Topic: [ANN][GAP] Gapcoin - Prime Gap Search - New Math Algo - CPU / GPU - Zero Premine - page 103. (Read 287695 times)

tried 8 threads , but the speed went almost 32000. I saw at harware mining comparation section that there are some older CPUs , which give much more speed , because of this I was curious about the speed...

./gapcoind signmessage "GMhhQ2Evh5P6mjoQRHokkz17gakvLjAizC" "Yes thats my address - dcct"
H654Z+DUgc78Q3L/gcg7lHCxiARH3unyEC08tZ6/d868ZZvtHz1P3j/G0k8cijEUbHpa5sRL3pv3xIoL1XonhkM=


I think we will find a lot more high merits soon. 35.4245 is the largest known.

New merit record!

http://coinia.net/gapcoin/blockexplorer.php?q=9ba300fb1ccee419d6c99c9a7b1f3434f5c09124a66dee9d2f1274c5115b2eb4

merit: 32.92211489
gap length: 6396
start prime: 2362114383441832500322011387943037977164194150997773091406607778787608479543714 567191
end prime: 2362114383441832500322011387943037977164194150997773091406607778787608479543714 573587

The i7-4790 has hyperthreading. Try 8 threads.

Note: everyone's hashrate will show lower for now until we have a chance to fix the stats. (Not that they are that accurate anyway)

I didnt get which is the supposed speed with i7-4790 for example?
EDIT: I cant reach above 27000 pps with mine.
command line:

I'll pay someone 200 GAP to set up a GAP p2pool on my VPS, pm me if interested.

Yes the pool is a bit overloaded for the rpc miner, we are working on a stratum server and hope to have it ready in a couple of  days.

We will be raising the pool difficulty shortly to reduce the server load.

i think the pool has to many worker.

whats up with the pool @Nonce ? i keep getting Stales

and this msg: curl_easy_perform()failed: time out was reached

Pool is severely under-reporting my hashrate. I'm running 6 servers with each running at 40,000pp/s for a 240,000pp/s total, being reported as 130,000pp/s on the site. Absolutely ridiculous.

ok.. @dcct running lets see.. gcc 4.9 , latest gmp latest mprf latest mpc... intel opts flags set

sorry not yet

ldd gapminer

libgmp.so.10 => /usr/lib/x86_64-linux-gnu/libgmp.so.10 (0x00007f76be2a4000)
libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6 (0x00007f76bd92e000

linking not right.. brb

brilliant! you da' man  BIG bump in PPS, if you have many instances for every 10 you can gain one server power with this.. nice one mate

:~/GapMiner/bin$ ./run-gapminer
[2014-11-02 14:12:25] Got new target: 22.4545401 @ 22.4545401
[2014-11-02 14:12:35] pps: 0 / 0  10g/h 0 / 0  15g/h 0 / 0
[2014-11-02 14:12:45] pps: 116851 / 116851  10g/h 13359 / 13359  15g/h 0 / 0
[2014-11-02 14:12:55] pps: 117309 / 117079  10g/h 17881 / 15620  15g/h 341 / 170

but you said ~ + 44K PPS hmmm... must be a different processor..

on the record link flags for compiler were:

export PATH=/home/ubuntu/gcc-4.9.2/dist/bin:$PATH

export LD_LIBRARY_PATH=/home/ubuntu/gcc-4.9.2/dist/lib:/home/ubuntu/gcc-4.9.2/dist/lib/gcc/x86_64-unknown-linux-gnu/4.9.2:/home/ubuntu/gmp-6.0.0/dist/lib:/home/ubuntu/mpc-1.0.2/dist/lib:/home/ubuntu/mpfr-3.1.2/dist/lib

make -j32

Makefile:

CXXFLAGS  = -Wall -Wextra -c -Winline -Wformat -Wformat-security \
            -pthread --param max-inline-insns-single=1000 -lm \
                                                -Wno-write-strings \
                                                -I/home/ubuntu/gmp-6.0.0/dist/include \
                                                -I/home/ubuntu/mpfr-3.1.2/dist/include
LDFLAGS   = -lm -lcrypto -lmpfr -lgmp -pthread -lcurl -ljansson -lboost_system \
                                                -L/home/ubuntu/gmp-6.0.0/dist/lib \
                                                -L/home/ubuntu/mpfr-3.1.2/dist/lib
OTFLAGS   = -O3 -march=corei7-avx -mtune=corei7-avx

and execute

export LD_LIBRARY_PATH=/home/ubuntu/gcc-4.9.2/dist/lib64:/home/ubuntu/mpfr-3.1.2/dist/lib:/home/ubuntu/gmp-6.0.0/dist/lib

Also compiled gmp with the following flags:

export CFLAGS="-O3 -march=corei7-avx -mtune=corei7-avx"
export CXXFLAGS="-O3 -march=corei7-avx -mtune=corei7-avx"

I want you to all have extra PPS   see if you can repeat it and achieve maybe the 44K PPS + and report if you can

thanks

no problem appreciate it.. this will be fastest gcc build , its ok.. although I could use 40 extra kilos of PPS  

it truly is however about GPU miner.. I feel we will have it in less that 2 or 3 weeks, it is important it is opened sourced and no freaking bounty or fee.. this so we reach high difficulty all of us very soon..

btw this is slightly easier to read and contains foundational code for fermat decomposition and sieve

https://github.com/eXtremal-ik7/xpmminer

will take a specialist though versed in OpenCL and GAP POW ... above needs RPC for pool, as this would be the best option iif opened sourced if hidden it will be attacked solo and you will see this in the increase of block counts in the rapid fashion (if GPU farm) that would be bad

./gapminer
./gapminer: /usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.15' not found (required by ./gapminer)
./gapminer: /lib64/libc.so.6: version `GLIBC_2.16' not found (required by ./gapminer)
./gapminer: /lib64/libc.so.6: version `GLIBC_2.15' not found (required by ./gapminer)
./gapminer: /lib64/libc.so.6: version `GLIBC_2.17' not found (required by ./gapminer)
./gapminer: /lib64/libc.so.6: version `GLIBC_2.14' not found (required by ./gapminer)

the static miner is btw. not working on any of my Nodes.

I was using 4.8.2. 4.9 was about the same speed.

yes, sudo killall gapminer; make clean;   building libgmp next and wiring, report soonish

shoot.. gcc 4.8 uses older version of gmp/mpfr compiling my own verison quick linking with gmp 6.0.0 and latest mpfr..

might as well gcc 4.9 ... funny I think someone was doing what I am doing now..

building gcc 4.9 with 32 cores ... what I won't do for GAP lol

you did make clean first?
While tune is running you need to stop the miner of cause.

awesome! you probably added few blocks for me just now.. be right back

ok..

-O3 -march=corei7-avx -mtune=corei7-avx

not much effect, optimizing for libgmp next...

[2014-11-02 12:58:17] pps: 108848 / 108592  10g/h 15910 / 13667  15g/h 319 / 160
[2014-11-02 12:58:27] pps: 109088 / 108756  10g/h 9263 / 12198  15g/h 0 / 106
[2014-11-02 12:58:37] pps: 108834 / 108775  10g/h 15621 / 13050  15g/h 0 / 80
[2014-11-02 12:58:42] Got new target: 22.4406137 @ 22.4406137
[2014-11-02 12:58:47] pps: 108882 / 108796  10g/h 14987 / 13437  15g/h 0 / 64

thanks! makes sense, would 1/2 my cost if successful, but from experience and I have done this before with primecoin in the beginning , it will go between 25 to 55 cents and costs pile up... also noticed one is much better off with solo mining at this level of computing as I did parallel test pool/solo , solo comes on the top , blocks are found.. so I don't really understand logic behind 8 mil PPS in the pool as you can "get to blocks" faster in solo and win that way , whatever wining word means here 

Try us-east-1e.

Also there is a beautiful api you can use. If it gets terminated just let it automatically spawn a new one in a different data center. Copy the miner and start it using a automated small script.


Let me guess: Not optimized for avx, using stock libgmp.

-O3 -march=corei7-avx -mtune=corei7-avx

+tune for libgmp.

And you are at ~144K pps