Everything went quickly and cleanly - perhaps 5 minutes of downtime.  So far, no issue.

One concern.  Per the website provided:


Yet the stated power consumption is 70J/TH.  The S9 is 0.098J/GH, or 98J/TH.  70 is not "less than half" of 98.

Also, 70J/TH @ 22TH = 1540 Watts, not 1340W as advertised, and thats just for the chips not including power supply losses, fans, controllers, etc.

Also concerned that the same chip number, BM1387, is being used for both the current 16nm chip and this proposed chip.  That is rather unusual to say the least.

This level of "sloppy" rings of a fake ad.

Just for the record, I just upgraded all (11) of mine per Fanatic26's comment without a problem.

Please encourage him to PM me and I'll provide my personal e-mail.  My data center can hold 200 units (power limit) so I suspect we have a few things in common.  Of course, I too will encourage him to come to Kano since I came from Slushpool myself.  <Just for clarity, I am not renting out space>

I second what Fanatic26 said.  Although it was a cool idea, it never did work well.  I experimented with it a LOT back in 2014 and my total hashrate took serious hits when either of the approaches were used.

Yeah, my reply was deleted as well.  My only rationalization was that it was on the BlokForge thread instead of something specific to Avalons - thus my slightly reworked post here.

Presuming you have set privileged API access, I don't see why it wouldn't.  Perhaps just try it vs debate it?

Under Awesome Miner rules:

Add a new rule with a TIME trigger and a "Stop Miner" action would do the trick I would imagine.

Of course, you would want another TIME rule to "Start Miner" later.

Presuming you exhaust the hot air directly into the environment, you only need to be able to cool the intake air to keep it within specifications.

The T9s, like most Bitmain products, require cooling air < 40C (104F).  

How much cooling capacity you require is dependent upon your ambient air temperature (and to a lesser degree, its humidity).

Simply never allow your exhaust air to mix with your intake air.

If your trying to make a closed environment, then every watt used will need to have the associated watt of heat rejected outside via your cooling system.  That it not efficient or recommended.

Will do.

In my mind, this is more of a "recall" than just a change in specs.  My data centers, with over 150 miners in it, is outside air cooled.  In my area, I might have a few afternoons a year, in an exceptionally hot year, where the outside air is > 40C.  I will have MONTHS of afternoon air temps > 30C.  This is a huge deal.

(Full disclosure:  Most of those miners were previously bought S9s with 40C temp ceilings.  With my latest order, there will only be 16 Avalon 8s in there, but they WERE my expansion choice - now I may need to reconsider.)

Agree with that, however many have been purchased with the understanding they were class A3 devices (at least in terms of acceptable maximum cooling air temp).

A3 is really considered the minimum for modern data centers.

As was reported (and deleted) on another thread, it appears Canaan has reduced the intake air specification from 40C to 30C (104F to 86F).

https://canaan.io/wp-content/uploads/2018/03/841-datasheet.pdf

This means the equipment is no longer spec'ed to work within ASHRAE guidelines (the industry standard).

Most new computing equipment is designed to meet the ASHRAE 2015 thermal guidelines of at LEAST class A2 (up to 35C) and most handle class A3 (up to 40C) or class A4 (up to 45C).  30C is not even class A1 (32C).  

So Canaan is now making "industrial" miners that are not designed for modern industrial data centers??!?!

The only way to be in spec now is to either be in an old inefficient data center that still provides the classic 18-27C cooling air or some similarly costly environment (unless your data center's intake air never exceeds 86F - possibly doable in northern regions).

Personally very disappointed with this, since I previously considered Avalons my premium minor (over my S9s).

Well... EXCUSE ME!  We had a member posting about how BCH had risen.  I responded on why, and how its likely just being pumped now, and why other pools deceive people - ending it on a pro Kano note.

If you don't mind, how about leaving the moderation comments to Kano.

Welcome to the world of speculative trading!  It appears Bitcoin Cash is on a bit of a ride due to their May 15th bump to 32MB blocks (and other tweaks).  That comes across to me as 100% headline grabbing only news... their current average blocksize is just 75kb, 1/10th that of Bitcoin.  There is no way they need 32MB blocks right now... its just PC to have them.   If you haven't come across is, I highly recommend looking over http://fork.lol - tonnes of good data there comparing BTC to BCH.  One note on the charted "Profitability": it presumes you could mystically sell block reward coins immediately, at the price each coin had at the moment the block was found - not after the required 100 confirms, nor after your wallet received them, and you sent them to an exchange, or the time it would take you to place an order and have that order fulfilled.  Basically, to really have a useful profitability metric, you would need to be able to predict the price and market demand for each coin at least 100 blocks INTO THE FUTURE.  Then you could decide what was most profitable.  Keep that in mind when considering any profit-switching alternatives to Kano.

As Fanatic26 stated, latency is what matters.  Do a ping to your chosen pool from each time of connection and pick whatever is fastest.  I wouldn't go with anything over 200ms, pings < 100ms should be fine.

My mining facility currently has about 125 S9s in it and is air cooled, similar to the way Yahoo data centers do it.  I recirculate hot aisle air in the winter to keep the cold aisle around 55F.  The rest of the heat, and all of it during warmer seasons, is vented out the top of the roof by natural convection.  Only fans other than the recirculating one are those on the miners, plus two small units at the top of the cold aisle that exhaust any accumulated hot air at the ceiling of that aisle.  e.g. My facility was designed for mining, its not a retrofitted data center.

Water cooling is efficient, in terms of extracting heat, but your still releasing it into the environment.  Any hot water you eject is either going into a ground loop and heating the soil, or into a stream and heating that water with associated environmental impacts (reduced oxygen capacity of the water until it radiates the extra heat to the atmosphere for you).  Hopefully your using a closed ground loop.

So 2500 watts for 20Th = 125 watts/TH right?  A normal 13.5 S9 draws ~1452 watts for about 108 watts/TH.  So running at that speed decreases efficiency about 16%, and requires submerged cooling.  Other than bragging rights and maybe noise control, why would one want to do this?

Somewhere around September 28th, 2017, they stopped accepting BTC.  I only know because I paid with BTC on the 27th of that month, and they had stopped accepting it shortly thereafter.

Sounds like more math.  Let me take a swing at it using a 13.5TH 1452 (at the wall) Watts as a basis.  Of course, everything would scale linearly with that.  Lets calc on a week.

168 hours at $0.1205 = $20.24 for power and yields 13.5TH * 168 hours = 8.1648 EHashes = $2.4789/EHash
143 hours at $0.07 = $10.01 for power and yields 13.5Th * 143 hours = 6.9498 Hashes = $1.44/EHash

That is the cost side.  You need to figure your profit side, subtract the cost, and see which is better.