Still lots of profitable gear out there. S17pros on LPM, or some S17s with aftermarket firmware, are pretty close to S19pro in terms of profitability per TH. M30S is still doing OK too. Texas power shortage is going to be just a temporary blip, that'll recover quickly. It does point out that the risk of mining in Texas is growing as they shut down reliable power sources in favor of wind and solar. I wonder how much $ those farms in Texas are getting for shutting down. I've heard of farms getting paid the current rate to take power offline. When things get close to overload that rate can be really high. Farms shutting down because of power price hikes will still just be temporary blips... that gear is just going to wind up running at a different site. I think price would need dip below 13k before we'd see a real sustained dump in diff.  Gear prices are still at 24 month ROI levels. I haven't noticed much movement in price in the last week or so, guess someone must be buying.  |
|
|
|
|
Well sounds like immersion cooling could solve all the issues. Just need to oversize your dry cooler enough to get the heat out. If the psu isn't in air, there would be no issue with arcs vs altitude.
But where would you find a location >2000m elevation with cheap abundant power?
|
|
|
|
I think you are completely correct when you say there is risk in not actually owning your mining hardware. But I also think that same principle of risk goes for everything else which you do not completely own. Just look at what happens when you leave your Bitcoin on the wrong exchange and they close up shop and never give you your Bitcoin back. Furthermore when it comes to mining, you are depending on the people who sold you the miners not to have included some kind of malware or spyware in the miners. And that the people who built the miners did not use defective chip parts and so on...
There is risk for everything, the thing to be concerned with is how much risk. And the risk of getting screwed by cloud mining is way way way larger than ending up with some flaky gear or getting stuck with hacked firmware. Also, the risk from your two examples can be nearly eliminated with a few precautions. If you buy used gear, reflash all of it with the latest firmware on an isolated network before putting it online. As for defective chips etc., .... just don't buy Bitmain. Problem solved  . |
|
|
|
As people have mentioned earlier, farms that had future orders in before the price crash are probably still installing that hardware when it shows up. So that would put several months of delay before we'd see any major slowdown. I think it's more likely to see gear prices crash than diff go down. As more miners realize that it's probably going to take a long time to get back to all-time-highs, gear manufacturers are going to have a harder time finding miners that will pay > 2-year ROI prices. I think the best we can hope for as far as diff going down is a few modest and temporary drops as gear gets shuffled around from farms that go under from spiking power costs to sites that are still profitable. Here's a table that shows the $/th you'd need to pay for gear to ROI in 12 months for different power costs and miner efficiencies at $19k and current diff. The best price I'm seeing for a new S19pro is ~$40/th.  |
|
|
|
The voltage setting stored in the eeproms (for the current mode if a s17) needs to be the same on all hashboards in a miner for stock firmware to work. If you have logs from the miners with the boards you want to mix, you can tell in advance if they will work together. You just need to find the following section in the logs and check that the voltage setting is the same in the two miners you want to mix and match boards from/to. 022-05-17 22:44:27 uart.c:71:set_baud: set fpga_baud to 115200
2022-05-17 22:44:38 driver-btm-api.c:915:check_asic_number_with_power_on: Chain[0]: find 48 asic, times 0
2022-05-17 22:44:49 driver-btm-api.c:915:check_asic_number_with_power_on: Chain[1]: find 48 asic, times 0
2022-05-17 22:45:00 driver-btm-api.c:915:check_asic_number_with_power_on: Chain[2]: find 48 asic, times 0
2022-05-17 22:45:05 driver-btm-api.c:345:set_order_clock: chain[0]: set order clock, stragegy 3
2022-05-17 22:45:05 driver-btm-api.c:345:set_order_clock: chain[1]: set order clock, stragegy 3
2022-05-17 22:45:05 driver-btm-api.c:345:set_order_clock: chain[2]: set order clock, stragegy 3
2022-05-17 22:45:05 driver-hash-chip.c:502:set_clock_delay_control: core_data = 0x34
2022-05-17 22:45:05 uart.c:71:set_baud: set fpga_baud to 3000000
2022-05-17 22:45:05 driver-btm-api.c:1574:check_clock_counter: freq 50 clock_counter_limit 6
2022-05-17 22:45:07 uart.c:71:set_baud: set fpga_baud to 115200
2022-05-17 22:45:07 voltage[0] = 1780
2022-05-17 22:45:07 voltage[1] = 1780
2022-05-17 22:45:07 voltage[2] = 1780
2022-05-17 22:45:07 power_api.c:219:set_working_voltage_raw: working_voltage_raw = 1780
|
|
|
|
If a hashboard errors-out for any reason during the tuning cycle, I think the Vnish firmware restarts the tuning process from the beginning. So if you have a board that occasionally has issues, it can end up tuning forever. Your log shows some temperature sensor errors on chain 0, so my guess is that board is your problem. You might be better off trying to manually tune it from the "chip frequencies" tab. And the log mikeywith asked for is different than the one you posted. The tuning log is not the "miner log", it is in the spot he indicated in his reply: I can't tell for sure without seeing the tuning log, you can find them in Miner configuration > mining profiles > at the bottom of the page, copy and paste them here, please use the code tag when doing so, without it will be messy and hard to read.
|
|
|
|
|
Ha... at your own risk! I'm only giving you the info I observed in the past, and I didn't look into it further to get any definitive answers. I'd suggest doing your own experiment to verify what I saw. You can run stock firmware immersed if you have some fan spoofers. If you run stock in the same setup and the temps come out to be 10degC lower, then make your own conclusions...
