Coming in v2.0 of my rigs.  Currently they're just PCs on standard metal shop shelves with fans circulating air accordingly.  My BTC profit model has been to get up and running as fast as possible first, then optimize for electrical efficiency using the profits.  As I pull hardware out of OEM cases, they'll be going into the cube stacks.

I've seen a few other rigs put into these kinds of things, not necessarily exactly the same, but close.  I think they're the way to go IMHO.  Can't beat the price and they're great for airflow. 

I still fail to understand why you'd want a "real" for bitcoin mining.

Why not just connect the green and black wires on each power supply?  They can be daisy chained afaik.  I've only had need to try it with two so far.  $20 for an adapter cable is steep.  Decent 500w power supplies can be had for just shy of double that.

Besides "no case" which gets hard to maintain when dealing with larger numbers, I humbly submit the following:

http://www.target.com/Wire-Cube-Shelving-System/dp/B000789RSQ

Rack and stack.  The above and zip ties are really all you should need.  Why waste money on a case?

Yeah, I don't think there is a perfect way to do what I'm trying.  But I wouldn't mind even some rough estimates.  Would be nice if exchanges also reported transaction numbers too.  That and pool numbers would be a decent sum to factor out of the global transaction figure.  The remainder, I contend, would be mostly goods and service based transactions.

Wait, what?  You have no clue how much current each machine pulls.  Nor do you have a clue what I pay for in electricity.  In fact, I humbly contend that you simply have no clue.

Yes, and I will be upgrading.  The thing is, I paid next to nothing for my machines (eg., mobo, cpu, ram, cables, chasis and fans).  

Go ahead, try and build 24 mining rigs for nothing more than the price of the video cards.  

Upon initial investment, $/mh/s is more important than $/kw/h.  

I've already made more than $2000 from the machines, AFTER the cost of electricity.  Had I spent more than double my initial investment (see below) I might have saved $100 on my electric bill, but that would have effectively doubled my initial investment, perhaps more.

Now, I can benefit from low cost bitcoin mining and easily upgrade to more efficient systems AT NO ADDITIONAL RISK to my investment.

I can easily set up rigs with 4 cards each in them running much more efficiently kw/h wise.  But that requires an investment in the motherboard, cpu, ram, chasis, cables and fans.

Balancing $, kw/h and initial investment is what I'm all about.  And as my friend, kookiekrak, illustrates ... not everyone has a solid grasp on how to do that.

If you'd like to get a ton of machines up and running quickly, at a very low cost of entry, I can help.  I still have industry contacts selling hardware at ridiculously low prices.  If, however, you'd like to build "state-of-the-art" rigs maximized for electrical efficiency I can help you there too.  Of course, you may want to balance the two just like I've done for myself.

As well as the physical manpower to put together efficiently (in $ and kw/h) built and spec'd bitcoin mining rigs.  

I'm sure many here on these boards can set up rigs.  I'm just selling my ability to do it quickly, as I said efficiently, and solidly under Linux.  I've basically come up with my own linux distro (eg., Ubuntu plus lots of tools I've written) to make managing dozens or even hundreds of machines pretty simple.

Oh, and I guess it goes without saying, I accept bitcoins as payment and will even work off a % share of your rig's profit.

Have a lot of money but don't know how to set up a bitcoin farm?

Have no money, but know people who do, and want to set up a bitcoin farm?

Any combination of the above, and want to set up a bitcoin farm?


Send me a note.  I have 24 machines currently running myself at around 8 Gh/s.  Can set up similar for you in just a few days (after parts arrive).  10 Gh/s to 1000 Gh/s ... it all depends on how much you want to spend and how much current your home or office space can handle.

All Linux.  All secure.  Rigs can be monitored via web page or mobile phone as well as reconfigured for custom power and temp settings.

I've been setting up Linux boxes and developing software for Linux since before Kernel 1.0.

./glakkeclock -ogcm -ogpl -cdi 0
2
1250

# It won't let me set mem clock here either to anything that low

./glakkeclock -oscm 400 -ostcs -cdi 0
./glakkeclock -ogcm -cdi 0
1250

./glakkeclock -hgil -cdi 0
Name: AMD Radeon HD 6900 Series
UDID: 256:26393:4098:12579:5762
Index: 0
Adapters: 4
Part: 113-695ZNF72-113-AC93301-100_XF
Version: 013.012.000.009.000000
Date: 03/30/11 23:45

Current Performance Level: 2 [0-2]
GPU (MHz): 831 [125-840]
Memory (MHz): 1250 [75-1325]
VDDC Volt (mV): 1100 [900-1100]
Temperature (C): 74
Fan RPM: 4317 [800-6000]
Fan percent (%): 85 [0-100]
GPU Activity (%): 99
Current Bus Lanes: 16/16
Current Bus Speed: 5000

Performance Level: 0
GPU (MHz): 250 [125-840]
Memory (MHz): 150 [75-1325]
Vddc (mV): 900 [900-1100]

Performance Level: 1
GPU (MHz): 831 [125-840]
Memory (MHz): 400 [75-1325]
Vddc (mV): 1100 [900-1100]

Performance Level: 2
GPU (MHz): 831 [125-840]
Memory (MHz): 400 [75-1325]
Vddc (mV): 1100 [900-1100]
topper@digger2:~/bin$

Oh well, not a big deal.  The cards are running 95% and above factory core peak.  Worst case I'm burning some watts on the higher memory clock.

No dice.  Same behavior.  All glakkeclock does is seem to reset the lower end for peaks not actual clocks.  Even if I use aticonfig to then subsequently reset the clocks to somewhere in the new lowered range, it only does the peak.  The effective clock remains the same.

I've tried under 0 load and full load.  Same behavior.

./glakkeclock -oscm 300 -oscg 600 -ostcs -cdi 0

Default Adapter - AMD Radeon HD 6900 Series
                            Core (MHz)    Memory (MHz)
           Current Clocks :    600           1250
             Current Peak :    600           300
  Configurable Peak Range : [600-840]     [300-1325]
                 GPU load :    99%

Default Adapter - AMD Radeon HD 6900 Series
                  Sensor 0: Temperature - 72.50 C
Fan speed query:
Query Index: 0, Speed in percent
Result: Fan Speed: 79%


Then:

aticonfig --odsc=700,300


Gives:

Default Adapter - AMD Radeon HD 6900 Series
                            Core (MHz)    Memory (MHz)
           Current Clocks :    700           1250
             Current Peak :    700           300
  Configurable Peak Range : [600-840]     [300-1325]
                 GPU load :    0%

Default Adapter - AMD Radeon HD 6900 Series
                  Sensor 0: Temperature - 72.00 C
Fan speed query:
Query Index: 0, Speed in percent
Result: Fan Speed: 75%

So, it seems as though glakkeclock works for core clocks and peaks, but not for memory clocks.  

It only seems to reset memory "peak" which from what I can tell does nothing.  My chip runs at the same temp and mH/s.  Similarly, the memory clock remains unchanged.

EDIT:  If I try to set my memory clocks somewhere ABOVE the factory default minimum of 1250, that still does work.  So there seems to be something in the card (?) that is only allowing glakkeclock to mess with core settings, not memory.

EDIT2:  Tried backing down clocks more slowly.  Guess what, glakkeclock WILL let me set memory to as low as 700, but that's it.  Anything below 700 and it's the same behavior as above.

Sadly, that doesn't give me much help in terms of temp and/or speed.  My chips are already running close to max core (835/840) at 75C.  Changing the memory clock to 700 does save me a few deg C (down to 72.5) but no love in terms of speed.  ;-(

Here's an example.  Current clocks:

Default Adapter - AMD Radeon HD 6900 Series
                            Core (MHz)    Memory (MHz)
           Current Clocks :    824           1250
             Current Peak :    824           1250
  Configurable Peak Range : [500-840]     [1250-1325]
                 GPU load :    99%

Default Adapter - AMD Radeon HD 6900 Series
                  Sensor 0: Temperature - 74.00 C
Fan speed query:
Query Index: 0, Speed in percent
Result: Fan Speed: 82%


Now I try:

glakkeclock -oscm 300 -cdi 0
[ERROR] <ADLManager> Generic Error. Most likely one or more of the Escape calls to the driver failed!
[ERROR] <ADLManager> ADL_Overdrive5_ODPerformanceLevels_Set
[ERROR] <Main> Error when setting clocks/vddc.
[WARNING] <Main> Performance level 0's values are higher then 2's specified clocks.
[WARNING] <Main> Performance level 1's values are higher then 2's specified clocks.


