Pages:
Author

Topic: [XPM] [ANN] Primecoin High Performance | HP14 released! - page 86. (Read 397616 times)

legendary
Activity: 1063
Merit: 1048
I used your version hp5.  I'm running Windows 8 R2 Standard 64 bit.  The processor is: Intel Xeon CPU E5-2420 @1.90 ghz

My settings are:

maxconnections=256
genproclimit=-1
sievesize=4000000
gen=1


My server processor shows 12 cores running and my desktop shows only 8 cores.  but it's still slower than the desktop by almost half.  Also my server processor cost 4x more than the the desktop processor...

Well, your server shouldn't be lagging behind by that much. Which operating system is the server running? Which version of the client are you using? Did you compile it yourself?

'genproclimit=1' will cause it to only create one thread for mining.  Is that really what you want?

sr. member
Activity: 378
Merit: 255
I have primecoin-qt how dose it find nodes?

It has just sat there for 20 hours?


Did you try this?

https://bitcointalksearch.org/topic/m.2769201
newbie
Activity: 7
Merit: 0
I have primecoin-qt how dose it find nodes?

It has just sat there for 20 hours?
sr. member
Activity: 476
Merit: 250
I've been mining with sievesize=4M for about 22 hours.
I found a block 4 hours after mining with this setting and nothing more for the next 18 hours.

So yes we still find blocks with this setting but it's not a "magical" setting even if the period is too short to draw any solid conclusion...

So I end up not knowing anymore after this test, the random aspect of mining is not helping finding the best sievesize.

It would be great if everyone shared their tests on this variable but i have a feeling some don't want to...

I've never found a block with 4M - only 1M and 2M.  But I think this has to do more with the randomness than anything else..

Did you find more blocks at 1M or at 2M ? And yes the randomness of the process is not helping finding the best settings as I've said earlier.
sr. member
Activity: 378
Merit: 255
Quote
primecoind: checkqueue.h:167: CCheckQueueControl::CCheckQueueControl(CCheckQueue*) [with T = CScriptCheck]: Assertion `pqueue->nTotal == pqueue->nIdle' failed.

I pushed a fix for this crash to bitbucket. If someone wants to try it, it would be appreciated. The crash is quite random.

Looks like this bug has been squashed.
legendary
Activity: 1820
Merit: 1001
Help needed AMD (Piledriver) FX-8350 4.00GHz

Only getting 4500 primes any idea whats going on?
member
Activity: 97
Merit: 10
I used your version hp5.  I'm running Windows 8 R2 Standard 64 bit.  The processor is: Intel Xeon CPU E5-2420 @1.90 ghz

My settings are:

maxconnections=256
genproclimit=-1
sievesize=4000000
gen=1


My server processor shows 12 cores running and my desktop shows only 8 cores.  but it's still slower than the desktop by almost half.  Also my server processor cost 4x more than the the desktop processor...

Well, your server shouldn't be lagging behind by that much. Which operating system is the server running? Which version of the client are you using? Did you compile it yourself?
sr. member
Activity: 301
Merit: 250
My server processor shows 12 cores running and my desktop shows only 8 cores.  but it's still slower than the desktop by almost half.  Also my server processor cost 4x more than the the desktop processor...

Well, your server shouldn't be lagging behind by that much. Which operating system is the server running? Which version of the client are you using? Did you compile it yourself?
member
Activity: 97
Merit: 10
My server processor shows 12 cores running and my desktop shows only 8 cores.  but it's still slower than the desktop by almost half.  Also my server processor cost 4x more than the the desktop processor...

No.

Desktop and server processors are identical. Server Xeons are just rebranded versions of the current era of desktop processors.

Your xeon is probably based on an architecture before the i7.
sr. member
Activity: 406
Merit: 250
No.

Desktop and server processors are identical. Server Xeons are just rebranded versions of the current era of desktop processors.

Your xeon is probably based on an architecture before the i7.
member
Activity: 97
Merit: 10
Hi,

I have a server running a Xeon that Xeon CPU E5-2420 and it only gets 3500 "primespersec"

I also have a I7 3770K processor and that gets over 6500 "primespersec"

Does desktop processors get better performance than server processors when mining XPM?

Is "primespersec" the accurate way to measure performance in mining?
full member
Activity: 122
Merit: 100
I have released Linux binaries for -hp5.

Please note that my binaries were linked against Berkeley DB 4.8.30. Sunny King's original binaries are linked against version 4.8.24 which should be compatible. If you have compiled from source before, your database format may be incompatible with my binaries. Always make sure to backup your database files and your wallet first.

