RPC Miners (CPU/4way/CUDA/OpenCL)

[pla]

If i try to run any of the exe's of this program on my Windows XP SP3 machine, I get "this application has failed to start because the application configuration is incorrect", it runs fine on my Windows 7 machine though. Am i missing something? Thanks.

On XP, you need the MSVC80 redistributable to run Puddin's miners:
http://www.microsoft.com/downloads/en/details.aspx?FamilyID=200B2FD9-AE1A-4A14-984D-389C36F85647&displaylang=en.

[1714280491]

1714280491

[1714280491]

1714280491

[1714280491]

1714280491

[yomi]

I just got rpcminer working with cuda as well.  Puddinpop is awesome!  Thanks for making your software opensource and being so patient with people   I'm using the 8600m gt gpu in my macbook pro, osx 10.5.8 leopard.  I'm getting about 3600khash/s with just the gpu.  If anyone needs help with linux or mac, feel free to ask!

Code:

3583 khash/s
3543 khash/s
2011-02-21 04:01:07 Found Hash!
Sending to server: {"method":"getwork","params":["000000011547005097397526c56d1062895102170ae1d8f7a3039112000091d900000000b698719b56b688cad527b7276586e1d59a341455b7dc11fbe91323bc806870804d61e37b1b01cc260290e900000000800000000000000000000000000000000000000000000000000000000000000000000000000000000080020000"],"id":1}
Server sent: {"result": true, "id": "1", "error": null}
3598 khash/s
3559 khash/s

Can you please give details on how you compiled rpcminer and rpcminer-cuda on Mac OS? I am running Snow Leopard and have installed boost and cmake via hombrew, but I keep getting compilation errors.

Thanks.

[NicholasBell]

Is there any particular reason that the CUDA miner would only perform at maximum khash/s if its CLI status window were the foreground window? I'm seeing consistent drops from 16000khash/s to 6000-9000 khash/s when the window is not in the foreground, even if absolutely nothing is happening in whatever window is the foreground (even away from the computer).

System: Windows 7 Enterprise 32 bit, 2.9ghz Core2Duo, Geforce 9600GT, 3gb RAM.

I'm another person that has issues with the CUDA miner speed dropping significantly if it isn't the currently highlighted window (109000 khash/s down to around 16000 khash/s). I really like this miner, but I pay ~100khash/s opportunity cost to actually use my PC. I run the 4way miner at the same time and see no slowdown when it isn't the active window.

I have W7 Professional 64-bit, mining on a gtx 480.

[JollyGreen]

I just got rpcminer working with cuda as well.  Puddinpop is awesome!  Thanks for making your software opensource and being so patient with people   I'm using the 8600m gt gpu in my macbook pro, osx 10.5.8 leopard.  I'm getting about 3600khash/s with just the gpu.  If anyone needs help with linux or mac, feel free to ask!

Code:

3583 khash/s
3543 khash/s
2011-02-21 04:01:07 Found Hash!
Sending to server: {"method":"getwork","params":["000000011547005097397526c56d1062895102170ae1d8f7a3039112000091d900000000b698719b56b688cad527b7276586e1d59a341455b7dc11fbe91323bc806870804d61e37b1b01cc260290e900000000800000000000000000000000000000000000000000000000000000000000000000000000000000000080020000"],"id":1}
Server sent: {"result": true, "id": "1", "error": null}
3598 khash/s
3559 khash/s

Can you please give details on how you compiled rpcminer and rpcminer-cuda on Mac OS? I am running Snow Leopard and have installed boost and cmake via hombrew, but I keep getting compilation errors.

Thanks.

What's your build error?  What cmake command line did you use?

[os008]

On XP, you need the MSVC80 redistributable to run Puddin's miners:
http://www.microsoft.com/downloads/en/details.aspx?FamilyID=200B2FD9-AE1A-4A14-984D-389C36F85647&displaylang=en.

Thank you; that worked. I think i installed the 2010 package before.

[puddinpop]

Update first post with latest release.  The connection to the RPC server will now be kept open if the server supports it instead of opening a new connection for each request.

[slush]

Update first post with latest release.  The connection to the RPC server will now be kept open if the server supports it instead of opening a new connection for each request.

Thanks a lot! All users of this miner & my pool, please update!

[JoeGTN1]

I'm getting compilation errors on Linux:

Code:

OPTION(BITCOIN_ENABLE_CUDA "Enable CUDA miner" ON)
OPTION(BITCOIN_ENABLE_OPENCL "Enable OpenCL miner" OFF)
OPTION(BITCOIN_ENABLE_REMOTE_SERVER "Enable remote miner server" OFF)
OPTION(BITCOIN_BUILD_GUI "Build Server GUI (bitcoin)" OFF)
OPTION(BITCOIN_BUILD_DAEMON "Build Server Daemon (bitcoind)" OFF)
OPTION(BITCOIN_BUILD_REMOTE_MINER "Build remote miner (bitcoinr)" OFF)
OPTION(BITCOIN_BUILD_RPC_MINER "Build RPC miner (rpcminer)" ON)

Code:

/root/z/rpc/src/rpcminer/../minercommon/../serialize.h: In function ‘unsigned int GetSerializeSize(const boost::tuples::tuple<T0, T1, T2>&, int, int)’:
/root/z/rpc/src/rpcminer/../minercommon/../serialize.h:579:31: error: ‘get’ was not declared in this scope
/root/z/rpc/src/rpcminer/../minercommon/../serialize.h: In function ‘void Serialize(Stream&, const boost::tuples::tuple<T0, T1, T2>&, int, int)’:
/root/z/rpc/src/rpcminer/../minercommon/../serialize.h:588:19: error: ‘get’ was not declared in this scope
...

"get" is a pretty hard search term to figure out what templated function I'm looking for.

I'm also not exactly sure which pieces I'm supposed to be building, but with the other "BITCOIN_BUILD" options I get other errors, some with WX (no target set), others with json_spirit (int64 isn't a valid type for some reason, I got it to keep going once by changing them all to int64_t).

Thoughts?  Different place I should be asking?

[JollyGreen]

I'm getting compilation errors on Linux:

Code:

OPTION(BITCOIN_ENABLE_CUDA "Enable CUDA miner" ON)
OPTION(BITCOIN_ENABLE_OPENCL "Enable OpenCL miner" OFF)
OPTION(BITCOIN_ENABLE_REMOTE_SERVER "Enable remote miner server" OFF)
OPTION(BITCOIN_BUILD_GUI "Build Server GUI (bitcoin)" OFF)
OPTION(BITCOIN_BUILD_DAEMON "Build Server Daemon (bitcoind)" OFF)
OPTION(BITCOIN_BUILD_REMOTE_MINER "Build remote miner (bitcoinr)" OFF)
OPTION(BITCOIN_BUILD_RPC_MINER "Build RPC miner (rpcminer)" ON)

Code:

/root/z/rpc/src/rpcminer/../minercommon/../serialize.h: In function ‘unsigned int GetSerializeSize(const boost::tuples::tuple<T0, T1, T2>&, int, int)’:
/root/z/rpc/src/rpcminer/../minercommon/../serialize.h:579:31: error: ‘get’ was not declared in this scope
/root/z/rpc/src/rpcminer/../minercommon/../serialize.h: In function ‘void Serialize(Stream&, const boost::tuples::tuple<T0, T1, T2>&, int, int)’:
/root/z/rpc/src/rpcminer/../minercommon/../serialize.h:588:19: error: ‘get’ was not declared in this scope
...

"get" is a pretty hard search term to figure out what templated function I'm looking for.

I'm also not exactly sure which pieces I'm supposed to be building, but with the other "BITCOIN_BUILD" options I get other errors, some with WX (no target set), others with json_spirit (int64 isn't a valid type for some reason, I got it to keep going once by changing them all to int64_t).

Thoughts?  Different place I should be asking?

I solved this build error by adding "using namespace boost;" just after the boost include files in "serialize.h".

[yomi]

I'm getting compilation errors on Linux:

Code:

OPTION(BITCOIN_ENABLE_CUDA "Enable CUDA miner" ON)
OPTION(BITCOIN_ENABLE_OPENCL "Enable OpenCL miner" OFF)
OPTION(BITCOIN_ENABLE_REMOTE_SERVER "Enable remote miner server" OFF)
OPTION(BITCOIN_BUILD_GUI "Build Server GUI (bitcoin)" OFF)
OPTION(BITCOIN_BUILD_DAEMON "Build Server Daemon (bitcoind)" OFF)
OPTION(BITCOIN_BUILD_REMOTE_MINER "Build remote miner (bitcoinr)" OFF)
OPTION(BITCOIN_BUILD_RPC_MINER "Build RPC miner (rpcminer)" ON)

Code:

/root/z/rpc/src/rpcminer/../minercommon/../serialize.h: In function ‘unsigned int GetSerializeSize(const boost::tuples::tuple<T0, T1, T2>&, int, int)’:
/root/z/rpc/src/rpcminer/../minercommon/../serialize.h:579:31: error: ‘get’ was not declared in this scope
/root/z/rpc/src/rpcminer/../minercommon/../serialize.h: In function ‘void Serialize(Stream&, const boost::tuples::tuple<T0, T1, T2>&, int, int)’:
/root/z/rpc/src/rpcminer/../minercommon/../serialize.h:588:19: error: ‘get’ was not declared in this scope
...

"get" is a pretty hard search term to figure out what templated function I'm looking for.

I'm also not exactly sure which pieces I'm supposed to be building, but with the other "BITCOIN_BUILD" options I get other errors, some with WX (no target set), others with json_spirit (int64 isn't a valid type for some reason, I got it to keep going once by changing them all to int64_t).

Thoughts?  Different place I should be asking?

I solved this build error by adding "using namespace boost;" just after the boost include files in "serialize.h".

Thanks! Adding that line I managed to compile the rpcminer using CUDA.

Now I am stuck with the cubin problem:

Code:

$ ./rpcminer -url=http://mining.bitcoin.cz:8332 -user=xxxx -password=yyyy -gpu
Client will start 1 miner threads
Work will be refreshed every 4000 ms
1 CUDA GPU devices found
Setting CUDA device to first device found
Loading module bitcoinminercuda_11.cubin
Unable to load CUDA module

I also tried to generate the cubin file with:

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

But I still get the same error when I start the miner.

How did you do it? :-)

Thank you.

[JollyGreen]

There are a couple ways around the .cubin issue.

I found if I do the original "cmake -i", eventually it will prompt you asking if you want to generate a cubin file.  You say "yes" and then it builds a cubin file with a ".txt" extension for whatever reason.

If you take the cubin.txt file and copy it to bitcoinminercuda_11.cubin rpcminer with cuda will work

the other option is to run nvcc and generate your own cubin from the ".cu" file

"nvcc ../src/cuda/bitcoinminercuda.cu -gencode arch=compute_10,\"code=sm_10,compute_10\" -gencode arch=compute_11,\"code=sm_11,compute_11\" --keep"

this will generate .cubin files, copy the one you want to bitcoinminercuda_11.cubin and then run rpcminer

[edric]

I'm having a problem getting the RPC Cuda miner set up on a secondary PC on my LAN. I've got it all set up on my main computer and running fine, but trying to run the miner on my other pc is giving me trouble. I've tried with -server=192.168.0.7 -user=xxx -password=xxx -port=8333 -rpcport =8333 and am getting CURL return value = 7
I've also tried with -url=192.168.0.7:8333 with user and pass, I'm getting no errors, but after it loads up CUDA it does absolutely nothing, not even 0 khash. Any insight on what I'm doing wrong?

edit: I figured out how to (supposedly)enable the server to accept rpc connections from other computers but I'm still getting CURL return value = 7. Seems that using rpcallowip=x.x.x.x in the .conf file is being ignored because I still cannot get the rpc miner on my other computer to connect. I've even tried specifying a port (8332) but it does not matter. Am I missing something??


EDIT AGAIN: Fixed. A faulty network cable causing extremely unstable connections. I knew I was missing something, lol. By the way, for some reason it only worked if I used -url but not -server.

[ddod]

So what am I doing wrong?  I downloaded the pre-compiled binaries for windows yesterday.  I have tried them on two machines (both with win7 64).  

Once machine has 2 Radeon HD 5770s (not in crossfire) and I am getting 0 K/hash/s not matter which GPU I try to select (although from the output it appears to only be finding one GPU anyway)

c:\bitminer>rpcminer-opencl.exe -gpu -url=http://localhost:8332 -user=uname -password=pass
Client will start 1 miner threads
Work will be refreshed every 4000 ms
Target = 0000000000012dcd0000000000000000000000000000000000000000000000
1 OpenCL platforms found
1 OpenCL GPU devices found on platform 0
Setting OpenCL device to first device found
Available extensions : cl_khr_global_int32_base_atomics cl_khr_global_int32_exte
nded_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics
 cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_amd_d
evice_attribute_query cl_amd_printf cl_amd_media_ops cl_amd_popcnt cl_khr_d3d10_
sharing
Defining AMDMEDIAOPS
Create context rval=0
Create command queue rval=0
Creating program with source
Building program with options -D AMDMEDIAOPS
0 khash/s

I get similar output with opencl on the other machine which has an Nvidia 8800gts in it.  When trying the CUDA option on that machine I get better results.  However with the default optionsit was getting about 1000khash/s which seemed a little low.  With tweaking (I now have the gpugrid set to 512) it is now getting between 10000 and 15000.  Is that still too low?

Thanks for this tool.  I look forward to getting it to work.

[edric]

So what am I doing wrong?  I downloaded the pre-compiled binaries for windows yesterday.  I have tried them on two machines (both with win7 64).  

Once machine has 2 Radeon HD 5770s (not in crossfire) and I am getting 0 K/hash/s not matter which GPU I try to select (although from the output it appears to only be finding one GPU anyway)

c:\bitminer>rpcminer-opencl.exe -gpu -url=http://localhost:8332 -user=uname -password=pass
Client will start 1 miner threads
Work will be refreshed every 4000 ms
Target = 0000000000012dcd0000000000000000000000000000000000000000000000
1 OpenCL platforms found
1 OpenCL GPU devices found on platform 0
Setting OpenCL device to first device found
Available extensions : cl_khr_global_int32_base_atomics cl_khr_global_int32_exte
nded_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics
 cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_amd_d
evice_attribute_query cl_amd_printf cl_amd_media_ops cl_amd_popcnt cl_khr_d3d10_
sharing
Defining AMDMEDIAOPS
Create context rval=0
Create command queue rval=0
Creating program with source
Building program with options -D AMDMEDIAOPS
0 khash/s

I get similar output with opencl on the other machine which has an Nvidia 8800gts in it.  When trying the CUDA option on that machine I get better results.  However with the default optionsit was getting about 1000khash/s which seemed a little low.  With tweaking (I now have the gpugrid set to 512) it is now getting between 10000 and 15000.  Is that still too low?

Thanks for this tool.  I look forward to getting it to work.

I'm having the same problem with the new build. 0 khash trying to use openCL. On my other machine, it just spams the window with text at warp speed, but that's using the earlier version.

[jkminkov]

hello, I have win7 x64 and Nvidia GTX560 Ti, factory overclocked to 900/1800/2000

last night another application caused conhost.exe to crash and thus ending rpcminer-cuda.exe

the problem is that afterwards your program uses at most 95% of the card, something like 95% throttling, instead of 99%, causing my hash/s to drop from previously average 86700 to 81500

any ideas how to correct that?

[edric]

hello, I have win7 x64 and Nvidia GTX560 Ti, factory overclocked to 900/1800/2000

last night another application caused conhost.exe to crash and thus ending rpcminer-cuda.exe

the problem is that afterwards your program uses at most 95% of the card, something like 95% throttling, instead of 99%, causing my hash/s to drop from previously average 86700 to 81500

any ideas how to correct that?

I've had some sort of mixed results similar to yours, using the same miner. My usual thing is to run it with aggression=6 while I'm at work and my wife is watching movies on the projector. It's not a problem and everything runs well but usage usually sits at %96 and even after VLC and everything is shut off, it stays the same and hashes will sit around 89. When I run my miner, I always set its priority to Above Normal. Sometimes it seems to keep this trend, even after other activities have stopped, of running 96% and giving me low hashes. I've found if I just shut it off, for a few minutes, then start it back up again, it seems to work just fine. Ask me why, I dunno, but it does. It'll go up to 99% just fine. Could possibly be a driver thing.

[jkminkov]

I found that when initial allocation of resources shows something different from 128,256(speed benchmark?), gpu is not used @ 100% happens mostly after using 3D/Cuda resources and launching miner in less than 1 minute

[TurboK]

Hi.
I'm running rpcminer-opencl.exe (20110227), and all is going right... except for one rather nasty problem. If I open any video file (local using MPC-HC with EVR renderer and ffdshow, or online on Youtube), or open Gpu-Z, my computer hangs up completely. This only happens while running the rpcminer.
I'm using a Radeon 6950 with Catalyst 11.2 on Win7 x64, GA-P35-DS3 motherboard and E6550 cpu. Upgrading to 11.4 preview made no difference. Switching from RPC Miner to poclbm made no difference, it still hangs up.

In the past, I had similar problems with another GPU accelerated app on my previous videocard - a Radeon 5850, it hanged up the whole system the exact same way when using ighashgpu + opening any video file. This was later fixed in one of the Catalyst updates.

Overheating etc. cannot be an issue, as I have adequate cooling and a good power supply (corsair hx520w).

Is there anything I can try (some setting in rpc miner), or am I limited to waiting a catalyst update that fixes this issue?

[jkminkov]

for radeon use SDK 2.2
run poclbm with -f120 -v -w128
lower priority of poclbm with task manager to low

if you play youtube clips, expect hashrate to drop by 50%
if still lags, rise value to -f150 or more

[TurboK]

The problem isn't performance. My computer flat-out hangs. No BSOD, no error message, no trace of anything in the event log. The screen will just freeze permanently. Only a hard reset will revive it.

I've tried various -f switches for poclbm-mod, and I've tried Stream SDK 2.1 to 2.4. All of them hang.