Well unless you add a block to the CPU, don't count it in (same goes with PSU inefficiencies), since all the heat will only be coming from what you have blocks on. You could do 2 or 3 5970s, and even downclock them for a better mhash/watt while still getting enough to keep the pool warm.
If it's underwater, if the heat does not go into the water, where else will it go? =) If heat is the desired product, there are zero inefficiencies. The only thing there needs to be is something to make sure the CPU and PSU get their heat carried away so they don't burn up. There will not be any net air flow - the air pocket will get hot in proportion to how well the heat can transfer out of the air and into the water. A CPU block and a water-cooled power supply would probably be necessary. So you you mean a completely sealed capsule/module that contains a miner that produces heat? And then you dunk it in? |
|
|
|
Those DPS-2000BB power supplies are a nice find.
Only problem I just discovered is that they do not have their own fans. So it will be necessary to hook up some fans to them so they don't overheat and die. Also, the 2000 watt version is less efficient than the 2500 watt version, and the 2500 watt version isn't for sale anywhere that I can find. did you find testing reports? link please Actually, I thought I had found the report for the 2000 watt version, but I must have been mistaken as I cannot find it anymore. The report for the 2500 watt version is here: http://www.plugloadsolutions.com/psu_reports/IBM%20CORP_DPS-2500BB%20A_2500W_SO-172_Report.pdf However it is not available for sale anywhere that I have found. |
|
|
|
Those DPS-2000BB power supplies are a nice find.
Only problem I just discovered is that they do not have their own fans. So it will be necessary to hook up some fans to them so they don't overheat and die. Also, the 2000 watt version is less efficient than the 2500 watt version, and the 2500 watt version isn't for sale anywhere that I can find. Are there mounting holes for fans and for what size? Not sure, those PSUs are intended for use in blade servers that have their own cooling. The PSUs that I actually currently own have their own fans. |
|
|
|
|
Well unless you add a block to the CPU, don't count it in (same goes with PSU inefficiencies), since all the heat will only be coming from what you have blocks on. You could do 2 or 3 5970s, and even downclock them for a better mhash/watt while still getting enough to keep the pool warm.
|
|
|
|
Ok, maybe i am not articulating myself clearly enough. DaT: Those figures were examples, completely from hat, meant to make a point. Plus you misunderstood what i were saying  rjk: Such a lookup of table of finished hashes or something, was not at all my intention. What i meant, there is cases where at certain situation a certain outcome is certain to 100%. Expanding THOSE lookups where we know that something will result into something 100% of the time is my intention. Let's say beginning of the hash is e5da3, and for some statistical miraculous reason if char[0] == e, char[3]==d, char[5] == 3 then char[6] == a when we know that what we are hashing is N characters long. Or maybe if the beginning of hash is 6be3c118da8acd56 we know the last character is 5, thus cannot be our match. Or maybe if the first round beginning is 6be3c118da8acd56 we know that next round first 3 chars will be dcc, and therefore not a match. Pending statistical analysis naturally. I don't understand why that kind of statistical analysis is so difficult to fathom. Similar things ARE being done with FPGAs, and probably on the GPU code. Also, we know the format the original data is in, which might result in some odd coincidences in the resulting double hash. Simple answer: No, won't work - have a look at the SHA256 algorithm to find out why. Slightly longer answer: actually, some optimization has been made requiring less than 64 full rounds to complete, but the savings was very little, and only was at the tail end of a round. The statistics are not going to be able to do what you are asking of them because of the final parts making the output completely random. You cannot statistically analyze random numbers in order to predict them or to predict parts of them, because then they would not be random. If you have an understanding of how SHA256 works, and have had an epiphany of how to create some kind of side channel attack (which is what your proposal amounts to), then show us some code so we can have a look at it and pick it apart. |
|
|
|
Could it be avatars? It appears the forum software does not host them locally, but simply redirects to the original site hosting the image.
That would be it it. There is an option for local storage, but no one seems to use it. |
|
|
|
When first landing at the website https is good, secure. As I drill down to post into a topic it becomes normal or insecure. With the latest chrome 18 it is fine until in a topic, then https is lost, backing out and refreshing 'secure' returned, enter topic, https is lost. Chrome 17 and Opera 11.61, once you drill down into a topic, the https is lost; up one directory, refreshing does not return https, it remains insecure.
Is this a site issue, a certificate issue, or a browser issue?
I am assuming that you mean you lose the padlock icon, or the blue bar? That could be caused by loading external images from non-secure sites. Or do you mean it actually switches between https:// and http:// ? I haven't seen that happening. |
|
|
|
I'm starting to see how you are thinking now. Most hot tub owners that I know of keep the thing on 24/7 so it is ready to go, but it sounds like you wish to only turn it on when you want to use it. In that case, you will need a lot of wattage, otherwise it will take forever to heat up.
I mean something like the opposite: I have to keep it warm 24/7 (or else it would freeze over), but only use it for 15 minutes here, an hour there. The miner would keep it warm the rest of the time while it wasn't being used, and having it underwater would eliminate the noise problem. The very fact that I HAVE to keep it warm in the winter (the only other choice is empty and unusable) means the power cost has already been justified the moment I decided to keep it full. OK I was confused by your statement of unplugging the box and taking it out, sorry. Still the extra cost would probably not be worth it. Let's start with this: how many watts is the current heater, and what is the approximate duty cycle that it is running at? I am assuming it is not running full on all the time, but is rather switching in and out to keep the temp steady. |
|
|
|
My RSVP still stands. I will be there. I can't wait to meet you gentlemen.