So then I try:

glakkeclock -Cpl 0 -oscg 800 -oscm 300
glakkeclock -Cpl 1 -oscg 800 -oscm 300
glakkeclock -Cpl 2 -oscg 800 -oscm 300


And now I have:

Default Adapter - AMD Radeon HD 6900 Series
                            Core (MHz)    Memory (MHz)
           Current Clocks :    800           1250
             Current Peak :    800           300
  Configurable Peak Range : [800-840]     [300-1325]
                 GPU load :    99%

Default Adapter - AMD Radeon HD 6900 Series
                  Sensor 0: Temperature - 74.50 C
Fan speed query:
Query Index: 0, Speed in percent
Result: Fan Speed: 82%


But I can't set the core memory CLOCK to anything below 1250.  Just the peak low end. 

aticonfig --odsc=800,500

Default Adapter - AMD Radeon HD 6900 Series
                  New Core Peak   : 800
                  New Memory Peak : 500


Which gives me:

Default Adapter - AMD Radeon HD 6900 Series
                            Core (MHz)    Memory (MHz)
           Current Clocks :    800           1250
             Current Peak :    800           500
  Configurable Peak Range : [800-840]     [300-1325]
                 GPU load :    99%

Default Adapter - AMD Radeon HD 6900 Series
                  Sensor 0: Temperature - 74.50 C
Fan speed query:
Query Index: 0, Speed in percent
Result: Fan Speed: 84%


So while it seems to let me reset the low end of things, it won't let me actually change the clocks.  I must be doing something wrong, but I'm not sure what.

Really?  How?  I compiled it and it "works" but whenever I try (for example) to set my clocks to 900,300 it only sets the peak, not the current clock for memory.  It also does something odd with core, narrowing the range to something like 824-889 or something like that.  I think it's an issue with the "smooth" flag or whatever that was.

Did you do something else prior in order to help glakkeclock set your memory clocks so low?

Well, pools and whatnot (who I think make more than 50% of all mining) might voluntarily report their "send from" addresses.  That would allow us to subtract those transactions and leave a rough estimate for the others.  The assumption there would be that "the others" constitute "real" transactions.

Come to think of it, it should be possible to do this without any type of voluntary reporting on the part of pools. 

Simply taking the total off all the pool hashrates and mapping them to the corresponding BTC/day should do it.  The assumption here being that everyone takes payout pretty often and doesn't store BTCs with their favorite pool(s).

I guess there needs to be some work done on the # of transactions and the avg. value of those transactions.  EG., 10 transactions at 0.1 BTC each != 10 transactions at 10 BTC each with respect to what I'm trying to find out.

I'm talking about BTC transfers for goods, not from mining pools to miners or miners directly by finding blocks.

I think that's what we need to start looking at.  Unless people start using it more to buy and sell goods & services ... BTC won't survive.

Thanks.  I'm trying to mull over the pros/cons of pool hopping.  The morality argument smacks (to me) of youth.  If anyone can pool hop, then it doesn't seem immoral since the opportunity is there for everyone.  Similarly, if more people hopping would make hopping less profitable, then even more so.

It's like saying you're "cheating" by driving your car more efficiently than its EPA MPG rating.  Technically, you're cheating the oil companies, who like it or not DO influence Washington when it comes to MPG legislation.  So by driving your car more efficiently, you're hurting the oil companies who then have to raise the price of oil, which hurts everyone.  Pretty ridiculous argument, no?  Logically equivalent though.

Good for people (especially young folks) to be *thinking* about morality though.  Shows it's not totally dead.  So I'm not gonna rail on them for it.

Been trying to Google and am not coming up with much.

Ok, I'll wait a few days and re-post my question in (hopefully) a more hijack proof manner.

I was mainly curious about which would be most profitable:  BTC mining, NMC mining, or BTC pool hopping.

Seems like the latter is the winner but I'm still somewhat at a loss as to how it's done best.  I mean, if I just rotate my miners every 10 minutes or something I'm not clear on how that improves my profit.

I think my lack of understanding is specific to how "shares" are translated into BTCs on various pools.  I know Slush's pool uses a score based mechanism that discourages hopping.  But "straight up" share based mechanisms (like Deepbit, BTC Guild, etc...) seem as though hopping would net you no real gain.  My thinking is that when you hop, you get a proportionately smaller share. 

So I'm not clear on how your "hopped" shares from differ from the sum total you'd normally get by staying with one pool.

Got glakkeclock built but it doesn't allow me to underclock memory.  It "thinks" it is.  But my cards still clock as they were before.  ;-(

Cool, thanks.  I don't think we need to solder wires.  These wire taps should work fine, no?

http://www.radioshack.com/product/index.jsp?productId=2104093

How do you know when a pool reaches 43.5% (or any %) since block solution is random?

Nice job.  Only minor critique is that you've got the GPUs pretty close together.  Why not space them farther apart to allow more airflow?  You can also angle them differently to catch air from something like a box fan, if you're dedicating a room in your house for tech that is.