Pages:
Author

Topic: 1GH/s, 20w, $700 (was $500) — Butterflylabs, is it for real? (Part 2) - page 55. (Read 146936 times)

sr. member
Activity: 475
Merit: 265
Ooh La La, C'est Zoom!
The other "interesting" thing is nobody makes 2500W powersupplies (due to lack of need and lack of access).  
...
So maybe the device can only work @ 240V using a pair of 1500W PSU?

Getting a 2500W power supply is not hard, it's just not as easy as driving to your favorite PC hobby supplier to buy a couple. This is because they are more of an "industrial" solution than a "personal" solution.

I am not familiar with the US National Electrical Code and applications of it in households, but I imagine that you could easily install a dedicated circuit in the electrical panel that feeds a NEMA L6-20R (receptacle) and then get a cord that converts from L6-20P (plug) to the the appropriate receptacle on the power supply, likely an IEC C19. Standard PC power supplies use IEC C13 (on cord)/C14 (on power supply).

The compute chassis' that I use in my day job have four 2500W power supplies, one of which is a hot spare. I don't have one to look at right now to check voltage range, but I know we feed each one from one leg of three-phase 208v coming off our UPS, and I know that we have fed them off 240v single phase circuits. They use an IEC C20 socket.

I just looked at some IBM x3650M2 (single quad-core Xeon [email protected]) servers I have and the power supplies (two) are 675w and can be fed using anything from 100v to 240v. They appear to be made by Emerson and have an IEC C14 socket on them.

- Zed


And what is your day job, honey?


Not that it's any of your business, but it's the one that pays the bills, including the mortgage, electricity, ramenfood, cars, RV, etc., and allows me to play with expensive toys.  Grin

- Zed
full member
Activity: 168
Merit: 100
The other "interesting" thing is nobody makes 2500W powersupplies (due to lack of need and lack of access).  
...
So maybe the device can only work @ 240V using a pair of 1500W PSU?

Getting a 2500W power supply is not hard, it's just not as easy as driving to your favorite PC hobby supplier to buy a couple. This is because they are more of an "industrial" solution than a "personal" solution.

I am not familiar with the US National Electrical Code and applications of it in households, but I imagine that you could easily install a dedicated circuit in the electrical panel that feeds a NEMA L6-20R (receptacle) and then get a cord that converts from L6-20P (plug) to the the appropriate receptacle on the power supply, likely an IEC C19. Standard PC power supplies use IEC C13 (on cord)/C14 (on power supply).

The compute chassis' that I use in my day job have four 2500W power supplies, one of which is a hot spare. I don't have one to look at right now to check voltage range, but I know we feed each one from one leg of three-phase 208v coming off our UPS, and I know that we have fed them off 240v single phase circuits. They use an IEC C20 socket.

I just looked at some IBM x3650M2 (single quad-core Xeon [email protected]) servers I have and the power supplies (two) are 675w and can be fed using anything from 100v to 240v. They appear to be made by Emerson and have an IEC C14 socket on them.

- Zed


And what is your day job, honey?
sr. member
Activity: 475
Merit: 265
Ooh La La, C'est Zoom!
Didn't mean to say no 2000W+ PSU exists but rather as you start to get into niche industrial gear costs goes way up.  Those 2500W PSU you are using in your workplace chassis likely cost 4x (maybe more) of a ATX 1200W PSU.

For sure. Volume and quality are amazing things. Reduce one and increase the other, and there you go. Also corporations are not so worried about 4x cost of a "minor" component. They'll just pass the cost on to the consumer of their product(s).

On US National Electrical Code...
Sure installing a dedicated circuit is fine.  I am considering putting in dedicated 240V branch in my garage.

I'm a car guy, so that circuit is for a welder, and another for the compressor, and a bigger circuit/sub-panel for the standby generator (45kW, 240v 3-phase)  Roll Eyes .

Wife is an artist so her circuit is for a kiln.

Good thing FPGAs are coming along. Can you imagine the heat from a GPU hash farm and the Kiln? Yikes!

Thinking 10 gauge NMC and L6-30R locking outlet.  That should be good for 250*30*0.8 = 6KW (not sure if 80% derate is necessary for dedicated 30A circuits?

Probably not, but safe is good.

So I guess if the magical rig box unicorn does exist someday I could run 2 rig boxes of the 30amp circuit. Probably need better alarm system and doors if I am going to have 100GH farm in my garage.

I'd go with a dual use system. It's called a Dog.  Grin  But you're right, probably going to want to do something. Pity the poor thief who grabs the shiny boxes only to discover his fence looks at him like he's from another planet when trying to unload them.

Still I was just pointing out that wattage >1400W guarantees 208V 3phase (non-existent outside of data centers and industrial buildings) or 240V connection.  Without modification homes in the US simply aren't wired for it.  Now international it is a non-issue.

Very true.

That's the cost of entry for what most would consider a hobby.  Wink

- Zed
donator
Activity: 1218
Merit: 1079
Gerald Davis
The other "interesting" thing is nobody makes 2500W powersupplies (due to lack of need and lack of access). 
...
So maybe the device can only work @ 240V using a pair of 1500W PSU?

Getting a 2500W power supply is not hard, it's just not as easy as driving to your favorite PC hobby supplier to buy a couple. This is because they are more of an "industrial" solution than a "personal" solution.

I am not familiar with the US National Electrical Code and applications of it in households, but I imagine that you could easily install a dedicated circuit in the electrical panel that feeds a NEMA L6-20R (receptacle) and then get a cord that converts from L6-20P (plug) to the the appropriate receptacle on the power supply, likely an IEC C19. Standard PC power supplies use IEC C13 (on cord)/C14 (on power supply).

The compute chassis' that I use in my day job have four 2500W power supplies, one of which is a hot spare. I don't have one to look at right now to check voltage range, but I know we feed each one from one leg of three-phase 208v coming off our UPS, and I know that we have fed them off 240v single phase circuits. They use an IEC C20 socket.

I just looked at some IBM x3650M2 (single quad-core Xeon [email protected]) servers I have and the power supplies (two) are 675w and can be fed using anything from 100v to 240v. They appear to be made by Emerson and have an IEC C14 socket on them.

- Zed


Didn't mean to say no 2000W+ PSU exists but rather as you start to get into niche industrial gear costs goes way up.  Those 2500W PSU you are using in your workplace chassis likely cost 4x (maybe more) of a ATX 1200W PSU.

On US National Electrical Code...
Sure installing a dedicated circuit is fine.  I am considering putting in dedicated 240V branch in my garage.  Thinking 10 gauge NMC and L6-30R locking outlet.  That should be good for 250*30*0.8 = 6KW (not sure if 80% derate is necessary for dedicated 30A circuits?  If not it is more like 7.5KW).  Smiley  I picked up an L6-30R to C14 PDU for cheap off ebay.  So I guess if the magical rig box unicorn does exist someday I could run 2 rig boxes of the 30amp circuit.  Probably need better alarm system and doors if I am going to have 100GH farm in my garage.

Still I was just pointing out that wattage >1400W guarantees 208V 3phase (non-existent outside of data centers and industrial buildings) or 240V connection.  Without modification homes in the US simply aren't wired for it.  Now international it is a non-issue.
sr. member
Activity: 475
Merit: 265
Ooh La La, C'est Zoom!
The other "interesting" thing is nobody makes 2500W powersupplies (due to lack of need and lack of access).  
...
So maybe the device can only work @ 240V using a pair of 1500W PSU?

Getting a 2500W power supply is not hard, it's just not as easy as driving to your favorite PC hobby supplier to buy a couple. This is because they are more of an "industrial" solution than a "personal" solution.

I am not familiar with the US National Electrical Code and applications of it in households, but I imagine that you could easily install a dedicated circuit in the electrical panel that feeds a NEMA L6-20R (receptacle) and then get a cord that converts from L6-20P (plug) to the the appropriate receptacle on the power supply, likely an IEC C19. Standard PC power supplies use IEC C13 (on cord)/C14 (on power supply).

The compute chassis' that I use in my day job have four 2500W power supplies, one of which is a hot spare. I don't have one to look at right now to check voltage range, but I know we feed each one from one leg of three-phase 208v coming off our UPS, and I know that we have fed them off 240v single phase circuits. They use an IEC C20 socket.

I just looked at some IBM x3650M2 (single quad-core Xeon [email protected]) servers I have and the power supplies (two) are 675w and can be fed using anything from 100v to 240v. They appear to be made by Emerson and have an IEC C14 socket on them.

- Zed
sr. member
Activity: 448
Merit: 250
They only have the Bitcoin Dedicated (but oddly multipurposed) line, despite 10 years of experience or business or whatever it was.
The performance of the units are measured in 2 step SHA256 hashes, while they claim to be multipurpose with 'drivers' available.
Since it's FPGA, it can be reconfigurable and really perform other tasks that don't require external memory and input/output pins (in theory, if it's not Easypath-like chip).

Performance may be mentioned in pairs of hashes because those products were intended for bitcoin mining and other purposes added just to make it look more like serious business. (This is my opinion and it may be not true. I do think that their product really exists, but adding "other purposes" stuff on the website make things look worse for skilled bitcoin people. Truth is more suitable sometimes.)

Yeah, and they stated something earlier about how they didn't figure out that it was a 2-cycle process until they had already built it??

Delivery in 4-6 weeks, bitches :/ Shouldn't it be 2-4 weeks now, BFL?
donator
Activity: 532
Merit: 501
We have cookies
They only have the Bitcoin Dedicated (but oddly multipurposed) line, despite 10 years of experience or business or whatever it was.
The performance of the units are measured in 2 step SHA256 hashes, while they claim to be multipurpose with 'drivers' available.
Since it's FPGA, it can be reconfigurable and really perform other tasks that don't require external memory and input/output pins (in theory, if it's not Easypath-like chip).

Performance may be mentioned in pairs of hashes because those products were intended for bitcoin mining and other purposes added just to make it look more like serious business. (This is my opinion and it may be not true. I do think that their product really exists, but adding "other purposes" stuff on the website make things look worse for skilled bitcoin people. Truth is more suitable sometimes.)
hero member
Activity: 504
Merit: 504
Decent Programmer to boot!
Assuming the Rig box is legit, I estimate that the box will take about 10,000 Watts (yes 10k).

But I'd like to point out some things I have discovered that I find odd:

They claim to have made this line for Bitcoin, but they also 'have' drivers available for other uses. These drivers appear to be hyperlinks that lead back to the drivers page.
They only have the Bitcoin Dedicated (but oddly multipurposed) line, despite 10 years of experience or business or whatever it was.
The performance of the units are measured in 2 step SHA256 hashes, while they claim to be multipurpose with 'drivers' available.

The driver page thing is what inspired me to write this, but the rest of it confuses me a bit.

Anyone else find it odd? Feel free to prove me wrong.
hero member
Activity: 518
Merit: 500
Will be good to see what is inside the unicorn.
legendary
Activity: 3878
Merit: 1193
Get real. Take a look at the box:



When the Rig Box was designed, their "design" was the power output from a simulator. We now know how accurate that is. The first time they try to plug 50 GH/s into that little box they'll be in for a big surprise.

I'll send 5 btc to the first person to post what FPGA is actually on The Single. Let's see what makes it tick.
hero member
Activity: 504
Merit: 500
One thing as I've read all this that has concerned me is that I didn't stumble on BFL. I had been on google doing a search for "Bitcoin Rig Boxes" or something similar to that and at the very top of the search results was the adword/adsense? result that was of course a paid result. I clicked on it and the rest is history.

In previous posts I get the impression that their company was brought to light accidentally and that they were not ready for that. Maybe this is referring to the early October matters but I wanted to bring it up because it seemed like a contradiction.

Interesting.  BFL has always indicated they didn't want to start pre-sales but some people found then and they were dragged reluctantly into accepting large sums of irreversible funds.  Strange they would be advertising.

It is a tad interesting, isn't it?  BFL had touched on it a bit in the old thread. They basicly just agreed to what I had suggested about the 'seo' being accidental in its back linking. There was however, no mention of Adsense at the time. Which would be near impossible to accidently do with any seo software. I had at one point an Adsense 'sniffer' type app that would let me see the dates and variations of ads. Not sure I have it now but am curious if anyone else is able to find such data.?

@jjddebug, do you have any way to verify it was an Adsense ad, what was the text and approximate date?

Not that it really means much, but it is curious, none the less..  Still patiently awaiting to see some units delivered.  Undecided

Cheers
donator
Activity: 1218
Merit: 1079
Gerald Davis
One thing as I've read all this that has concerned me is that I didn't stumble on BFL. I had been on google doing a search for "Bitcoin Rig Boxes" or something similar to that and at the very top of the search results was the adword/adsense? result that was of course a paid result. I clicked on it and the rest is history.

In previous posts I get the impression that their company was brought to light accidentally and that they were not ready for that. Maybe this is referring to the early October matters but I wanted to bring it up because it seemed like a contradiction.

Interesting.  BFL has always indicated they didn't want to start pre-sales but some people found then and they were dragged reluctantly into accepting large sums of irreversible funds.  Strange they would be advertising.
hero member
Activity: 798
Merit: 1000
The good news (for me at least) is that you anecdotal confirms something I have long suspected.  Most FPGA buyers will be doing so because GPU mining is no longer profitable thus they are replacing GPU for FPGA which should result in flat or lower aggregate hashing power.  If that tends to be true for most buyers that means GPU will "last" longer in lower electrical cost environments.


For sure.  In my case, my hash rate will actually remain the same, so that's nice.

I haven't bothered replying to your other points as I'm already well aware of them.

EDIT:  Paypal payment has been made just now.
legendary
Activity: 1260
Merit: 1000
Not yet, but I've been pretty busy these last couple weeks.  I'm hoping things calm down after the New Year.
sr. member
Activity: 448
Merit: 250
Ok, I agree, maybe saying fairly easy is understating things.  I guess my point is that cooling it isn't going to be like trying to keep 8 5970's cool.  I can immediately envision several designs that would make cooling far more efficient given an effectively unrestrained design approach and a 4U box to do it in.

I agree, though, that if the power requirements merit 240v or unusual amperage, it should be made clear to the consumer that you can't just buy one of these things and set it up in your broom closet.


Has there been any progress on getting you a unit for testing?

I know I am like, 90% trolling in this thread, but I am very interested to see how this all pans out, and as I stated before, I might even buy one if one exists...
sr. member
Activity: 341
Merit: 250
yeah i saw the bfl web page before i even saw this forumn.  i looked at it my very first day of learning about bitcoins.  its right there at the top if type in "bitcoin mining".  i'm following this thread really closely bc if its real i will buy one.  but for me 600 is a ton of money so i can't risk just throwing it away on anything less than 800 m/hash 100w.  i'm not trying to make money with it, i just enjoy mining bitcoins and would like to solo mine ( i know it will still takes months to get a block) just to do it.  anyway, i'm really glad this thread is here so i can make an informed decision.
legendary
Activity: 1876
Merit: 1000


just my 2 bitcents about the power.  It will require 240v,  that is a no brainer, even for my little brain Smiley 
sr. member
Activity: 446
Merit: 250
I ordered my singles middle of November. I came across this or actually the previous thread about a week later and got very concerned about it possibly being a scam. I decided to ride it out and wait and see.

BFL has kept in touch with me or at least they reply to my emails whenever I have contacted them. I'm currently told that I can expect shipment third week of January. (Originally delivery would have been about right now)

One thing as I've read all this that has concerned me is that I didn't stumble on BFL. I had been on google doing a search for "Bitcoin Rig Boxes" or something similar to that and at the very top of the search results was the adword/adsense? result that was of course a paid result. I clicked on it and the rest is history.

In previous posts I get the impression that their company was brought to light accidentally and that they were not ready for that. Maybe this is referring to the early October matters but I wanted to bring it up because it seemed like a contradiction.

I'm looking forward to having my "faith" in this proven valid. If not then its an expensive lesson as I have purchased quite a few.

legendary
Activity: 1795
Merit: 1208
This is not OK.
The power issue would require a slight house re-wire. Most houses in the US have 2-phase supply... each phase 120v and 180 degrees off each other. Normally these are referenced to ground and different rooms/floors use different phases. Put together though that's 240 Smiley (used for electric dryers, stoves etc). Cooling would be equally as difficult to maintain - hence the dream. Not stopping me thinking of possible solutions though Smiley
legendary
Activity: 1260
Merit: 1000
Ok, I agree, maybe saying fairly easy is understating things.  I guess my point is that cooling it isn't going to be like trying to keep 8 5970's cool.  I can immediately envision several designs that would make cooling far more efficient given an effectively unrestrained design approach and a 4U box to do it in.

I agree, though, that if the power requirements merit 240v or unusual amperage, it should be made clear to the consumer that you can't just buy one of these things and set it up in your broom closet.
Pages:
Jump to: