Pages:
Author

Topic: Butterfly Labs - Bitforce Single and Mini Rig Box - page 76. (Read 186946 times)

vip
Activity: 166
Merit: 100
I just looked at some air con models available here. 12,000 BTU, split and window type, around 1.05 to 1.2x kWh. I hope that's max and not some sort of average.

Pay close attention to the EER (Energy Efficiency Ratio). For window types, most don't go above 10.8 (divide the BTUs by the ratio to get watts used). The highest I've seen is 11.7 EER (1025 watts)
hero member
Activity: 489
Merit: 500
Immersionist
Thanks for clarifying that about the PSU.

I just looked at some air con models available here. 12,000 BTU, split and window type, around 1.05 to 1.2x kWh. I hope that's max and not some sort of average.


vip
Activity: 166
Merit: 100
Interesting. I'll have to do some more research on all of this.

I also just realized the dimensions of the case they are using for the rig box contains an internal power supply. Try to imagine a 2500Watt power supply. It definitely couldn't fit in there. Even two of those 1500 Watt monsters from Newegg would not be enough (80% efficiency, so only 2400 Watt) and two would physically already be bigger than the "Rig Box".

The ratings on a power supply is the amount of watts it can deliver to the components, so two 1500 watt PSUs would be enough. Regarding heat, I ran 1,500 watts of GPUs and it easily overpowered the room they were in, even with the central A/C that was running around the clock. I agree with  SgtSpike, you will absolutely need A/C to cool down 2,500 watts.
hero member
Activity: 489
Merit: 500
Immersionist
If you use 2500w in a small space, you most certainly will need 1 ton of A/C to keep it cool.  And 1 ton might not be enough come summertime!

But, remember an A/C is a heatpump, so it is more efficient than a resistive heating heat source.  Hard to explain.  But basically, a 1 ton A/C will NOT be anywhere close to 3500 watts.  I have a 10,000 BTU window A/C and it uses 1,020w.  And it's fairly old, so I am sure there's better and even more efficient units out there.  I bet you could get 12,000 BTU of cooling for under 1,000w now.

Interesting. I'll have to do some more research on all of this.

I also just realized the dimensions of the case they are using for the rig box contains an internal power supply. Try to imagine a 2500Watt power supply. It definitely couldn't fit in there. Even two of those 1500 Watt monsters from Newegg would not be enough (80% efficiency, so only 2400 Watt) and two would physically already be bigger than the "Rig Box".
legendary
Activity: 1400
Merit: 1005
hero member
Activity: 489
Merit: 500
Immersionist
I am trying to figure something out about the Rig Box. For people not following all the BFL threads, I had initially paid 20 singles and made a down payment for a Rig Box. But I cancelled my order a week or two ago, until I see more single units out there (3 rev ones, without extra fans and holes in the bottom) and at least one or two Rig Boxes in the wild.

I am sitting here in a friends office who is designing a new server room for their business. I've been joking he could house my mining rigs in there once he is done, which started a lengthy discussion about heat and A/C.

He has been reading this article:

Calculate heat dissipation for better server room design
http://www.techrepublic.com/article/get-it-done-calculate-heat-dissipation-for-better-server-room-design/5031678

Now here is what I am trying to figure out.  BFL says the Rig Box does 50.4 Giga Hash @ 2,500 Watt. They haven't said anything about the "required" or "recommended" temperature, but they said it's 22C with the singles (correct me if I am wrong). Now for the Rig Box I don't expect it to be different, on the contrary, 2500W is a lot of heat for a box with dimensions of 24.5x17.7x13.5cm. Honestly, I believe the case is just a preliminary picture, and they are probably using something completely different, but that's not of any concern in this post. The point is that the heat dissipation is 2,500 Watt. Probably more than that given their track record of getting it wrong.

Since we don't have heat dissipation specifications from BFL, we are going to use a formula provided by the above server room design article.

(2 500 * 0.80 * 1.73) / 0.293 = 11 809 BTU.

An air conditioner rated at 1 ton is roughly capable of handling 12,000 BTU (or 3517 watts!!) per hour of capacity, as specified by the American Society of Mechanical Engineers.

So does that mean you need a 1 ton (12,000 BTU) A/C for a Rig Box and pay an additional 3517 watts of electricity?

I can't imagine you'd need an A/C if you run a couple of GPU mining rigs with a total of 2,500 watt (to stay with the same number) and clearly, people run mining rigs in hotter climates (ie. Goat in Thailand). Why would you need one for servers? Or would it only matter if you have multiples of those (as it is usually the case for server rooms)? As you can see, I am clearly confused about his.

For reference, temperature averages here in my city:

Code:
Jan-Dec, Degree Centigrade
18.6 18.6 21.5 25.1 28.4 30.4 31.3 31.1 30.2 27.7 24.0 20.3
legendary
Activity: 2114
Merit: 1031
any updates on bfl rigbox?
Well, when someone emailed asking if they could visit them with a $30,000 cashier's check they didn't reply.
So I guess it's not ready yet.

Have one on order in the first batch, last I heard they are expecting delivery in May.

And you shelled out 25-30k?  Wow, my wife would not let me do that... heh
full member
Activity: 168
Merit: 100
any updates on bfl rigbox?
Well, when someone emailed asking if they could visit them with a $30,000 cashier's check they didn't reply.
So I guess it's not ready yet.

Have one on order in the first batch, last I heard they are expecting delivery in May.
sr. member
Activity: 349
Merit: 250
You guys are all noobs.

I am waiting for the "SuperComputer"

Nothing else suffices. Grin

Hopefully it will surface before the end of the world Cheesy
Silk Road beware, bulanula's coming and loaded with bitcoin.  Grin
hero member
Activity: 518
Merit: 500
You guys are all noobs.

I am waiting for the "SuperComputer"

Nothing else suffices. Grin

Hopefully it will surface before the end of the world Cheesy
sr. member
Activity: 350
Merit: 250
any updates on bfl rigbox?
Well, when someone emailed asking if they could visit them with a $30,000 cashier's check they didn't reply.
So I guess it's not ready yet.
Well, I really like the Single.  It's interesting that the rig box will use a different chip than the single. I wouldn't mind seeing a roll out of a "rig box"-chip Single. Don't have to layout 30k for a box, but can receive some of the efficiency of the more powerful or efficient chip. To me, it would also serve to shake out any problems with the newer chip prior to shipping out a 30k product. Certainly not required, but why not?

I also would like to hear of more singles in the wild.  Any new reports?

USPS have mine.  Could be a while Wink
sr. member
Activity: 349
Merit: 250
any updates on bfl rigbox?
Well, when someone emailed asking if they could visit them with a $30,000 cashier's check they didn't reply.
So I guess it's not ready yet.
Well, I really like the Single.  It's interesting that the rig box will use a different chip than the single. I wouldn't mind seeing a roll out of a "rig box"-chip Single. Don't have to layout 30k for a box, but can receive some of the efficiency of the more powerful or efficient chip. To me, it would also serve to shake out any problems with the newer chip prior to shipping out a 30k product. Certainly not required, but why not?

I also would like to hear of more singles in the wild.  Any new reports?
legendary
Activity: 4592
Merit: 1851
Linux since 1997 RedHat 4
any updates on bfl rigbox?
Well, when someone emailed asking if they could visit them with a $30,000 cashier's check they didn't reply.
So I guess it's not ready yet.
newbie
Activity: 28
Merit: 0
any updates on bfl rigbox?
full member
Activity: 217
Merit: 100
I was thinking about buying some 7970's instead.  I'm thinking about investing some serious dough and if bitcoins flop then who is going to want the fpga,  at the least I could get something back for he 7970s.  What is everyone's thoughts on this.  

Bitcoin survived a price collapse from $30 to $3, I dont think it is going to flop anytime soon.
sr. member
Activity: 350
Merit: 250

Back on topic, America does some things right, but their postage system isn't counted in that.   USPS is, and always has been, a bloody disgrace Smiley   I'm sure it might work well enough interstate, but internationally it's bloody terrible.

Shouldn't come as a shock to anyone, but felt like venting anyway Cheesy

I disagree. Had many shippments from the USA to Switzerland in the past. Never had an issue with USPS international priority mail ! They ship fast and you can afford it ! UPS or FedEX are the real disgrace. You pay way more and don't get your goods any faster. So i only do USPS and i did more than 20 shippments.

Wow.  I've never had anything other than horrible service from them unfortunately.   People in Australia are just now getting packages that were sent a month ago.   I always go out of my way to avoid USPS wherever humanly possibly, but unfortunately it wasn't possible this time.

UPS and Fedex make it here in 3 days 90% of the time in my experience.   USPS quote 3-5 days for Express and it's usually a couple of weeks (the good news being that they're daft enough to offer a money back guarantee if it's not delivered in those timeframes.  It never ever is, so it usually works out to free postage.  I'd still rather pay and have it quicker though).
sr. member
Activity: 349
Merit: 250
Butterfly Labs - can you tell us how many backorders you have at this point?

I see so many people saying they have ordered 1 or 2 (or 10).  If we assume 300 backordered singles, and say, 10 of the box rigs, that's around 750 GH/s of added hashing power!  Increase of difficulty around 7.5% from where we are now...  I guess it's not a huge amount, but still interesting to think about!
The difficulty is going up by around 15% a month anyway. Even putting all bitcoins earned back into new singles, it will not be enough to maintain a set income. Unless bitcoin prices go up or the difficulty stops going up at the current rate.
So ... where did you get that 15% number from?
The current jump 24 blocks ago was approx 9%
Each adjustment happens about every two weeks. Last time it was 9%. I think 15% a month is about right, I don't see what you think is wrong with it.
The difficulty on 14 October 2011 was 1468195.43, the current difficulty is 1496978.59.  Your estimate of 15% a month is absent of fact.  Maybe that is what kano finds wrong.
After the price crash when the bottom price was uncertain, the difficulty dropped with it. Confidence in the price did not start until December. Estimating future difficulty off a time period with a price crash only makes sense if you expect the price to crash again. The difficulty has gone up steady this year, I see no reason for it to stop.
Don't let the facts get in the way of a good opinion, huh?  Difficulty will follow price. The reason for it to stop is price. That said, I think price will head up, but the 15% a month is bubble talk. No bubble, no 15%. If we have a bubble all bets are off, 15% will be conservative.
Code:
1468195.43		10/14/2011 02:44																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																																													
1203461.93 -18.03% 10/31/2011 03:42
1192497.75 -0.91% 11/14/2011 06:56
1090715.68 -8.54% 11/29/2011 14:20
1155038.33 5.90% 12/12/2011 19:42
1159929.50 0.42% 12/26/2011 18:43
1250757.74 7.83% 01/08/2012 18:26
1307728.36 4.55% 01/22/2012 03:55
1379647.44 5.50% 02/04/2012 10:32
1376302.27 -0.24% 02/18/2012 11:24
1496978.60 8.77% 03/02/2012 08:25
sr. member
Activity: 392
Merit: 250
Butterfly Labs - can you tell us how many backorders you have at this point?

I see so many people saying they have ordered 1 or 2 (or 10).  If we assume 300 backordered singles, and say, 10 of the box rigs, that's around 750 GH/s of added hashing power!  Increase of difficulty around 7.5% from where we are now...  I guess it's not a huge amount, but still interesting to think about!

The difficulty is going up by around 15% a month anyway. Even putting all bitcoins earned back into new singles, it will not be enough to maintain a set income. Unless bitcoin prices go up or the difficulty stops going up at the current rate.
So ... where did you get that 15% number from?
The current jump 24 blocks ago was approx 9%
Each adjustment happens about every two weeks. Last time it was 9%. I think 15% a month is about right, I don't see what you think is wrong with it.
The difficulty on 14 October 2011 was 1468195.43, the current difficulty is 1496978.59.  Your estimate of 15% a month is absent of fact.  Maybe that is what kano finds wrong.
After the price crash when the bottom price was uncertain, the difficulty dropped with it. Confidence in the price did not start until December. Estimating future difficulty off a time period with a price crash only makes sense if you expect the price to crash again. The difficulty has gone up steady this year, I see no reason for it to stop.
legendary
Activity: 1022
Merit: 1000
BitMinter

Back on topic, America does some things right, but their postage system isn't counted in that.   USPS is, and always has been, a bloody disgrace Smiley   I'm sure it might work well enough interstate, but internationally it's bloody terrible.

Shouldn't come as a shock to anyone, but felt like venting anyway Cheesy

I disagree. Had many shippments from the USA to Switzerland in the past. Never had an issue with USPS international priority mail ! They ship fast and you can afford it ! UPS or FedEX are the real disgrace. You pay way more and don't get your goods any faster. So i only do USPS and i did more than 20 shippments.
legendary
Activity: 4592
Merit: 1851
Linux since 1997 RedHat 4
Hmm ... anyone know if current can be updated if BFL come out with a new firmware?
(I'm wondering if the reason he's selling them is the current firmware and p2pool do not go together well)

Any FPGA can be upgraded with any compatible bitstream.  Icarus and ztek for example have issued dozens of bitstream updates.   

However BFL indicated the JTAG header wouldn't be useful to determine the chip being used which shows they are keeping (or at least trying) to keep the devices a black box where end user has no internal access.  So it is more a question of if they allow the bitstream to be upgraded.
Yep - that's exactly what I'm asking Smiley
(no idea if my wording was right or wrong Tongue)

The BFL <-> P2Pool issue is reasonably straight forward (and I'm pretty sure they know also?)
When you abort a work request to the BFL it returns nothing (same as Icarus - no idea about the others)

However, if the BFL had actually found any shares, it won't tell you before you abort it - it tells you when it completes the work.
(Icarus does the opposite which actually seems bad also, but in reality is fine - it stops processing when it finds a share and returns it)

Thus for a BFL, if an LP comes before the work completes, even if the work found shares before the LP, you don't know of them.
Not a big % when you are mining 10 minute LPs - but with P2Pool, 10 second LPs means you lose a lot of work ...
Pages:
Jump to: