Topic: RPC Miners (CPU/4way/CUDA/OpenCL) - page 10. (Read 548522 times)

Thanks, that got the wheels turning upstairs.  I had to register a worker first on bitcoin.cz before it would allow me to get work units. I thought that simply signing up for an account was enough. Now I am getting work and hashing roughly 45000 khash/s on my GTX 260.

If you don't see a line that says "Target = " with something other than all ffs, then the miner couldn't connect to the RPC server and couldn't download work.  Check that your arguments are correct, particularly the login info.

After adding the #define _BITCOIN_MINER_CUDA_ or turning on the flag I got a valid .cubin file that rpcminer can load.  However, now I only get 0 khash/s


Any ideas?

I got rpcminer to compile on mac osx, now it needs the ".cubin" file to load.  I'm new to cuda programming, but this is what I have tried so far

nvcc -cubin ../src/cuda/bitcoinminercuda.cu -gencode arch=compute_11,code=sm_11 -o bitcoinminercuda_11.cubin


Any of you experts out there have any ideas what I might be screwing up?

Do you have to have a bitcoin wallet program loaded on the computer to make things work?

I am having this same problem too...  Win 7 64  / 8gbmem / ati HD 5450  rpcminer-opencl.exe   0 khash/s...

I'm getting 0/khashs on the windows binaries you compiled. doesn't matter if it's CPU, CUDA, or screensavers.

Hello All,

I've been working on compiling the rpcminer for mac osx leopard 10.5.8.  I've gotten really far and think I'm almost there, but I'm running into an issue related to boost, I believe.  It seems the function "get" is not resolving.  I'm using macports and I've done a "sudo ports -dv install boost", it seems to be finding the boost headers, because it doesn't complain about those, but it can't resolve the "get" function for some reason.

I would like to hear your 2 cents about the issue if you have some spare time.  I would really appreciate it.  I'm willing to donate some bit coins, but at this point I only have 0.07

How I compile on windows with NVCC?

Aggression would be the main parameter that affects desktop responsiveness.  Set it to 1 and slowly increase it until you find a balance you like.



258.96 is an old driver.  You need a driver that supports at least CUDA 3.2 if you use the supplied binaries.  Alternately, you can compile the source with whatever CUDA toolkit you have available.



As I said before, if you don't see a line that says "Build program rval=0" the kernel didn't get built successfully and you're not going to be doing any hashing.  You can try this kernel.  If that doesn't work you'll have to edit the kernel and find out what's causing it not to build.

I can't seem to get the CUDA miner to work.  It only generates 0/khashs. My command line is as follows

D:\RPCMiner\rpcminer-cuda.exe -gpu -url=http://mining.bitcoin.cz:8332 -user=***** -password=*****

What am I doing wrong?

I have the same problem dduane:

 procedure entry point cuMemFree_v2 Could Not Be Located In The Dynamic Link Library nvcuda.dll 

I'm running Windows 7 64 bit with Nvidia GT 240M

Hello,

I'm having problems with rpcminer-opencl.exe. I'm still getting 0 khash/s even if it connects correctly to the local server.




rpcminer-cpu.exe is working flawless.



My system:
Windows 7 (64 bit)
2x ATI HD 6950 CrossfireX
Catalyst 11.2
ATI Steam 2.3

[12:16]flags8192@Desktop:~/bitcoin-remote-rpc-20110213-src$ cmake CMakeLists.txt
-- The C compiler identification is GNU
-- The CXX compiler identification is GNU
-- Check for working C compiler: /usr/bin/gcc
-- Check for working C compiler: /usr/bin/gcc -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working CXX compiler: /usr/bin/c++
-- Check for working CXX compiler: /usr/bin/c++ -- works
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Boost version: 1.42.0
-- Found the following Boost libraries:
--   date_time
--   filesystem
--   program_options
--   regex
--   system
--   thread
-- Found wxWidgets: TRUE
-- Found OpenSSL: /usr/lib/libssl.so;/usr/lib/libcrypto.so
-- Found BerkeleyDB: /usr/lib/libdb.so
-- Found CUDA: /usr/local/cuda
-- Found CURL: /usr/lib/libcurl.so
-- Configuring done
-- Generating done
-- Build files have been written to: /home/flags8192/bitcoin-remote-rpc-20110213-src
[12:17]flags8192@Desktop:~/bitcoin-remote-rpc-20110213-src$ make
[  1%] Building NVCC (Device) object cmake-bitcoin/./bitcoin_generated_bitcoinminercuda.cu.o
Scanning dependencies of target bitcoin
[  2%] Building CXX object cmake-bitcoin/CMakeFiles/bitcoin.dir/__/src/db.cpp.o
In file included from /usr/local/include/wx-2.9/wx/wx.h:15,
                 from /home/flags8192/bitcoin-remote-rpc-20110213-src/src/headers.h:33,
                 from /home/flags8192/bitcoin-remote-rpc-20110213-src/src/db.cpp:5:
/usr/local/include/wx-2.9/wx/defs.h:49: error: #error "No Target! You should use wx-config program for compilation flags!"

help compile, already the second week of agony: (

Hi Puddinpop

Thanks so much for you brilliant work.

I'm getting the message: 
The procedure entry point cuMemFree_v2 could not be located in the dynamic link library nvcuda.dll

I'm running on a windows 7 professional laptop (HP) which has an NVIDIA GeForce GT 230M in it. 

I downloaded the latest driver from NVIDIA  258.96 and double checked that I didn't need to have a driver update from NVIDIA's website. 

I'm running rpcminer-cuda.exe -url http://mining.bitcoin.cz:8332 -user=hidden -password=hidden -gpu

Thanks for any insight you might have

--Darrell

Heya... First, thanks for a great set of RPC miners!

I have a question about the CUDA miner... I have a crappy GeForce 8400GS, which pushes about 1500kh/s.  Using it, however, brings my system to a complete crawl (almost no CPU load, but it spends 100% of its time trying to redraw windows really really slowly).

So, any way I can throttle the CUDA miner to make my system useable?  If I could have a functional system at 90% performance, I could keep it running all day rather than just at night.  I tried setting the "aggression" lower, but to no effect.

Not complaining - Thanks again for writing this - Just wondering if I might have missed a combination of settings that would help tame it just a bit.  

Thanks!

Not. I tried to make small changes, but it was unsuccessful.

I'm seeing o khash/s as well, with both rpcminer-cpu and rpcminer-cuda. Connecting to host http://mining.bitcoin.cz:8332.

Anyone else seeing this behavior?

Did it build the kernel successfully or not?  You'll just have to find out what in the kernel is causing it not to build on your machine.


You don't need wxWidgest to compile just the rpc miners.  Set the appropriate CMake options to not compile the GUI.


4way uses SSE2 instructions, which may be faster depending on the CPU.  The -cpu miner just uses the standard CPU mining found in the vanilla GUI client.


You need to compile the .cu file targeting the 1.1, 1.2, and 2.0 CUDA architecture, and give the resulting cubin the appropriate name, and place it in the working directory before you start the miner.


Maybe try to #include in minerheaders.h, before serialize.h is included.