Hey Mikaelh,

Do the

const unsigned int nPrimes = (uint64)nTotalPrimes * 8 / 100;  // down from 10

static const unsigned int nPrimorialHashFactor = 9; // up from 7

changes make sense? I looked at your repo and I did not see either.

On another note how is the gpu miner going? I hope you are getting lots of tips for all the hard work. I'm definitely going to send some your way.
full member
Activity: 201
Merit: 100
I've been mining with sievesize=4M for about 22 hours.
I found a block 4 hours after mining with this setting and nothing more for the next 18 hours.

So yes we still find blocks with this setting but it's not a "magical" setting even if the period is too short to draw any solid conclusion...

So I end up not knowing anymore after this test, the random aspect of mining is not helping finding the best sievesize.

It would be great if everyone shared their tests on this variable but i have a feeling some don't want to...

I've never found a block with 4M - only 1M and 2M.  But I think this has to do more with the randomness than anything else..
full member
Activity: 227
Merit: 100
HP 5 works great, it took my athlon 1055t from 5100pps (hp4) to 5500pps (6cores ftw).

That said, I haven't found anything in 3 days  Cry.  Does changing the -sieve size help?  If so how do I change it using the windows client?

Try using a shortcut or the conf file...shortcut is easier.
Or create a bat file like for cgminer:

primecoin.bat

primecoin-qt.exe -sievesize 2000000

That in the same folder as primecoin-qt is ought to work.

Thank you, how do I check to see if that worked?  I don't see a sieve value when I getinfo or getmininginfo.  

Edit:
From the FAQ:

"Q: How do I change the sieve size?
A: You can either use the config file or command line parameters. Either type "sievesize=1000000" in the config file or add "-sievesize=1000000" to the command line (without the quotes). Replace 1000000 with your desired sieve size. The command line switch should be added to the command that starts the Qt wallet or the Primecoin daemon."

There is no config file in appdata - roaming - primecoin, there is no config file in the qt folder either (x64 client).  
sr. member
Activity: 476
Merit: 250
I've been mining with sievesize=4M for about 22 hours.
I found a block 4 hours after mining with this setting and nothing more for the next 18 hours.

So yes we still find blocks with this setting but it's not a "magical" setting even if the period is too short to draw any solid conclusion...

So I end up not knowing anymore after this test, the random aspect of mining is not helping finding the best sievesize.

It would be great if everyone shared their tests on this variable but i have a feeling some don't want to...
sr. member
Activity: 301
Merit: 250
Quote
primecoind: checkqueue.h:167: CCheckQueueControl::CCheckQueueControl(CCheckQueue*) [with T = CScriptCheck]: Assertion `pqueue->nTotal == pqueue->nIdle' failed.

I pushed a fix for this crash to bitbucket. If someone wants to try it, it would be appreciated. The crash is quite random.
hero member
Activity: 812
Merit: 1000
How do you "install" the binaries in ubuntu server then?

Well, if you really want to "install" it you can copy the binary to /usr/local/bin:
Code:
sudo cp -f primecoind /usr/local/bin/

That only makes a bit easier to use (you can type "primecoind" anywhere instead of "./primecoind" in some specific directory).

Then just create a config file and maybe a script for starting it up.

Thanks a lot man...I'll donate a few yac for the help Smiley
sr. member
Activity: 363
Merit: 250
Sorry if this has already been brought up, but why will certain sievesize's work on testnet (and work very well) yet they crash the client as soon as live mining starts?

I spent 4 hours finding what I thought was the best setting for highest block count and find out these settings crash off the testnet.  Huh
sr. member
Activity: 301
Merit: 250
How do you "install" the binaries in ubuntu server then?

Well, if you really want to "install" it you can copy the binary to /usr/local/bin:
Code:
sudo cp -f primecoind /usr/local/bin/

That only makes a bit easier to use (you can type "primecoind" anywhere instead of "./primecoind" in some specific directory).

Then just create a config file and maybe a script for starting it up.
hero member
Activity: 812
Merit: 1000
HP 5 works great, it took my athlon 1055t from 5100pps (hp4) to 5500pps (6cores ftw).

That said, I haven't found anything in 3 days  Cry.  Does changing the -sieve size help?  If so how do I change it using the windows client?

Try using a shortcut or the conf file...shortcut is easier.
Or create a bat file like for cgminer:

primecoin.bat

primecoin-qt.exe -sievesize 2000000

That in the same folder as primecoin-qt is ought to work.
Pages:
Jump to: