Topic: [ANN] [XMG] MAGI | CPU mining | mPoW | mPoS | [MagiPay] - page 1140. (Read 2375926 times)

If you are having a specific problem then please post here and I will try to sort it out. Read the read.me file first, then run Setup.bat which does all the setting-up for you according to your individual requirements. You only have to provide a few bits of information such as your pool login name, pool worker name and password. The questions asked by Setup.bat are mostly self-explanatory. As a quick confidence check you can just run automine.bat and see if that works straight out of the box without running Setup.bat, which it should.

I honestly do not have the time right now to write up a full tutorial

Spexx can you guys a tutorial how to config / setup your miner ?

Tor nodes were stuck, checking!

No help. That wallet does not work!

Alright! here's the thing .. As I am the winner of the first real round on http://lotto.dhs.org/magilotto and I didn't want to.. I'll give an extra 25 XMG to the 4 next round winners. (game id 5,6,7 and 8.)
those rounds has to last at least one hour and have 5 players. (don't want a single person to buy the 100 ticket and then receive 120 XMG..)

the extra 25 XMG will be sent manually by me so don't worry if I don't send it rightaway after the round.

Hang on for a bit. I am working on a newer 32 bit version too, but first I will have to build a new 32 bit compiler with my custom optimizations which takes time.

Really great thanks for accepting XMG.
 

I will accept MAGI for my hot sauce (https://www.pexpeppers.com) just choose your sauce(s) and send me an email to [email protected]

I'm not one of these people ... I have a 32-bit miner for other's 32bit-OS   

Don't forget!

And here I thought you owned a hat...
No worries mate, glad to help out.

Yea,  those 5 extra kh/s does do a difference.

And folks, don't forget to update your miners!

Edit: added the hole quote since i got a new page in this thread

Dev or someone can you help this member with this problem?

Windows 64 bit. Updated minerd software now available!

You lucky people I am claiming a further 3 percent increase in hashrate with this one. This makes it over 10 percent faster than my first minerd release for the new algorithm.

My thanks to ex33s who inadvertently pointed out an alternative source for the minerd code at https://noncepool/m7magi-cpuminer-v2 and has been sent a bounty of 25 XMG as a reward. My challenge and 100 XMG bounty still stands for anybody else who can compile a faster minerd. Please see https://bitcointalksearch.org/topic/m.9412187 for details.

Download links as before.

Full package:-

.rar https://www.dropbox.com/s/hiadf8o5u7bsit7/XMGpoolminerWin64.rar?dl=0

.zip https://www.dropbox.com/s/6hmlajdp0qgmcfi/XMGpoolminerWin64.zip?dl=0

self-extracting .exe (extracts to C:\XMGminer by default) https://www.dropbox.com/s/dw7yjssjrrqp0m1/XMGpoolminerWin64.exe?dl=0

Minerd binaries only:-

.rar https://www.dropbox.com/s/q8puq9w6uc240mx/Spexx-Win64-generic-XMGminerd.rar?dl=0

.zip https://www.dropbox.com/s/o8i726sxnwe2wxa/Spexx-Win64-generic-XMGminerd.zip?dl=0

self-extracting .exe (extracts to C:\XMGminer by default) https://www.dropbox.com/s/uudj2npeyixesjb/Spexx-Win64-generic-XMGminerd.exe?dl=0

Happy Hashing

PS: Message for ex33s - I cheated because I don't actually own a hat Your comparison on Ubuntu was very interesting too - a remarkable difference. Thanks again.

Thanks for the 25 xmg tip, glad to hear that you managed to get better performance, saved 75 xmg but i still want to see you eat your hat as you won your own bet

But for a comparison with the miner you used but on my system, compiled the same way as my previous post.
http://i.imgur.com/Hdu6q7Z.png

I am so glad that you did! I was using the source at https://github.com/noncepool/m7m-cpuminer-v2 instead of the one you used at https://github.com/noncepool/m7magi-cpuminer-v2 of which I was unaware. I have squeezed another 3 percent higher hashrate by using the latter source and have therefore just won my own prize of 100 XMG (not strictly according to the rules, but hey-ho).

I will therefore update my challenge to refer to this latter source code - that is fair. Also, to be fair, I will send a bounty of 25 XMG to ex33s for the tip (thanks mate), which probably saved me 100 XMG for the time being at least.

My latest minerd will be available on the same links as before in the next hour or so.

I noticed that the Suprnova Dashboard hashrate figure was all wrong too. The figure quoted when you look at your account worker(s) is correct though - so is the hashrate figure reported by the miner when shares are accepted.

The figure compared to the original algo would be closer to divide by 2.9 by my reckoning. However, the CPU Passmark ratings are only a guide and some actual results with the miner may be disappointing. The miner is a maths-intensive CPU-bound process which is not necessarily the type of benchmark used to gain the Passmark ratings.

I noticed in particular that laptop and notebook CPUs would run about 20 percent slower than the Passmark ratings would suggest, which is probably something to do with CPU throttling to avoid overheating but that is only a guess. I have not managed to fry any CPUs yet but they will certainly run hot. Those people who are inclined to overclock their CPUs would be well-advised to keep an eye on temperatures and to make sure their heatsinks, fans and airways are cleaned frequently.

When compiling the miner code for Windows x86_64-w64-mingw32 (gcc version 4.9.1), I noticed that experimenting with CFLAGS etc made no difference at all in most cases. Code compiled with -mtune=generic differed from code compiled with -march=native (or -march=anything else which is valid) in only 7 bytes which were clearly some sort of bitmask. The actual hashrates obtained were identical and ran on any 64 bit CPU no matter what -march was used, which was a bit of a surprise really.

The only CFLAG which made any measurable difference was -fomit-frame-pointer which gave a 0.06 percent increase in hashrate - right on the limit of measurement uncertainty.

Running the miner as multiple instances/single thread as opposed to single instance/multiple threads gives a speed advantage of around 0.5 percent. Not a lot, but measurable.

I updated wallet to v1.1.0.2-magi-tor, found the wallet can not be synchronized, and I want to know how to solve

Actually, I'm getting higher then that (Running on all 8 cores).

My CPU (/proc/cpuinfo)

Checking the site https://www.cpubenchmark.net/cpu_list.php (That Spexx listed) is giving me a benchmark of 9390.

So 9390 / 1000 * 3.6 = 33.804 kh/s for my cpu

Here's what I'm getting with my miner running on ubuntu 13.04 (x64) on all 8 cores, 35.65 kh/s as lowest and 36.03 kh/s as highest. See screenshot link below
http://i.imgur.com/rFTHeB2.png

I get an average rate of 35.952666667 kh/s
(36.02+36.02+36.02+36.03+35.98+36.02+35.93+35.93+35.92+35.97+35.96+35.97+35.95+35.92+35.65) / 15 =  35.952666667

Witch means i get a multiplier of 3.828824991

I've compiled the miner with gcc version 4.7.3 and used CFLAGS = -march=native -mtune=native -Ofast and CXXFLAGS = -Ofast -pipe -march=corei7-avx

Miner being used is https://github.com/noncepool/m7magi-cpuminer-v2

And please remember that using your computer while your mining will decrease the kh/s you are getting. When i'm running Spotify and Chrome at the same time as the miner i'm getting around 30 kh/s instead of ~35.95 kh/s

Edit: Added link to the miner i'm using

After the algo change, the hashes are way less than a division by 2.5
Also the hashes reported by the miner are different from the hashes reported in suprnova dashboard.
The miner says it is around 28 khash most of the time, the pool dashboard changes from 18 to 25 most of the time.