Did atlas show up like he promised to? |
|
|
|
It doesn't need to be anything fancy with connections. Just a box you stick underwater when you aren't using the hot tub.
As for "shockingly good time" just unplug the box and take it out when using the tub silly. And use a gfi outlet so the power is cut if any leaves the box into the water as a secondary measure (gfi is already installed on most outdoor outlets anyway on modern construction).
I'm starting to see how you are thinking now. Most hot tub owners that I know of keep the thing on 24/7 so it is ready to go, but it sounds like you wish to only turn it on when you want to use it. In that case, you will need a lot of wattage, otherwise it will take forever to heat up. |
|
|
|
You sure that OK Gear one is 16ga or better? One other issue I see is ... is 12" enough to actually run 3 units spread out enough? Might be kinda tight depending on heat envrionment.
You can always do a double splice connector ---- butt splice ----- extension wire 2 ft ----- butt splice ---- wire terminated barrel adapter but the longer the cable the more important it is to have thick wires to avoid voltage drop and the corresponding current / temp increase. The reason I mention the size is because ngzhang just recently burnt some such similar extenders when testing Icarus. Not sure how many Icarus he was testing per wire, but it must have been more than 2, cause the connections went brown fairly quickly. |
|
|
|
|
Having such a lookup table would require you to know the end result of the solution that we are bruteforcing, which is impossible. It sounds like you are saying to proceed through about half of the SHA256 calculation and then create a lookup - this won't work because you can't predict the output from only running half the algorithm.
|
|
|
|
You sure that OK Gear one is 16ga or better?
Not sure, that's why I said buy 1 to find out. Might be able to look them up and find a spec sheet, but I doubt it. |
|
|
|
The multiplier in the DSP48-block is not needed in SHA-256, hence what he obviously uses is the 18-bit adder
BCOUT = B + D.
Nah; I use the DSP48s as big fat FIFOs; they have lots of registers inside and if you configure them right everything's a no-op. Interesting use case; so essentially no added latency using them this way? |
|
|
|
well either way it looks like that server that was hosting the bitcoind has been shut off. Seeing as it hasn't made any blocks in the last 2h50m. But its worrying that no block has been made it in the last 30m. (According to blockchain.info)
Even 1thash pools have bad luck. Don't rule it out just yet. |
|
|
|
Probably that the loss in efficiency isn't much to worry about when the farm makes so much money during power outage. FWIW, I wouldn't bother with a UPS though, power is never out very long around here.
My thinking was just that since he is overly concerned about the quality of the power a line interactive UPS would ensure a very high quality input AC power. Near perfect sinewave with the UPS boosting sags and absorbing spikes. Ah yes I didn't consider that. A double-conversion UPS would certainly add protection at the cost of some efficiency (not sure how much though). |
|
|
|
The important thing will be to ensure you're not overloading any wires. The setup that ngzhang posted looked a little "ghetto" lol (I suspect he was rushing and just slapped it together, as I've done in the past myself once or twice  ) It looked to be hand twisted wire with tape sealing it up (no offence ngzhang, and if I'm incorrect, let me know). This is prone to high resistance at the twisted connection, and will result in a lot of heat generated there (and risk of fire). Your best bet is to make sure not to exceed the rating (4 pin molex connectors are rated for 11A at 12V or 132W) The Icarus draws about 20W. So that would mean if everything is perfect, you could run 6x Icarus boards. But all is rarely perfect. And Molex connectors are renowned for arcing and other problems (which can generate excess heat). Best to keep it to no more than 4. Also ensure a clean connection (well soldered joints, not just crimped) and so on. I'm making several "Molex to Barrel DC Jack" connectors, which I'm building to a fairly high standard. To allow me to run up to 4x Icarus per Molex. On that 700W supply I'll be using 2 Molex, (8 Icarus total). And I am building adapters from the PCI-E connectors in the same fashion (rated 150W each and designed to deliver high DC current) to power 8 Icarus each. That will give me a total of 24x Icarus boards off that PSU. (24x 20W = 480W roughly, and that supply sends nearly 600W to the 12V rails, so it should be able to handle that just fine) Anyway, basically ensure clean connections, good joints, and always give yourself plenty of headroom from the rated spec on all your components (keep an eye on the weakest link). Once you have the rig built, I would check all power lines for heat. If you have wires running hot to the touch that's probably bad.  It looks like the actual Molex connectors are burnt. This is likely due to inexpensive underrated pins and tiny gauge wire. |
|
|
|
I have trouble finding high enough capacity UPS units that don't cost more than it's worth to protect, though...
For GPUs I agree but he is talking about ztex FPGAs. At $250 per board (even in bulk) and 8W it is roughly $31 per watt. One can get a UPS for 1% of total system cost. Sounds like a good investment to me. Now GPUs I wouldn't consider a UPS. I'm not sure I see how that matters, if he's loading down a 1200w PSU with 125 Zetex or 14 BFL units, the load is the same to the UPS. Am I missing something? Probably that the loss in efficiency isn't much to worry about when the farm makes so much money during power outage. FWIW, I wouldn't bother with a UPS though, power is never out very long around here. |
|
|
|
|
Show Posts
