiins installed, it's currently sitting at 1.9kw with chip temperature in late 80C. I don't understand how a 1.4kw S9 is 65C and a 500W increase in a T17 shoots chip temperature to almost 90C! Ambient temp is 30C, fluid inlet temp is 45C. Here is an image showing both settings for S9 and T17:  Here's my setup https://imgur.com/a/LFZIk8ECould it be something about T17 sensors? My setup is oversized for this kind of power consumption! |
|
|
|
New Antminer S17 Pro arrived today and plugged it in. It boots up, I can access the web UI, I set my pool, and I set it to "Low Power" mode for testing. After hitting apply, the fans spin up -- similar to the S9 -- and I'm expecting to start hashing about 30 seconds later. Instead, after about 60-70 seconds, the red "Fault" light on the machine starts blinking. Fans keep spinning, web UI still works fine, but it's not hashing. I've verified that it has an internet connection using the network diagnostics tool in the Antminer Web UI. And the pool isn't down because I have other S9s hashing on it right now. If I put it in "sleep mode", the fault light turns off -- although it still isn't hashing. Trying to set it to Low or Normal power mode causes the blinking again about 60-70 seconds. I captured the entire kernel log here: https://gist.github.com/tjwebb/8f75822baf28f346110ffca359790364. I don't see any obvious failures there, but any insight is appreciated. Latest bitcoin Antminers are bad quality. Ill have same problems , kernel log shows it has initialized fine but doesnt mine and then reboots after while. It is most likely one of your hashing boards died. At least i disconected one and it was working fine after. Sick of these asic miners, installed 5 of them in my mining farm 3 of them have issues after 2 weeks. Hi - What issues are you facing exactlly? |
|
|
|
Sorry, you're living in the GPU scamcoin world with the statement that current miners are force down-clocked by miner producers.
If you've been around mining for 10 years like I have, and been involved in producing modified firmware, and in my case I was possibly the first to do it, you'd know that the miners are not made to be run slower to make people buy them cheaper.
Also, there are no 3rd party firmware versions for current miners, only previous ones.
As for 3rd party firmware making them run faster, yeah like you can make anything run faster, with extra cooling and shorter lifespan.
As if the manufacturers want to handle warranty on miners that the buyers have destroyed by over clocking them ...
Bitcoin is about producing money, so people do all sorts of stupid things trying to get more money and usually end up getting less.
Anyway, your original statement was about making things more open and freely available ... to run closed source, fee driven firmware, that violates licenses.
That seems to be an obvious oxymoron ...
Meanwhile, this week I released open source code to use the S17 chip (that someone else worked out the commands) ... oh well ...
10 years! that's impressive. I want your take on immersion cooling? I've just built an immersion cooling system that I'm pretty happy with and looking to buy an S9 to miss around with. Running a miner on immersion cooling easily matches the stock hash rate at lower consumption. Wouldn't you be intrigued to custom firmware it and overclock it? I think it's a no-brainer only if you're into immersion cooling. |
|
|
|
|
Hello,
I know I can export my key through the GU suite, but can someone please breakdown this command for me:
gpg --armor --export KEY_ID | base64 -w0
the w0 dosesn't get executed, what does it mean?
Thanks!
|
|
|
|
Hello, hope this is the right place for this kind of post. So, There is this project (altcoin) called The Origin Protocol, their vision is a decentralized peer-to-peer eCommerce platforms, to compete with shopify. This product is called dshop ( https://www.originprotocol.com/en/dshop). They claim to achieve this by combining IPFS ,to pin the site's data, and Ethereum to store those IPFS hashes and send them over the blockchain to process orders/receive customer's informations. All customers' information are encrypted with GBG and can only be decrypted by the dshop owner to view the details and send the goods! Share your thoughts! Has anyone tried deploying their own dshop? tutorial: https://github.com/OriginProtocol/dshop |
|
|
|
|
