I have a question/idea about the Avalon's structure.
Someone said it would be composed of several chips (from 15 to 30 if my memory is correct)
What would happen if one chip drops dead ? It would be nice if in such case we only lose 1/30 of the hashpower; we could even remove the bad chip if it's presence gets in the way of important things, or physically cut lanes meant for, and then restart the board with the rest of the asics.
It's just that 30 chips on a single board increase the failure rate/probability enough to bother me.
Is that a bad idea ?
Topic: "Avalon" ASIC, announcement & pre-order. pre-order over. project started. - page 30. (Read 98445 times)
September 18, 2012, 02:12:18 PM
September 18, 2012, 01:34:27 PM
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
Two sides of the bitcoin :
- You need to deliver fast to be among the first to mine with ASICs
- ASICs will at the end place the profitability of mining to the lowest possible value, and within a year or something only the low consumption systems will be profitable.
These two concepts being contradictory, you'll have to find a compromise
September 18, 2012, 01:18:29 PM
LOL I just realized something. He's saying if he doesn't get 300 units ordered, then the entire batch gets canceled. Well 300 orders @ 60GH/s each is 18TH/s, all entering the market at the same time ~ Feb 2013. 
September 18, 2012, 01:00:03 PM
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
I might be alone in this, but shipping times and keeping the cost down is more important to me than power savings. That's just my opinion, though. September 18, 2012, 11:21:24 AM
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
I might be alone in this, but shipping times and keeping the cost down is more important to me than power savings. That's just my opinion, though. It really comes down to how much power is saved. If we're talking 25 watts for two months extra time, probably not worth it. There are always options around the power issue including hosting your farm somewhere else where power is cheaper.
September 18, 2012, 11:13:10 AM
+1 
September 18, 2012, 11:12:27 AM
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
I might be alone in this, but shipping times and keeping the cost down is more important to me than power savings. That's just my opinion, though. +1
September 18, 2012, 11:02:04 AM
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
I might be alone in this, but shipping times and keeping the cost down is more important to me than power savings. That's just my opinion, though. +1
September 18, 2012, 10:58:59 AM
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
I might be alone in this, but shipping times and keeping the cost down is more important to me than power savings. That's just my opinion, though. 
September 18, 2012, 04:46:02 AM
if the energy consumption optimizations take a lot of design time then it is a risk.
I bet on fast delivery at least for the first ASIC batch.
I bet on fast delivery at least for the first ASIC batch.
First batch ?
Problem being that the second batch HAS to be EXACTLY the same than the first one if you want cheap chips.
If you want to have a second batch with improved ASICS, you have to go through the whole design process again and pay all the money it costs again.
I think he probably means the second production run. Not a chip batch.
September 18, 2012, 03:58:16 AM
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
Two sides of the bitcoin :
- You need to deliver fast to be among the first to mine with ASICs
- ASICs will at the end place the profitability of mining to the lowest possible value, and within a year or something only the low consumption systems will be profitable.
These two concepts being contradictory, you'll have to find a compromise
I've been repeatedly surprised at the speed of developments regarding FPGAs & ASICs.
With so much competition - it seems to me that 'within a year or something' may be a short time indeed, and so low power consumption should be a pretty high priority.
September 18, 2012, 03:22:42 AM
if the energy consumption optimizations take a lot of design time then it is a risk.
I bet on fast delivery at least for the first ASIC batch.
I bet on fast delivery at least for the first ASIC batch.
First batch ?
Problem being that the second batch HAS to be EXACTLY the same than the first one if you want cheap chips.
If you want to have a second batch with improved ASICS, you have to go through the whole design process again and pay all the money it costs again.
September 18, 2012, 02:46:49 AM
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
Two sides of the bitcoin :
- You need to deliver fast to be among the first to mine with ASICs
- ASICs will at the end place the profitability of mining to the lowest possible value, and within a year or something only the low consumption systems will be profitable.
These two concepts being contradictory, you'll have to find a compromise
yes, you got the point.
if the energy consumption optimizations take a lot of design time then it is a risk.
I bet on fast delivery at least for the first ASIC batch.
I bet on fast delivery at least for the first ASIC batch.
yes, timing is the most important part.
September 18, 2012, 02:44:35 AM
if the energy consumption optimizations take a lot of design time then it is a risk.
I bet on fast delivery at least for the first ASIC batch.
I bet on fast delivery at least for the first ASIC batch.
September 18, 2012, 02:29:42 AM
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
Two sides of the bitcoin :
- You need to deliver fast to be among the first to mine with ASICs
- ASICs will at the end place the profitability of mining to the lowest possible value, and within a year or something only the low consumption systems will be profitable.
These two concepts being contradictory, you'll have to find a compromise
September 18, 2012, 02:09:40 AM
is there a power budget cap for pre-order deisgns?
I ask this because electricity costs are a major factor in Australia...currently I pay AU$0.22kW/hr (about US$0.25kW/hr) so the GH/s : kW/hr is my main assessment tool
I ask this because electricity costs are a major factor in Australia...currently I pay AU$0.22kW/hr (about US$0.25kW/hr) so the GH/s : kW/hr is my main assessment tool
if use front end simulation result, the power consume is only about 120~160watt for a 60G rig (just like BFL's pronounce), but....
by our engineering experience, it MAY higher than this value. but i think it will be in a "hundreds of watt" range.
or:
if you guys all consider that, the MH/W is a very important value, we could pay more effort on energy consumption optimize.
September 18, 2012, 01:59:20 AM
is there a power budget cap for pre-order deisgns?
I ask this because electricity costs are a major factor in Australia...currently I pay AU$0.22kW/hr (about US$0.25kW/hr) so the GH/s : kW/hr is my main assessment tool
I ask this because electricity costs are a major factor in Australia...currently I pay AU$0.22kW/hr (about US$0.25kW/hr) so the GH/s : kW/hr is my main assessment tool
September 18, 2012, 12:29:06 AM
Count me in for one or two, If only for the sake of decentralization.
Edit : If this thing is to consume 600w, I think we can save almost 80$ for PSU + shipping. I'd guess, like many miners, I won't have much use for these 80+ Gold PSUs
Edit : If this thing is to consume 600w, I think we can save almost 80$ for PSU + shipping. I'd guess, like many miners, I won't have much use for these 80+ Gold PSUs
I really think that 600W number was a wild guess. It will be a lot less.
I trust ngzhang to make good technical choices.
The BFL singles plug into a computer. These Avalon devices are stand alone. A computer is at least 200W probably, right? My question is how does BFL wattage + computer wattage compare with just the Avalon standalone wattage?
No computer consume as little as 30watt, think netbook or HTPC. BFL merely use an usb port and you can add as many as you want with USB hubs. (128)
the mk802 while not sure if you want to consider it a computer...but it does have linux builds that run on it.
Net books dont use very much power, the adapters are about 40 watts...take away the lcd and the hard drive and it can definitely go below 30 watts.
September 17, 2012, 11:24:18 PM
host PC NOT required
---so there's a controller embedded? it would be upgradeable, right? I hope the final product will support stratum mining protocol.
http://mining.bitcoin.cz/stratum-mining
it's a linux embedded system. which is running cgminer or something like that.---so there's a controller embedded? it would be upgradeable, right? I hope the final product will support stratum mining protocol.
http://mining.bitcoin.cz/stratum-mining
September 17, 2012, 08:29:24 PM
Good to see nzhang in there.
Jump to: