[XPM] [ANN] Primecoin Release - First Scientific Computing Cryptocurrency - page 57.

bcp19

hero member

Activity: 532

Merit: 500

Quote from: Aggrophobia on August 01, 2013, 06:18:30 AM

Quote from: bcp19 on August 01, 2013, 05:54:33 AM

Quote from: lucasjkr on July 31, 2013, 10:20:12 PM

I'm not a mathemetician but my understanding is that

A) There's essentially an unlimited number of primes. Even when youre talking numbers hundreds of digits long, yoiu'll still find primes in them pretty regularly (every couple hundred digits or so).

B) Primecoin isn't even trying to find the biggest primes out there - indeed if it was and did, Sonny would qualify for some pretty big reward money

I have a feeling that the biggest primes are safe since the top 10 are all Mersenne Primes found by GIMPS. Primecoin has helped find record Cunningham primes in the 120-160 digit length, GIMPS' record is something like 17 million digits. The next reward from EFF is for a 100,000,000+ digit prime ($100,000) and a 1,000,000,000+ digit prime ($150,000).

means (2^332.xxx.xxx)-1 , to confirm that a number this large is prime with LL would need 2000 hours with a GTX Titan or 3400 hours with an 8-core CPU
now you have to see that the chance to find a Mersenne Prime in that range is 1 : 370000 (many candidates are proven composite with trialfactoring)
so normally it would need ~168.000 years for one person to find a 100,000,000+ digit prime, try your luck Wink

I used 2 cores of an i5 2500k on such a number and it took over 200 days to complete. Using an 8 core CPU would be a huge power waste for very little speed gain. The P95 program is very efficient, so memory bandwidth becomes an issue. In addition, each iteration depends on the one prior, so parallelation is impossible. IIRC, on my 2500k, using a single core would run at 144ms per iteration, 2 cores at 75ms, 3 cores at 60ms and all 4 core 55ms. VERY inefficient with multiple cores.

Stinky_Pete

hero member

Activity: 560

Merit: 500

Primecoin isn't trying to find all primes, even small ones. It is looking for Cunningham chains (and, no, I don't know what they are either)

Aggrophobia

legendary

Activity: 1106

Merit: 1001

Quote from: bcp19 on August 01, 2013, 05:54:33 AM

Quote from: lucasjkr on July 31, 2013, 10:20:12 PM

I'm not a mathemetician but my understanding is that

A) There's essentially an unlimited number of primes. Even when youre talking numbers hundreds of digits long, yoiu'll still find primes in them pretty regularly (every couple hundred digits or so).

B) Primecoin isn't even trying to find the biggest primes out there - indeed if it was and did, Sonny would qualify for some pretty big reward money

I have a feeling that the biggest primes are safe since the top 10 are all Mersenne Primes found by GIMPS. Primecoin has helped find record Cunningham primes in the 120-160 digit length, GIMPS' record is something like 17 million digits. The next reward from EFF is for a 100,000,000+ digit prime ($100,000) and a 1,000,000,000+ digit prime ($150,000).

means (2^332.xxx.xxx)-1 , to confirm that a number this large is prime with LL would need 2000 hours with a GTX Titan or 3400 hours with an 8-core CPU
now you have to see that the chance to find a Mersenne Prime in that range is 1 : 370000 (many candidates are proven composite with trialfactoring)
so normally it would need ~168.000 years for one person to find a 100,000,000+ digit prime, try your luck Wink

bcp19

hero member

Activity: 532

Merit: 500

Quote from: lucasjkr on July 31, 2013, 10:20:12 PM

I'm not a mathemetician but my understanding is that

A) There's essentially an unlimited number of primes. Even when youre talking numbers hundreds of digits long, yoiu'll still find primes in them pretty regularly (every couple hundred digits or so).

B) Primecoin isn't even trying to find the biggest primes out there - indeed if it was and did, Sonny would qualify for some pretty big reward money

I have a feeling that the biggest primes are safe since the top 10 are all Mersenne Primes found by GIMPS. Primecoin has helped find record Cunningham primes in the 120-160 digit length, GIMPS' record is something like 17 million digits. The next reward from EFF is for a 100,000,000+ digit prime ($100,000) and a 1,000,000,000+ digit prime ($150,000).

itod

legendary

Activity: 1974

Merit: 1077

^ Will code for Bitcoins

Quote from: cyberstef on July 31, 2013, 10:01:55 PM

I was wondering if a clone of primecoin will have any scientific value? Is primecoin eventually will find ALL the primes or only some? So it make a primecoin clone valueless? Am I missing something? Thanks.

If all the mankind turn every computer that has ever being built, and ever will be built, to running prmecoin finding primes, and do that till the end of mankind, all primes that would be found would be drop in the ocean, while the rest of the ocean are non-found primes. And even this is overly optimistic, that drop should be divided infinite number of times and only one tiny-tiny part of that drop would be found primes.

notme

legendary

Activity: 1904

Merit: 1002

Quote from: lucasjkr on July 31, 2013, 10:20:12 PM

I'm not a mathemetician but my understanding is that

A) There's essentially an unlimited number of primes. Even when youre talking numbers hundreds of digits long, yoiu'll still find primes in them pretty regularly (every couple hundred digits or so).

B) Primecoin isn't even trying to find the biggest primes out there - indeed if it was and did, Sonny would qualify for some pretty big reward money

Primecoin finds chains of smaller primes because they are easier to verify. Verification needs to be quick for a proof of work to function properly. Verifying primality of very large numbers is hard.

lucasjkr

hero member

Activity: 644

Merit: 500

I'm not a mathemetician but my understanding is that

A) There's essentially an unlimited number of primes. Even when youre talking numbers hundreds of digits long, yoiu'll still find primes in them pretty regularly (every couple hundred digits or so).

B) Primecoin isn't even trying to find the biggest primes out there - indeed if it was and did, Sonny would qualify for some pretty big reward money

cyberstef

newbie

Activity: 49

Merit: 0

I was wondering if a clone of primecoin will have any scientific value? Is primecoin eventually will find ALL the primes or only some? So it make a primecoin clone valueless? Am I missing something? Thanks.

Disclosure: i'm not planning to launch any clone coin, i'm just curious to understand more.

jpmi1

full member

Activity: 214

Merit: 100

Quote from: gigawatt on July 31, 2013, 09:49:26 AM

Quote from: Trillium on July 31, 2013, 08:10:10 AM

Quote from: Lauda on July 30, 2013, 07:01:02 PM

Stuff has settled down in regards to XPM Cheesy

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!

I just went the lazy route. Spent ~48 hours running parameter benchmarks and found some hearty improvements to my average 5-chain/hr.
Its going pretty well though. Mined 5 blocks in the last 24hrs, so I can't really complain.

VPS?

crendore

sr. member

Activity: 363

Merit: 250

Quote from: gigawatt on July 31, 2013, 09:49:26 AM

Quote from: Trillium on July 31, 2013, 08:10:10 AM

Quote from: Lauda on July 30, 2013, 07:01:02 PM

Stuff has settled down in regards to XPM Cheesy

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!

I just went the lazy route. Spent ~48 hours running parameter benchmarks and found some hearty improvements to my average 5-chain/hr.
Its going pretty well though. Mined 5 blocks in the last 24hrs, so I can't really complain.

Mind i if i ask what settings you are using?

gigawatt

full member

Activity: 168

Merit: 100

Quote from: Trillium on July 31, 2013, 08:10:10 AM

Quote from: Lauda on July 30, 2013, 07:01:02 PM

Stuff has settled down in regards to XPM Cheesy

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!

I just went the lazy route. Spent ~48 hours running parameter benchmarks and found some hearty improvements to my average 5-chain/hr.
Its going pretty well though. Mined 5 blocks in the last 24hrs, so I can't really complain.

Trillium

hero member

Activity: 546

Merit: 500

Quote from: Lauda on July 30, 2013, 07:01:02 PM

Stuff has settled down in regards to XPM Cheesy

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!

Lauda

legendary

Activity: 2674

Merit: 2965

Terminated.

Stuff has settled down in regards to XPM Cheesy

Trillium

hero member

Activity: 546

Merit: 500

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

mechs

full member

Activity: 210

Merit: 100

Quote from: lucasjkr on July 30, 2013, 10:15:40 AM

Quote from: GSnak on July 29, 2013, 10:20:07 PM

Quote from: masterOfDisaster on July 29, 2013, 12:35:19 PM

Quote from: kendog77 on July 29, 2013, 12:02:12 PM

[...] so IMHO it's not worth the electricity to mine at this point.

...that's why I turned of my dual HD5850 Bitcoin miner in June 2011 - it was simply not worth investing the electricity for the lousy 0.8 BTC (per day) I could earn by pooled mining these days...
This time I'm going to make other mistakes Wink

When cost to mine>cost to buy, you're better off using cash to purchase it on the market.

It's still far too difficult to get cash into the ecosystem. At least for Americans.

I can use mtgoz, but that incurs wire transfer fees

I can use virwox, but you get hit with many 2.5% commissions (I think there a fee to make initial deposit, another to sell dollars and buy lindens, another to sell lindens and buy BTC, and yet another to withdraw those BTC.

I can use campbx, but I have to wait 30 days before dwolla will let them let me withdraw those btcs.

Or I can use local bitcoins. But everyone there seems to charge 10-20% more than the mtgox price.

That's why people mine even when mining costs more than just buying the coins.

Not true about CampBX. It only 30 days from when you first deposit funds - after those first 30 days, the account is verified and no longer any delays. The only delay is funding Dwolla (sending from Dwolla to CampBX is instant). CampBX is great for smaller purchases, their fee 0.55% - the only problem with them is lack of liquidity. Still always cheaper than both Mt. Gox and Coinbase.

Lauda

legendary

Activity: 2674

Merit: 2965

Terminated.

Quote from: mustyoshi on July 30, 2013, 01:57:30 PM

So the difficulty has been ~9.2x for over a week now.

Where will GPU mining take us?

Who knows, much much higher Cool

mustyoshi

sr. member

Activity: 287

Merit: 250

So the difficulty has been ~9.2x for over a week now.

Where will GPU mining take us?

lucasjkr

hero member

Activity: 644

Merit: 500

Quote from: GSnak on July 29, 2013, 10:20:07 PM

Quote from: masterOfDisaster on July 29, 2013, 12:35:19 PM

Quote from: kendog77 on July 29, 2013, 12:02:12 PM

[...] so IMHO it's not worth the electricity to mine at this point.

...that's why I turned of my dual HD5850 Bitcoin miner in June 2011 - it was simply not worth investing the electricity for the lousy 0.8 BTC (per day) I could earn by pooled mining these days...
This time I'm going to make other mistakes Wink

When cost to mine>cost to buy, you're better off using cash to purchase it on the market.

It's still far too difficult to get cash into the ecosystem. At least for Americans.

I can use mtgoz, but that incurs wire transfer fees

I can use virwox, but you get hit with many 2.5% commissions (I think there a fee to make initial deposit, another to sell dollars and buy lindens, another to sell lindens and buy BTC, and yet another to withdraw those BTC.

I can use campbx, but I have to wait 30 days before dwolla will let them let me withdraw those btcs.

Or I can use local bitcoins. But everyone there seems to charge 10-20% more than the mtgox price.

That's why people mine even when mining costs more than just buying the coins.

GSnak

full member

Activity: 182

Merit: 100

Quote from: masterOfDisaster on July 29, 2013, 12:35:19 PM

Quote from: kendog77 on July 29, 2013, 12:02:12 PM

[...] so IMHO it's not worth the electricity to mine at this point.

...that's why I turned of my dual HD5850 Bitcoin miner in June 2011 - it was simply not worth investing the electricity for the lousy 0.8 BTC (per day) I could earn by pooled mining these days...
This time I'm going to make other mistakes Wink

When cost to mine>cost to buy, you're better off using cash to purchase it on the market.

Tuck Fheman

sr. member

Activity: 363

Merit: 250

Quote from: refer_2_me on July 29, 2013, 03:54:29 PM

Quote from: itod on July 29, 2013, 03:46:20 PM

Quote from: bcp19 on July 29, 2013, 03:38:20 PM

Well, ~5 hours later and up to 4.87 XPM. Rough estimate = 4.3 XPM per day from 10k PPS

Can someone else running similar number of PPS confirm these result? Not that I don't believe it, but the return looks a bit high.

I'm running at about 11-12k PPS and for the last few days i've gotten about 1 XPM (+/- 0.15) per day on ypool.

Try tweaking your settings (if you haven't already). Find the sievesetting that yields the highest difficulty shares on average. A setting of 800000 was my rigs best setting by far.

Topic: [XPM] [ANN] Primecoin Release - First Scientific Computing Cryptocurrency - page 57. (Read 688824 times)