That might or might not pan out like you think it will. There are some fundamental mathematical reasons why GPU mining is difficult for primes. I expect that eventually mtrlt will succeed but the performance gains will not be as extreme as when SHA-256 (bitcoin) was going from CPU to GPU.

As long as the GPU miner doesn't come out publicly within 7 days I should break even on what I paid for these (but probably not recover the cost of my time spent on them lol). Also, I have all the old cases and parts, I could sell these as 'faulty / for parts' without PSUs on eBay for more than I paid. That being said, racing out and buying a rack full of new servers would be a bad idea, if you can't use them for anything else.


Profitable mining always depends on a balance of performance, efficiency, and initial expenditure. These old Dells have the last two which makes up for the first (performance). Core2Duo CPU's are the lowest I would consider using for mining now.

Got some really cheap Dell Optiplex PC's to add to my prime mining network at home. They were really cheap, apparently because 3 of 4 of them had dead PSU's. So just plug in another one right? Wrong. Dell and HP small form factor / slimline desktops use a quasi-proprietary mini 24-pin motherboard power connector that is about 1 cm smaller than a normal ATX motherboard connector. Sure, there are some adapter on ebay for about $15 from China or USA... but none in my country.

The solution was clear to me:





So I did that three times each for the three cables. Saved myself $30 on ebay + 1 month wait.

The next problem was that a normal PSU is twice the size of the Dell original. Also these cases get REALLY hot (50-60 deg C) when under 100% load for a long time, and with the original Dell PSU's they're even worse. The one with the working Dell PSU has the entire inside of the case at 50+ degrees C after 1 hour at load... not good for the components. The cooling in those Dell PSU really is terrible, you get what you pay for I guess.



And so it looked pretty silly with 3 big PSU power 3 small desktops, so I figure I will remove the cases entirely... FPGA stack style.

Taking shape!



Here's the 'installed' system on top of my bookshelf. Each has a USB to boot off.





Because installing windows for every machine takes too much time // I can't linux btw :S // I am investigating using VMWare ESXi 5.1 virtualization on them. But this is a learning curve in itself... and I have had to completely alter one of my servers just to run try and run them.

Oh yeah, each desktop is a core2duo @ 2.93 and with 3 or 4 GB ram.

EDIT: Oh yeah the three larger PSU are really old (~2002 or earlier). I'll be replacing all of them with new ones soon.

Can I play the big fan game too?  

• 
• 
• 
• 
• 
• 
• 
• 

It needs to run off a car battery jump-starter (600 A peak current) because none of my bench PSU could handle it. It needs about 200 watts or more. Even my small desktop PC ATX PSUs were no match.

I made a bag thing out of a painters dropsheet to show how much air it moves.

http://www.youtube.com/watch?v=ihvaFVsB4G8

Right, as has been pointed out, chainsperday is a count of how many chains should meet the integer difficulty requirement, not considering the fractional difficulty requirement that must also be met.

Simple example:

If your chains per day was 1.0 and diff was 9.0100 then you will probably find 1 block per day on average.
If your chains per day was 1.0 and diff was 9.9985 then you probably won't find a block

Please correct me if my understanding is wrong...

However, like with any solo mining, you are still subject to variance ("luck"). You could find 4 blocks the first day you mine then find nothing for a week or more... etc

Donations ≠ payment

If you sent money to him, you did it out the kindness of your own heart. It's cringeworthy if you expect him to make software for you just so you can get rich, and expect him to work on it 24/7.

Now in HD! Up to block 88877, 2013-07-29 22:48:52 UTC, anyway.

https://i.imgur.com/l9KCkEB.png



https://i.imgur.com/z2gW5XS.png

No. Thats not what he is saying/asking. 1l1l11ll1l has several relatively nice servers mining, and was wondering why the 24-core server seem to outperform those with a total of 32-cores. Refer to my response above for one possibility why.

From PassMark and the opteron wiki page:
http://www.cpubenchmark.net/cpu_list.php
http://en.wikipedia.org/wiki/List_of_AMD_Opteron_microprocessors


Dual CPU, 12-core opteron 6164HE's   PassMark CPU result: 5351/ea, 5351*2 = 10702 ||| Cache arrangement; L2:  12x 512 KB    L3:  2x 6 MB
[Dual CPU] AMD Opteron 6274   PassMark CPU result: 10809 (inclusive of both) ||| Cache arrangement; L2:  8x 2MB L3:  2x 8 MB

If I had to guess, the dual cpu, 16 core setup (6274's) is slower because it shares one unit of L2 cache between two cores. The HE's have dedicated L2 for every core.

Despite the disappointing(?) performance, those are still all nice systems and I would mine on them any day.

A good rule in life is: No, you can't get something for nothing. There is always a catch.

In this case you risk destroying your miner quickly, or possibly reducing its life. As the number of people who've tried changing the oscillators can be counted on one hand, we have no idea how long these modified units can run for. YMMV.

I think 51% is unlikely, estimates show there are already equivalents of thousands of servers (pretty much all of DigitalOcean amiright?  ) mining away. Then add all the miners using desktop CPU at home, laptops even, etc.

I think that not many (if any) persons would have the resources to pull off such an attack unless they devised an extremely efficient GPU miner that was privately released.

Any organizations looking to perform a 51% attack against primecoin would (at this time at least) need to devote huge server infrastructure to to the task. I don't think any college/universities high performance computing clusters would be up to the task (those are sometimes primarily run by 1-2 people and could lead to means of an attack). I don't think any company - even with their own datacenters - would have much interest in a primecoin attack even if they were capable, because those resources are far better used serving clients.

If you want to worry about 51% attack, I think primecoin is relatively safe. On the other hand some (all?) of the SHA-256 altcoins are in a bad place to be right now, eg. terracoin, but that was only possible because of ASIC.

The increase in network diff is not necessarily because of new miners and new hardware coming onto to the network.

It could simply be that the new code optimizations and defaults in HP9 are resulting in faster block generation and the network is adapting.

It's really easy to see those amongst you who have solo mined before and are used to having large variances in # of blocks found in any time period compared to those who have probably only ever used pool mining. I think the later have some unrealistic ideas about how block probabilities work and so get all prissy when they don't find a single block in a day.

Also be mindful of the scale of the primary Y axis in the linked graph, the full scale shown is only over the range 9.18 - 9.32, you can graph pretty much any data on a small enough range and make it look more/less meaningful than it really is.

From OP's posts it looks like they've done a lot more testing of those variables than anyone else, especially me, so I've started to use the HP9 defaults because of this.

It looks like a (fairly old) variac, which is a variable voltage AC toroidal transformer. They're useful for controlling the power input into simple motors and static-type loads, like heaters. They're also good for blowing up electronics, because they can usually be turned up to give a voltage higher than their mains input.

My tests show that memory is not very important for primecoin. I believe its mostly CPU speed and cpu-cache that are important. However, inadvertently on most high end systems with fast processors you will also have fast RAM, especially true of gaming systems (less so on servers but they are different story because of ECC RAM).

Difficulty has dropped throughout today. Within a week we might peek back under 9.0 diff. That would be interesting. I added some more servers today at my place and I'm getting a few thousand more PPS. Bring it on I say!

Fan-cooled PDU's. Love it.

The release of a GPU miner might just counteract all those expiring 30 day DigitalOcean trails? Amiright guys? 

There is no onboard temp monitoring because it keeps costs down. ASIC are internally simple (relatively) there are images of a the die on a site which I conveniently can't find where someone etched the epoxy away with hot nitric acid.

Adding a sensor would increase ASIC manufacturing cost and number of components in the circuit, and not really required for normal use.

You are assuming the motive of the attacker(s) is to make money.