Topic: [XPM] [ANN] Primecoin High Performance | HP14 released! - page 130. (Read 397647 times)
Why is your account created 2 hours ago ?
First coin found with this miner: http://puu.sh/3CdP6.png 3 minutes ago EST
So he checked the code in and released this an hour ago, and its show you mined that coin yesterday..how is that possible?
Found any blocks?
you think somebody after a mins of the release found a block very easy. i have more than 24 hrs before this release @900pps and not found nothing, give more than 3 hrs for check this
I sure hope this was tested before released.
It's likely folks will still find blocks with it, but without running on testnet for a few hours and comparing the results to another build it won't tell us much. Just reduce the prime sieve size on Sunny's client, boom your PPS goes through the roof.
Well, to be honest, I was slightly concerned about changing the sieve size. I'm not 100% sure about the full effects. What I do know is that when you reduce it to 100,000 you stop getting millions of L1 write misses.
I have found quite a lot of blocks on mainnet with the reduced sieve size. I will try to get some better comparison numbers on testnet.
About 4000 pps with A10.
hmmm
i dont need more pps, i just wanna find more blocks...
i dont need more pps, i just wanna find more blocks...
Error when compile on Ubuntu 12.04
In file included from main.cpp:16:0:
prime.h:10:17: fatal error: gmp.h: No such file or directory
compilation terminated.
make: *** [obj/main.o] Error 1
In file included from main.cpp:16:0:
prime.h:10:17: fatal error: gmp.h: No such file or directory
compilation terminated.
make: *** [obj/main.o] Error 1
You need to install libgmp-dev.
Found any blocks?
you think somebody after a mins of the release found a block very easy. i have more than 24 hrs before this release @900pps and not found nothing, give more than 3 hrs for check this
I sure hope this was tested before released.
It's likely folks will still find blocks with it, but without running on testnet for a few hours and comparing the results to another build it won't tell us much. Just reduce the prime sieve size on Sunny's client, boom your PPS goes through the roof.
First coin found with this miner: http://puu.sh/3CdP6.png 3 minutes ago EST
Found any blocks?
you think somebody after a mins of the release found a block very easy. i have more than 24 hrs before this release @900pps and not found nothing, give more than 3 hrs for check this
I sure hope this was tested before released.
Error when compile on Ubuntu 12.04
In file included from main.cpp:16:0:
prime.h:10:17: fatal error: gmp.h: No such file or directory
compilation terminated.
make: *** [obj/main.o] Error 1
In file included from main.cpp:16:0:
prime.h:10:17: fatal error: gmp.h: No such file or directory
compilation terminated.
make: *** [obj/main.o] Error 1
Installing the libgmp-dev package should fix that.
my semprons make 1000pps now
Found any blocks?
Doesn't reducing the sieve size reduce the likelihood you'll find a solution by the same amount you've increased the primes per second?
Can you compare the number of blocks found with your code versus sunny's code on the test net per some time interval? As sunny has said multiple times, the pps rate is not an effective measure of block finding increase.
I believe that has to do with the fact that the PPS varies widely ... this is a dramatic increase on each cpu, i think when PPS is "generally high" or "generally low" is what sunny meant.
Error when compile on Ubuntu 12.04
In file included from main.cpp:16:0:
prime.h:10:17: fatal error: gmp.h: No such file or directory
compilation terminated.
make: *** [obj/main.o] Error 1
In file included from main.cpp:16:0:
prime.h:10:17: fatal error: gmp.h: No such file or directory
compilation terminated.
make: *** [obj/main.o] Error 1
Found any blocks?
you think somebody after a mins of the release found a block very easy. i have more than 24 hrs before this release @900pps and not found nothing, give more than 3 hrs for check this
Found any blocks?
Any new extra libraries / dependences needed?
http://puu.sh/3CdpH.png <--- from 700 to 10,000 on a Phenom II 1100t
Here is my high performance version of Sunny King's Primecoin client. This version improves mining performance significantly.
Notable changes:
* Use GMP for bignum calculations in the mining threads
* Reduced sieve size to 100,000 so that important arrays fit into L1 cache
* Replaced some bignum calculations with 64-bit arithmetic inside the sieve
* Reduced the amount of memory allocations
Windows x64 binaries
http://sourceforge.net/projects/primecoin-hp/files/0.1.1-hp1/primecoin-0.1.1-hp1-winx64.zip/download
https://mega.co.nz/#!ksEDGYLY!M3orWiIf8_Twipn13oVRcA298EMU_GdxxtVvLZTFMDs
SHA1SUM: ee17da5f3be1cd1ca23ee3df57dcca71f560881b
Windows x86 binaries
http://sourceforge.net/projects/primecoin-hp/files/0.1.1-hp1/primecoin-0.1.1-hp1-winx86.zip/download
https://mega.co.nz/#!E1lXySbC!EIkrCi3olKURb0RXs8wwRifbvBhiXtu4OzT-FLNL8Kk
SHA1SUM: b557bce76691a391b5328d437180403d6f8eeae0
Source code
https://github.com/mikaelh2/primecoin
Linux users please compile it yourself for now.
Some primes per sec numbers (on 64-bit Linux):
~8700 on an Intel Core i7-4770k
~3100 on an Intel Core 2 Q6600
This release includes mustyoshi's patch from this post:
https://bitcointalksearch.org/topic/m.2689981
[/quote
Can you compare the number of blocks found with your code versus sunny's code on the test net per some time interval? As sunny has said multiple times, the pps rate is not an effective measure of block finding increase.
Notable changes:
* Use GMP for bignum calculations in the mining threads
* Reduced sieve size to 100,000 so that important arrays fit into L1 cache
* Replaced some bignum calculations with 64-bit arithmetic inside the sieve
* Reduced the amount of memory allocations
Windows x64 binaries
http://sourceforge.net/projects/primecoin-hp/files/0.1.1-hp1/primecoin-0.1.1-hp1-winx64.zip/download
https://mega.co.nz/#!ksEDGYLY!M3orWiIf8_Twipn13oVRcA298EMU_GdxxtVvLZTFMDs
SHA1SUM: ee17da5f3be1cd1ca23ee3df57dcca71f560881b
Windows x86 binaries
http://sourceforge.net/projects/primecoin-hp/files/0.1.1-hp1/primecoin-0.1.1-hp1-winx86.zip/download
https://mega.co.nz/#!E1lXySbC!EIkrCi3olKURb0RXs8wwRifbvBhiXtu4OzT-FLNL8Kk
SHA1SUM: b557bce76691a391b5328d437180403d6f8eeae0
Source code
https://github.com/mikaelh2/primecoin
Linux users please compile it yourself for now.
Some primes per sec numbers (on 64-bit Linux):
~8700 on an Intel Core i7-4770k
~3100 on an Intel Core 2 Q6600
This release includes mustyoshi's patch from this post:
https://bitcointalksearch.org/topic/m.2689981
[/quote
Can you compare the number of blocks found with your code versus sunny's code on the test net per some time interval? As sunny has said multiple times, the pps rate is not an effective measure of block finding increase.
Giving it a run now
Jump to: