It's Friday, which I'm told means something to regular people, and the MCP has some very interesting data for me.
I'm loving the value jump--albeit somewhat arbitrary until distribution day. We were at $675.32 last night (keep in mind, this mining batch started 01-Feb-2018) with a price of $144.44. We're still mining right around the 3.7 GH/s average hash rate-- seeing about 0.41 LTC/day. The drag from the 0.5 LTC/day at the same hash rate can be explained by two primary conditions on our end; One, the MCP tends to throw efficiency out of the window when it's finding good temps/speeds initially (usually happens after I've had the box open due to breaking the air-seal and letting much warmer air into the intake--something that will be rectified by the new HDPE air-box); Two, the MCP has been doing a lot of adjusting, which means a lot of restarting.
The miners take a couple of minutes from reboot to hashing, and then a couple more to find their speed. So, a lot of adjustments hurt. As you can see in the screenshot, at about 06:00 EST, the MCP started picking up a lot of hardware failures--I had set all the miners to 500 MHz to let the MCP bring them down to their comfortable clock speeds and temps.
When a chip reports a failure (not just the PCB HW failure, but the X you get in the chip table) it usually means that a chip is individually flawed and/or power draw fluctuations are too much--often the result of the PSU being at capacity. When the MCP detects a chip failure, it immediately checks the length of time since the last failure and if it's been too short of a time, it's likely the clock speed is overtaxing the PSU or the chip's maximum power efficiency. When this condition is detected, the MCP adjusts the clock speed down and restarts the miner.
So, just so the reader knows that the MCP doesn't just drag the clock speeds down, there are several conditions under which the MCP will increase the clock speed. For instance, if the failure history is clean for a certain amount of time, and the temperature drops significantly below the average until they almost meet, the MCP will try to increase the clock speed in increments of 25. This is done because the chips may not be flawed at all; They could simply be underpowered because their thermal-efficiency sucks. If the MCP tries to overclock a miner several times with a resulting chip failure, the MCP will flag the miner and the troubled blade (PCB) for me to look into replacing or swapping later (I'll get into moving the best blades up and the worst blades down in a later post).
In the end, the current mass restarts were caused entirely by me, more or less intentionally--at least, I knew it was going to happen when I set the MCP running after an update.
The intake and exhaust temps have been pretty steady despite the outside temp reaching 1.1°C (34°F), which gives me some relief from the panic that the warmup would drive the temps up significantly. You can see the miners' temps fluctuating there in the early morning from the restarts, but the temps remain beneath 60°C for the most part (remember, this is an average of the chip temps, not the PCB temps). We'll see what happens throughout the day as the temperature outside climbs to 6°C (43°F).
All in all, things look nice for the day. Happy with the temps, happy with the hash rates, and even happy with the market price--although, Binance hasn't fully recovered (I can get periodic, false-hope inspiring data on my movile app, but their site still times out), so there will likely be a big dip coming.