The chips were designed to be run hot, the temperature delta to ambient is needed to move all that heat with air cooling. So you can actually run into some problems if you run too cold, timing on the ASIC design is very complicated and is temperature-dependent. So if they designed the chip for say 70degC to 90degC range, it may not meet timing requirements at 65degC for a given frequency and voltage.
I'm sure there's a sweet spot that would be below 80degC, but you'd need to experiment on that to find out. My point about the 80deg target on factory firmware is only that Bitmain seems to think that their chips are happy at that temperature.
|
|
|
|
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 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.
|
|
|
|
2022-05-16 17:18:30 voltage[0] = 1940
2022-05-16 17:18:30 voltage[1] = 1970
2022-05-16 17:18:30 voltage[2] = 1960
2022-05-16 17:18:30 Voltage are different. Will exit.
Stock Bitmain firmware stores the chip tuning information along with the voltage it was tuned to run at in an eeprom on each hashboard. If the voltages are different between the boards, the firmware will not run. You can either buy a tool that will let you program the eeprom, or run aftermarket firmware like Brains or Vnish, both of which ignore all the data stored in the hashboard. |
|
|
|
|
Just keep in mind that the eeprom holds tuning information (chip frequencies & voltage) that was set at the factory, so although copying the data from the working eeprom may enable the board to work, it may end up operating well below optimal performance.
|
|
|
|
Looing like >4% increase in difficulty along with a ~9% drop in price. People keep saying this trend can't possibly continue, but then it just keeps going in the same direction... Profitability is still relatively high though, it's just the gear prices that are out of control. Might just need some actual low profitability time to clean house... maybe a doubling of difficulty at current price or halving in price at current difficulty would get us there. 
I wonder what's the average price at which those companies actually lose money on operations alone.
For the average S19 running at 5 cents per kWh electricity is 25% of the revenue right now, I wonder how much are the other costs, maintenance, security, electricity cost for cooling, and so on assuming a 1k or 5k miner operation.
I don't think costs other than power make up a significant percentage of the total for even relatively small farms... For a 5MW farm paying $0.05/kwh, that's $182,500 per month for a power bill. That's ~1600 3kw miners that would fit in 5 or 6 shipping containers. They could have several full-time employees and not even hit 5% of the power bill. |
|
|
|
If you have the pickit that is needed to flash the firmware, then you can also read the firmware out of one of your other boards. No need to buy it.  |
|
|
|
|
While it could technically be possible to run a board that has a bad chip, I don't think there are any firmwares out that will do it. Also, the way the communication works between chips, the bad chip would have to fail in a very specific way to make it possible. If the bad chips pulls the RO signal low, then it will prevent all other chips from responding to commands.
|
|
|
|
|
Sorry, but a single bad chip on these hashboards will prevent the whole board from working. Assuming that you don't have experience repairing these, your best option is to try to find a shop that can replace the chips for you.
You can run the other two hashboards just fine without this one plugged in though.
|
|
|
|
|
It looks like the chips have been damaged, part of the chip is still attached to the heatsink on both. So the chips will need to be replaced, just attaching the heatsink is not going to work.
|
|
|
|
2022-03-14 09:30:55 power_api.c:71:check_voltage: target_vol = 21.00, actural_vol = 21.17, check voltage passed.
2022-03-14 09:30:55 uart.c:72:set_baud: set fpga_baud to 115200
2022-03-14 09:31:06 driver-btm-api.c:1096:check_asic_number_with_power_on: Chain[1]: find 0 asic, times 0
2022-03-14 09:31:17 driver-btm-api.c:1096:check_asic_number_with_power_on: Chain[1]: find 0 asic, times 1
2022-03-14 09:31:27 driver-btm-api.c:1096:check_asic_number_with_power_on: Chain[1]: find 0 asic, times 2
...
2022-03-14 09:31:36 driver-btm-api.c:1119:check_asic_number_with_power_on: Chain[1]: find 0 asic, times 3
...
2022-03-14 09:31:51 driver-btm-api.c:1119:check_asic_number_with_power_on: Chain[1]: find 0 asic, times 4
...
2022-03-14 09:32:06 driver-btm-api.c:1119:check_asic_number_with_power_on: Chain[1]: find 0 asic, times 5
...
2022-03-14 09:32:15 driver-btm-api.c:1155:check_asic_number: Chain 1 only find 0 asic, will power off hash board 1 It looks like chain[1] has a problem and needs repair. Also, chain[0] is not detected at all, is its data cable plugged in? |
|
|
|
|
I think you can buy direct if you want 50 or more, otherwise, you'll need to find a reseller. I got mine from Kaboomracks, $325 each including domestic shipping in the USA.
|
|
|
|
|
Show Posts
