[XPM] Primecoin High Performance Linux Compilation Guide - page 5.

ivanlabrie

hero member

Activity: 812

Merit: 1000

Quote from: bidji29 on July 21, 2013, 01:51:01 PM

If i want to install HP6 on HP5, i just need to do the Step 3, right?

I downloaded the hp6 binaries for linux, did the tar xvjf command, cd into the bin/64 dir and do a sudo cp -f primecoind /usr/local/bin/
That replaces the old primecoind program.
I use the original run-primecoind script I created and erased the hp6 folder...all good.

bidji29

sr. member

Activity: 392

Merit: 250

If i want to install HP6 on HP5, i just need to do the Step 3, right?

mikaelh

sr. member

Activity: 301

Merit: 250

Quote from: maco on July 21, 2013, 06:42:29 AM

This is killing me... i can't shut down primecoind, its always running, and has incorrect rpcuser and password. It doesn't let me edit or attempt to shut down using pkill, primecoind stop.

cd
echo '#!/bin/bash
killall --older-than 10s -q run-primecoind primecoind
function background_loop
while :; do
primecoind >/dev/null 2>&1
sleep 1
done
background_loop &' > run-primecoind
chmod +x run-primecoind

You need to kill the auto-restart script first (that's what my stop script does).

Code:

killall run-primecoind

maco

sr. member

Activity: 294

Merit: 250

This is killing me... i can't shut down primecoind, its always running, and has incorrect rpcuser and password. It doesn't let me edit or attempt to shut down using pkill, primecoind stop.

cd
echo '#!/bin/bash
killall --older-than 10s -q run-primecoind primecoind
function background_loop
while :; do
primecoind >/dev/null 2>&1
sleep 1
done
background_loop &' > run-primecoind
chmod +x run-primecoind

dudeofthestick

member

Activity: 78

Merit: 10

Actually my company bundles OpenStack as a product for Hosters and Services Providers. OpenStack Nova supports snapshotting/cloning by default. Sadly, most companies disable this feature as a vendor-locking strategy. We don't.

We have our own cloud for testing, and one of the applications we run to stress test the cloud is... Mining. Obviously we normally connect to a pool of btc or ltc, something it's too early for xpm.

What I can tell you is most of the cloud providers don't want to host applications like mining rigs because they put all virtual CPUs at %100 for long periods of time. Since they oversubscribe the CPUs a lot, it means that less intensive apps hosted in the same server will have the performance severely affected.

Heavy users are not welcomed in the Clouds. Just like body builders are not welcomed in a gym ;-)

dudeofthestick

member

Activity: 78

Merit: 10

Ok. I have my VM up, running and mining in my Cloud.

Now I want to launch several instances of a snapshot of this VM. Before launching, I should delete the wallet.dat, to start from scratch on each mining node.

But what would happen if I try to mine using the same the old wallet.dat file in all my servers? All mined coins will be assigned to the same address?

Thanks again!

ivanlabrie

hero member

Activity: 812

Merit: 1000

Quote from: rethaw on July 21, 2013, 12:47:37 AM

Quote from: maco on July 20, 2013, 10:26:29 PM

Does anyone know how to delete the transaction id off the screen after you've already exported immature block to main wallet?

and what does genproclimit=-1 do?

Hey maco, remove the wallet.dat to get a new set of keys.

Oh, my lack of sleep made me misunderstand that xD

rethaw

sr. member

Activity: 378

Merit: 255

Quote from: maco on July 20, 2013, 10:26:29 PM

Does anyone know how to delete the transaction id off the screen after you've already exported immature block to main wallet?

and what does genproclimit=-1 do?

Hey maco, remove the wallet.dat to get a new set of keys.

ivanlabrie

hero member

Activity: 812

Merit: 1000

Quote from: maco on July 20, 2013, 10:26:29 PM

Does anyone know how to delete the transaction id off the screen after you've already exported immature block to main wallet?

and what does genproclimit=-1 do?

genproclimit defines the amount of cpu cores used to mine. -1 = all

Hit control + c or pause key

maco

sr. member

Activity: 294

Merit: 250

Does anyone know how to delete the transaction id off the screen after you've already exported immature block to main wallet?

and what does genproclimit=-1 do?

LlamaMaster

newbie

Activity: 51

Merit: 0

Quote from: ivanlabrie on July 19, 2013, 10:15:30 PM

Quote from: LlamaMaster on July 19, 2013, 09:18:47 PM

Is there any way to set the sievesize parameter for the windows client?

Create a conf file, add a line that reads: sievesize=x (replace x by 1000000 or 2000000, or whatever)

I created a primecoin.conf with notepad and pasted that sievesize line in it, but I can't figure out how I am supposed to tell if it's working. I tried putting the file in several different directories (including the one in appdata), but I didn't see any difference when I ran the client.

Also, is there any logic behind choosing between a 1M and 2M sievesize at this point, or is it just speculation?

relm9

hero member

Activity: 840

Merit: 1000

Quote from: rethaw on July 20, 2013, 03:00:01 PM

Has anyone shown increased performance using the locally compiled libgmp?

Using newer libgmp made a difference here, but the one on my server was rather outdated. 4.3.1 I get 3k PPS, with 5.1.2 8k PPS - dual E5-2620

rethaw

sr. member

Activity: 378

Merit: 255

Has anyone shown increased performance using the locally compiled libgmp?

relm9

hero member

Activity: 840

Merit: 1000

Finally fixed my issue. I had to increase the virtual memory ulimit for some reason.

mikaelh

sr. member

Activity: 301

Merit: 250

Quote from: dudeofthestick on July 20, 2013, 05:37:57 AM

I have access to big clusters of virtual servers running on KVM. Clearly the more cores the better, but what about memory, disk, swap and so on?

Thank you very much, the information in this thread is awesome!

Memory requirements:
0.5 GB is enough to run the client currently
Each mining thread probably needs a few megabytes
Memory requirements will increase a bit once the blockchain gets longer

Disk:
Few hundred megabytes for Linux base system and libraries
Few megabytes for the binary
63 MB for the blockchain currently

Moebius327

hero member

Activity: 770

Merit: 500

Thanks for this new release. I just found one block.

I am getting only around 8-9 connections compared to 20-25 with hp2. Is there going to be a problem if I don't add nodes manually?

dudeofthestick

member

Activity: 78

Merit: 10

I have access to big clusters of virtual servers running on KVM. Clearly the more cores the better, but what about memory, disk, swap and so on?

Thank you very much, the information in this thread is awesome!

maco

sr. member

Activity: 294

Merit: 250

Quote from: ivanlabrie on July 20, 2013, 03:47:38 AM

Thanks guys...
I'm starting over.

I may have found your problem as it was a problem of mine too.

The make file was killing me and my swap wasn't working

so I did the following

sudo swapoff /swapfile
sudo rm /swapfile

then tried these commands:

sudo dd if=/dev/zero of=/swapfile bs=64M count=16
sudo mkswap /swapfile
sudo swapon /swapfile

then went into
make -f makefile.unix USE_UPNP=-

mikaelh

sr. member

Activity: 301

Merit: 250

Quote from: dudeofthestick on July 20, 2013, 04:21:51 AM

Can I use a headless 13.04 ubuntu server version? I'm considering running it in an idle platform that uses Ubuntu cloud images

Headless should be fine.

dudeofthestick

member

Activity: 78

Merit: 10

Can I use a headless 13.04 ubuntu server version? I'm considering running it in an idle platform that uses Ubuntu cloud images

Topic: [XPM] Primecoin High Performance Linux Compilation Guide - page 5. (Read 43563 times)