[XPM] [ANN] Primecoin High Performance | HP14 released! - page 14.

Trillium

hero member

Activity: 546

Merit: 500

Quote from: abdulmad on November 14, 2013, 03:31:15 AM

Ok guys i managed to acquire a HP ProLiant DL585 G7 Server, fully poppulated with 4 AMD 64 FX dual-core Opteron 6278's, thats a total of 64 cores running at 2400 default and 2.7 Ghz turbo for 600$, good deal or bad deal. Which will be the best use for this server, solo or pool mine. Your thoughts?

You'll probably never recover your initial expenditure of $600. However, depending on other specs, getting a G7 for that price is probably not too bad. You can possibly resell it for more than that in the future (on eBay) when you're done mining with it.

How much power consumption at 100% load? I hope your running it at work where that doesn't matter. Also, noise.

Solo or pool depends on how long you can wait between payouts. I'd probably just aim for solo.

Although the number of cores sounds impressive, like many people here who've gotten high-core opterons you may be disappointed with the overall performance.

At 100% load what is your chainsperday in HP11? From there you can estimate how much coin it will generate and how quickly it'll start paying itself off.

abdulmad

sr. member

Activity: 672

Merit: 250

Ok guys i managed to acquire a HP ProLiant DL585 G7 Server, fully poppulated with 4 AMD 64 FX dual-core Opteron 6278's, thats a total of 64 cores running at 2400 default and 2.7 Ghz turbo for 600$, good deal or bad deal. Which will be the best use for this server, solo or pool mine. Your thoughts?

Trillium

hero member

Activity: 546

Merit: 500

Quote from: Ethera

I give a heads up for the guy, as the diff rising steep shows that the gpu computation did enter the scene.

A statement for which you can provide convincing evidence for, I am sure.

Ethera

member

Activity: 69

Merit: 10

Quote from: ?? on ??

Quote from: Stinky_Pete on November 10, 2013, 04:17:13 PM

I don't believe his claims, and won't until I see code running.

+1
We're still dealing with crybabies who have absolutely no idea how computers work.
Also his post made zero sense.

Actually his posts makes all sense. You claiming that someone has no idea how computers works just shows how childish you are (let me guess, around 16-21 age, no idea what you are talking about?)

I give a heads up for the guy, as the diff rising steep shows that the gpu computation did enter the scene.

Wow this is exciting times. Probably the first coin where cuda will have its share. Gamers will rejoice Cheesy

(IF he ever releases that to public. Heads down to him, GENIUS at the work.)

(disclaimer, i run a big amd farm for ltc but i can and do recognize when i see genius at work).

Neil

Stinky_Pete

hero member

Activity: 560

Merit: 500

Quote from: satriani on November 09, 2013, 09:16:27 AM

there is some guy here:
https://bitcointalk.org/index.php?topic=251850.3440
and he can mine 30 block/days on 5xGPU. could it be true ?

You are referring to Supercomputing? I don't believe his claims, and won't until I see code running. Anyone can claim to have found problems with the algorithm, writing improved code is a completely different thing.

I see he also brings up the cliched theory that 'someone has a GPU miner and is keeping quiet'.

satriani

member

Activity: 93

Merit: 10

Quote from: itod on November 09, 2013, 09:17:39 PM

Yes, everything he says makes sense. He refuses to take any money, that gives him some additional credibility.

so why the f... nobody discus about it? its just unbelievable.
we have some conspiracy or what?

itod

legendary

Activity: 1974

Merit: 1076

^ Will code for Bitcoins

Yes, everything he says makes sense. He refuses to take any money, that gives him some additional credibility.

satriani

member

Activity: 93

Merit: 10

there is some guy here:
https://bitcointalk.org/index.php?topic=251850.3440

who claims that previous inventor of GPU-miner were a scamer and he can mine 30 block/days on 5xGPU. could it be true ?

altsay

sr. member

Activity: 359

Merit: 250

I'm having a connection problem and i'm sure the port 9911 is open as it's listening:

Code:

$ netstat -anltp | grep "LISTEN"

tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN 972/apache2
tcp 0 0 0.0.0.0:22 0.0.0.0:* LISTEN 820/sshd
tcp 0 0 0.0.0.0:9911 0.0.0.0:* LISTEN 1740/primecoind
tcp6 0 0 :::22 :::* LISTEN 820/sshd
tcp6 0 0 :::9911 :::* LISTEN 1740/primecoind
tcp6 0 0 :::9912 :::* LISTEN 1740/primecoind

And the debug.log is:

Code:

trying connection 78.31.111.116:9911 lastseen=356666.4hrs
connect() failed after select(): Connection refused
trying connection 198.199.84.246:9911 lastseen=356666.4hrs
connect() failed after select(): Connection refused
trying connection 197.96.138.8:9911 lastseen=356666.4hrs

and so on...

This is my node list:

Code:

addnode=208.68.37.41
addnode=129.7.204.114
addnode=198.199.100.118
addnode=87.98.146.72
addnode=88.190.56.58
addnode=78.31.111.116
addnode=198.199.84.246
addnode=197.96.138.8
addnode=96.126.103.94
addnode=192.81.212.113
addnode=198.199.112.141
addnode=192.34.58.127
addnode=31.25.188.219
addnode=54.235.231.242
addnode=175.41.200.152

Removed everything inside .primecoin folder except conf file and restarted but it didnt help.

Trillium

hero member

Activity: 546

Merit: 500

Quote from: belltown on November 07, 2013, 12:48:21 AM

Quote from: Trillium on November 06, 2013, 11:39:45 PM

I have 192 E3-series xeon vCPU going offline today, maybe it will make the diff 0.001 lower for you guys Wink

192!!!! Wow. This does not look like botnet at all Wink

My special botnet that virtualizes the hosts CPU cores without them realizing, infects even and divisible numbers of machines/cores, is limited to spreading inside just one room, and effects only E3 series Xeons. Yes.

ivanlabrie

hero member

Activity: 812

Merit: 1000

Quote from: belltown on November 07, 2013, 12:48:21 AM

Quote from: Trillium on November 06, 2013, 11:39:45 PM

I have 192 E3-series xeon vCPU going offline today, maybe it will make the diff 0.001 lower for you guys Wink

192!!!! Wow. This does not look like botnet at all Wink

Not botnet, probably work pc's.

belltown

sr. member

Activity: 301

Merit: 250

still can't change my profile pic

Quote from: Trillium on November 06, 2013, 11:39:45 PM

I have 192 E3-series xeon vCPU going offline today, maybe it will make the diff 0.001 lower for you guys Wink

192!!!! Wow. This does not look like botnet at all Wink

Trillium

hero member

Activity: 546

Merit: 500

Great discussion it will be interesting to see how this plays out over the next week or two.

Quote

blocks/day = chains/day * (1 - fracDiff + 0.035)

This would be good to have in the FAQ on the OP since a lot of newbies seem to be in denial about how hard it is to mine these days.

I have 192 E3-series xeon vCPU going offline today, maybe it will make the diff 0.001 lower for you guys Wink

mikaelh

sr. member

Activity: 301

Merit: 250

Quote from: Sunny King on November 06, 2013, 02:24:04 AM

The difficulty adjustment works like an oscillator at the stepup/stepdown boundary, so the block spacing target is still maintained. For example it could stabilize at probability(difficulty <=9+255/256) = a, probability(difficulty >= 10) = 1-a. This should allow block spacing target to be maintained at 1 minute. Thus the difficulty jump (up/down) at the boundary has no negative effect on the network.

Thanks Sunny. That's what I suspected would happen.

satriani

member

Activity: 93

Merit: 10

sunny: so the diff will drop when block/min fall (and power of network doesnt change) ?

Sunny King

legendary

Activity: 1205

Merit: 1010

The difficulty adjustment works like an oscillator at the stepup/stepdown boundary, so the block spacing target is still maintained. For example it could stabilize at probability(difficulty <=9+255/256) = a, probability(difficulty >= 10) = 1-a. This should allow block spacing target to be maintained at 1 minute. Thus the difficulty jump (up/down) at the boundary has no negative effect on the network.

mhps

hero member

Activity: 516

Merit: 500

CAT.EX Exchange

Quote from: mikaelh on November 05, 2013, 06:13:27 AM

There will be a jump in the difficulty when difficulty goes to 10.0. That's because none of the 9-chains will qualify for blocks and we have to start looking for 10+-chains. I've actually been working on a paper related to this. Right now it looks like 10.0 will be more difficult than 9.996 will be which means we could get stuck between 9.996 and 10.0 for a while.

So now we know before reaching 10.0 ony 1/256 (0.4%) 9-chains found qualify. At the same time there are ~3% number of 9-chains of 10-chains qualify. So it looks like going over the 10-chain boundary would cause an 0.004/(0.03+0.004) = 10% decrease of chance to find a block.

Of course as mikaelh pointed out this assumes a flat distribution of Ferma-test remainder that is used to calculate the fractional part of the difficulty.

Quote

You can actually see that happening before in my charts:
http://xpm.muuttuja.org/charts/

If you look closely enough, the network block rate seems to have dropped when we went from 8.996 to 9.0. Of course we were using an older version of the mining algorithm back then which probably behaved slightly different.

From the chart the difficulty reached 8.996 at c.a. 15:10 on July 21, and reached 9 less than 1.5 hours later. I guess 10.0 will be reached before December.

cabin

sr. member

Activity: 604

Merit: 250

Quote from: mikaelh on November 05, 2013, 12:23:33 PM

Quote from: cabin on November 05, 2013, 11:15:38 AM

Probability of finding a 10th chain link when you have a 9 chain and are sieving on 9-chains: 1 / ln(2^256) = 0.005635

That's not quite right. The primorial that is embedded in the chain origin actually changes the probability a lot. If you filter out the prime p_i, the probability increases by

1 / (1 - 1/p_i)

If you take a product of those up to p_17 which is 59, you get 7.47493. So that's how many times bigger the probability will be if the number is a multiple of 59#. Also, I usually assume that the numbers are about 300 bits in size. So then the probability for the 10th number being prime becomes:

7.47493 * 1/ln(2^300) = 0.0359468

And if I plug that number into your calculations, I get:

Sieve Size 9:
9-chains per day: 0.54 (reported by miner)
10-chains per day: 0.54 * 0.0359468 = 0.0194113
block per day = 0.54*0.04 + 0.0194113 = 0.0410113

And now size 9 wins by a significant margin.

Interesting..that does change things! So all that sieving really just bumps you from 3.5% chance of a prime to ~9.5%.. not as big a help as I thought.

satriani

member

Activity: 93

Merit: 10

can anybody tell me what happen when we cross difficult 10 at PrimeCoin?
and how much supply will be at PrimeCoin (5-10mln or 20-30mln or 50-100mln -> which one is the most probability?)

mikaelh

sr. member

Activity: 301

Merit: 250

Quote from: cabin on November 05, 2013, 11:15:38 AM

Probability of finding a 10th chain link when you have a 9 chain and are sieving on 9-chains: 1 / ln(2^256) = 0.005635

That's not quite right. The primorial that is embedded in the chain origin actually changes the probability a lot. If you filter out the prime p_i, the probability increases by

1 / (1 - 1/p_i)

If you take a product of those up to p_17 which is 59, you get 7.47493. So that's how many times bigger the probability will be if the number is a multiple of 59#. Also, I usually assume that the numbers are about 300 bits in size. So then the probability for the 10th number being prime becomes:

7.47493 * 1/ln(2^300) = 0.0359468

And if I plug that number into your calculations, I get:

Sieve Size 9:
9-chains per day: 0.54 (reported by miner)
10-chains per day: 0.54 * 0.0359468 = 0.0194113
block per day = 0.54*0.04 + 0.0194113 = 0.0410113

And now size 9 wins by a significant margin.

Topic: [XPM] [ANN] Primecoin High Performance | HP14 released! - page 14. (Read 397587 times)