Avalon ASIC users thread

[bvoid]

If you need a pc for other mining devices, just grab a raspbeerry pi: power consumption is only 2-3W   

[1715213032]

1715213032

[1715213032]

1715213032

[Luke-Jr]

In the past, we still need an old cheap PC to control the mining rigs, now with Avalon, we don't even need that PC controller right? ( I have never mined bitcoin before, so I have to ask.)

Rather, it has a router builtin to serve that purpose.

[scrybe]

A picture is worth a thousand words 

...

It would be great if you gave the source of this image.

Reddit with link to Chinese language site: http://www.reddit.com/r/Bitcoin/comments/18qxtl/chinese_have_received_their_avalons/

[DobZombie]

Can someone tell me how many of these are CONFIRMED in the wild yet?  CNY wasn't all month was it?  Didn't shipping happen 32 days ago?

I'm not trolling here, so put your flamethrowers away!

 

[fcmatt]

Can someone tell me how many of these are CONFIRMED in the wild yet?  CNY wasn't all month was it?  Didn't shipping happen 32 days ago?

I'm not trolling here, so put your flamethrowers away!

 

I would say that the speculation most would guess at is that avalon needed batch #2 money in
order to complete batch #1 orders.. and get them shipped by the end of Feb in order to stay
on track.

So one could speculate that they are purposely quiet on this front and maybe outright fibbed
to people so they could keep this rolling. It does not sound very nice to state, "we ran out of
funds and need more money to finish up batch #1.. so please order batch #2 so we do not go
bankrupt".

The trick is.. the batch #1 money may have covered a lot of sunk costs. So the amount of money
needed to cover batch #1 and batch #2 is already in hand. So it is not like they are unable to deliver
what they are promising. It is just their lack of transparency has caused these types of speculations.
Very BFL like in ways when it comes to giving out information to customers who spent money with them.

[cedivad]

I remember bitsynch saying clearly that they went on red when they started shopping, because they underestimated the shipping cost by 200$/unit.

Anyway, we will see soon. My guess is that they will meet their deadlines.

[GenTarkin]

Im curious, does avalon use any 5V or 3.3V power rails of the PSU? ... if not, and depending on the particular PSU design, this would be a case called crossloading which could make a PSU even less efficient than its rated efficiency.
Perhaps finding out the exact model# of this PSU and seeing if any similar series is available at jonnyguru.com PSU tests would be good. He does crossloading tests for inefficiencies and voltages. Just some food for thought.
OTOH if the 5V / 3V rail is used at least some...then it should be workin well within its rating of efficiency.

[Aseras]

It's a PC power supply the 5v and 3v rails can't carry much of a load. The 12v rails are where the power and efficiency is. From what nzhang has said they re using the 12v rails and dropping it to 1.2v with dc to dc conversion. It's more efficient at 220v than 120v.

It was stated that they were using antec 650w? power supply. The 4 pcb version had an enemax revolution 1000w unit.

[GenTarkin]

It's a PC power supply the 5v and 3v rails can't carry much of a load. The 12v rails are where the power and efficiency is. From what nzhang has said they re using the 12v rails and dropping it to 1.2v with dc to dc conversion. It's more efficient at 220v than 120v.

It was stated that they were using antec 650w? power supply. The 4 pcb version had an enemax revolution 1000w unit.

I realize that, but still when crossloading certain PSU's of a certain design type...their effeciency can drop by as much as 5-10%...
crossloading is when u put load on 12v and not on 5/3v .. vice versa

[needbmw]

I realize that, but still when crossloading certain PSU's of a certain design type...their effeciency can drop by as much as 5-10%...
crossloading is when u put load on 12v and not on 5/3v .. vice versa

Modern high-power ATX PSU's usually have several primary 12V high-current rails, 12->5/3.3V secondary buck converters, and separate low-power converter for stand-by voltage.

[cypherdoc]

In the past, we still need an old cheap PC to control the mining rigs, now with Avalon, we don't even need that PC controller right? ( I have never mined bitcoin before, so I have to ask.)

Rather, it has a router builtin to serve that purpose.

so what's the best way to solo mine with it?  can u load bitcoind into open-wrt or do u have to use a pc?

[Hawkix]

For solo mining with ASIC, I am not sure if getwork calls from bitcoind will suffice. You will need some STRATUM enabled pool deamon software.

[Luke-Jr]

In the past, we still need an old cheap PC to control the mining rigs, now with Avalon, we don't even need that PC controller right? ( I have never mined bitcoin before, so I have to ask.)

Rather, it has a router builtin to serve that purpose.

so what's the best way to solo mine with it?  can u load bitcoind into open-wrt or do u have to use a pc?

Avalon/cgminer doesn't work with GBT, so you can't solo mine directly. You'll need to setup a PC with bitcoind + Eloipool.
Once I'm done with BFL SC support, I'll take a longer look at porting BFGMiner to Avalon so solo mining is more practical.

[Aseras]

It won't have the memory to hold the blockchain. You'd have to mine with a pc ( getwork to solo bitcoin is going to kill it, need stratum or gbt )or something like p2pool or with a pool that mines shares/earnings to your address.

[goxed]

http://bbs.btcman.com/forum.php?mod=viewthread&tid=868&page=13#pid10448

[cedivad]

16*5 = 80 chips for 22GH/s => 275MH/chip @ 2.5W.
BFL should do around 6GH/chip @ 13W

Maybe someone already did this math but i still find it interesting.

[SLok]

16*5 = 80 chips for 22GH/s => 275MH/chip @ 2.5W.
BFL should do around 6GH/chip @ 13W

Maybe someone already did this math but i still find it interesting.

Where did you get that from? Last thing I heard it was 72W at the wall for the 60GH/s single, which' 8 chips clock at 7.5GH/s btw, so the whole device uses 1.2 W per GH/s.

[scrybe]

16*5 = 80 chips for 22GH/s => 275MH/chip @ 2.5W.
BFL should do around 6GH/chip @ 13W

Maybe someone already did this math but i still find it interesting.

Fewer devices, smaller feature size, way less power.

Let me just use this 1970's pocket calculator to verify that (THUMP)

[GenTarkin]

The big difference in power probably comes from the smaller process node, 110nm that avalon uses is horribly huge by todays standards =/

[cedivad]

16*5 = 80 chips for 22GH/s => 275MH/chip @ 2.5W.
BFL should do around 6GH/chip @ 13W

Maybe someone already did this math but i still find it interesting.

Where did you get that from? Last thing I heard it was 72W at the wall for the 60GH/s single, which' 8 chips clock at 7.5GH/s btw, so the whole device uses 1.2 W per GH/s.

Sorry... i did the math with what i recalled by mind and i had an error of 20%...

The big difference in power probably comes from the smaller process node, 110nm that avalon uses is horribly huge by todays standards =/

Once that BFL deliveries you will have a point.