Chemisist
Member
Offline
Activity: 99
Merit: 10
|
|
July 12, 2013, 05:36:49 PM |
|
Did you just use the default flags that are in makefile.unix in the master branch?
|
btc 1ChemaH12nRmd75M8BmPSiqd8x7B2wxFNF ltc LaWX7jgJDyQ2oFaQYJvo5kqC1e1KYPoCfd xpm Ab8NSgxHgGUJvHgSHYqMYBMWai6ZdsA91s
|
|
|
fabrizziop
|
|
July 12, 2013, 05:39:43 PM |
|
Did you just use the default flags that are in makefile.unix in the master branch? Nope, I compiled makefile.release created by bitcoin-qt.pro.
|
|
|
|
jlspartz
|
|
July 12, 2013, 05:43:22 PM |
|
Running at a stable higher rate. 8 threads 30-40% increase
|
|
|
|
azwccc
|
|
July 12, 2013, 05:46:37 PM |
|
my i7-3770 got ~50% increase compared to official 0.1.1 Thank you and ckemisist.
|
Bitrated user: azwccc.
|
|
|
TheSwede75
|
|
July 12, 2013, 05:51:26 PM |
|
AMD 4 and 6-core seeing around 30% from 0.1.1, good job!
|
|
|
|
Kavre
Member
Offline
Activity: 64
Merit: 10
|
|
July 12, 2013, 05:52:48 PM |
|
I am a bit confused
can't get it work properly
if I set setgenerate true -1 (this is random and it should use all core/threads) work like 80pps +-20
with setgenerate true 3 works 300pps
i have 8cpu with 4 cores, which will be the best settings I really tried almost all of it but with no success
thanks
|
For support: - BTC: 143JzbV9RC4r5uLDsQQnVWay6fJfRyYgjS
|
|
|
cosurgi
|
|
July 12, 2013, 05:55:40 PM |
|
any progress replacing CBigNum with gmp ? It is supposed to be like 10x faster...
|
|
|
|
jlspartz
|
|
July 12, 2013, 05:55:48 PM |
|
Running at a stable higher rate. 8 threads 30-40% increase Its a i7 3940xm. Cpu core temps are the same. Seems to drop down to 10% increase gradually and then go back up again. Thanks! Good increase so far with O2
|
|
|
|
xyzzy099
Legendary
Offline
Activity: 1065
Merit: 1077
|
|
July 12, 2013, 06:00:25 PM |
|
Lol, installed fabrizziop's build and mined a block with it inside 20 sec Both fabrizziop and Chemisist will be getting a little tip from me shortly Thanks for your efforts.
|
Libertarians: Diligently plotting to take over the world and leave you alone.
|
|
|
|
Chemisist
Member
Offline
Activity: 99
Merit: 10
|
|
July 12, 2013, 06:06:08 PM |
|
march=native pretty much ensures that it will not be guaranteed to work on any other processor type than what you compiled it with (right?). Which processor did you compile this on?
|
btc 1ChemaH12nRmd75M8BmPSiqd8x7B2wxFNF ltc LaWX7jgJDyQ2oFaQYJvo5kqC1e1KYPoCfd xpm Ab8NSgxHgGUJvHgSHYqMYBMWai6ZdsA91s
|
|
|
romerun
Legendary
Offline
Activity: 1078
Merit: 1002
Bitcoin is new, makes sense to hodl.
|
|
July 12, 2013, 06:09:37 PM |
|
holyshit chemisist thing has ramped my pps up from 150 to 350, one core VPS.
|
|
|
|
Chemisist
Member
Offline
Activity: 99
Merit: 10
|
|
July 12, 2013, 06:10:15 PM |
|
holyshit chemisist thing has ramped my pps up from 150 to 350, one core VPS.
|
btc 1ChemaH12nRmd75M8BmPSiqd8x7B2wxFNF ltc LaWX7jgJDyQ2oFaQYJvo5kqC1e1KYPoCfd xpm Ab8NSgxHgGUJvHgSHYqMYBMWai6ZdsA91s
|
|
|
fabrizziop
|
|
July 12, 2013, 06:10:41 PM |
|
march=native pretty much ensures that it will not be guaranteed to work on any other processor type than what you compiled it with. Which processor did you compile this on? FX 8350, and the native build is working too on a Sempron 145, so that's why I guess it must work on other CPUs too.
|
|
|
|
dudeguy
Member
Offline
Activity: 182
Merit: 10
|
|
July 12, 2013, 06:12:25 PM |
|
i3 3225 at fsb 103 so 3.4 GHz (2 cores 4 threads with hyperthreading). Would you recommend this build or your previous one? BTW you're getting tipped if I push past 700 PPS. Currently at 490-540 PPS on Zalfrin's i5 build.
|
|
|
|
Chemisist
Member
Offline
Activity: 99
Merit: 10
|
|
July 12, 2013, 06:12:45 PM |
|
march=native pretty much ensures that it will not be guaranteed to work on any other processor type than what you compiled it with. Which processor did you compile this on? FX 8350, and the native build is working too on a Sempron 145, so that's why I guess it must work on other CPUs too. I just remember reading that the results may differ on different processors when compiled with the -march=native flag on the gcc page edit: http://gcc.gnu.org/onlinedocs/gcc/i386-and-x86_002d64-Options.html
|
btc 1ChemaH12nRmd75M8BmPSiqd8x7B2wxFNF ltc LaWX7jgJDyQ2oFaQYJvo5kqC1e1KYPoCfd xpm Ab8NSgxHgGUJvHgSHYqMYBMWai6ZdsA91s
|
|
|
jlspartz
|
|
July 12, 2013, 06:16:53 PM |
|
march=native pretty much ensures that it will not be guaranteed to work on any other processor type than what you compiled it with. Which processor did you compile this on? Hmmm working here... seems less stable. Goes from 10% decrease to 35% increase and system is a little laggy. Switching back to O2 win patch for now.
|
|
|
|
Chemisist
Member
Offline
Activity: 99
Merit: 10
|
|
July 12, 2013, 06:18:34 PM Last edit: July 12, 2013, 06:39:28 PM by Chemisist |
|
i3 3225 at fsb 103 so 3.4 GHz (2 cores 4 threads with hyperthreading). Would you recommend this build or your previous one? BTW you're getting tipped if I push past 700 PPS. Currently at 490-540 PPS on Zalfrin's i5 build. Try the -O2 and -O3 builds at my dropbox folder. To see which one runs better on your system: 1. delete the debug.log (rm ~/.primecoin/debug.log) 2. start one of the builds. 3. Let it run for > 10 minutes and grep the debug.log file for "primemeter" (grep primemeter ~/.primecoin/debug.log > primemeter-O3.txt) 4. stop the daemon and repeat 1-3 for the other build. Then you can compare the results of 10 time steps of pps and decide which one works best. (https://www.dropbox.com/home/primecoin%20compiled%20binaries) (I'm uploading an -O2 build in the next 5 minutes.)https://drive.google.com/folderview?id=0ByFxCU5Lp7TBZmRNVm01eU8xOGM&usp=sharing(Thanks @fabrizziop for pointing out the -O2 flag)
|
btc 1ChemaH12nRmd75M8BmPSiqd8x7B2wxFNF ltc LaWX7jgJDyQ2oFaQYJvo5kqC1e1KYPoCfd xpm Ab8NSgxHgGUJvHgSHYqMYBMWai6ZdsA91s
|
|
|
fabrizziop
|
|
July 12, 2013, 06:22:03 PM |
|
i3 3225 at fsb 103 so 3.4 GHz (2 cores 4 threads with hyperthreading). Would you recommend this build or your previous one? BTW you're getting tipped if I push past 700 PPS. Currently at 490-540 PPS on Zalfrin's i5 build. Try the -O2 and -O3 builds at my dropbox folder. To see which one runs better on your system: 1. delete the debug.log (rm ~/.primecoin/debug.log) 2. start one of the builds. 3. Let it run for > 10 minutes and grep the debug.log file for "primemeter" (grep primemeter ~/.primecoin/debug.log > primemeter-O3.txt) 4. stop the daemon and repeat 1-3 for the other build. Then you can compare the results of 10 time steps of pps and decide which one works best. ( https://www.dropbox.com/home/primecoin%20compiled%20binaries) (I'm uploading an -O2 build in the next 5 minutes.) (Thanks @fabrizziop for pointing out the -O2 flag) lol you forgot to share the dropbox folder. In my case the build I made with the best performance is the -O2 -march=native one, the O3 is worse.
|
|
|
|
Kujmandosz
|
|
July 12, 2013, 06:29:52 PM |
|
Can someone link the wallet for mac please
|
|
|
|
|