Open Source XPM Pool + GPU Miner (aka. madPrimeMiner)

[eXtremal]

OpenCL error: -11 at ..\xpmclient.cpp:888

Try with latest drivers, I use catalyst 14.3 for create this binary.

[XPM-guru]

Don't stop working on solo miner, please. We need solution for solo mining.
I personally can't stand ypool. And claymore miner uses 5x more cpu than yours.

Will start working on solo miner first thing tomorrow. It will be a stripped down version of the pool server, without database, without website, but with Qt GUI.
And you can connect the existing pool miner to it just like it was a pool server.

+1

Waiting for solo miner...

[incognitoworker]

Drivers should be 14.4 or 14.3?
How to aply the patch?

[kev7112001]

Yeah, looks like my fermat test wasn't top of the line 

Was working on the solo miner, but since Claymore is now faster, don't need to bother really.

Let's see if eXtremal can match claymores speed.

thanks for the xpm but u didnt have to give it back

[ozzy1926]

is this project dead like the pool ?

[kev7112001]

is this project dead like the pool ?

lol the miner isnt dead or the pool

[mfqrs3]

Also, fug ypool, mafiapool, they started being shady from day one.

+ 100

[primer10]

Peeps, if you want to make use of existing madmax's client with the improved fermat.cl, you can do the following:

1) mkdir ../gpu (while in the madmax dir where xpmclient is located)
2) copy the new fermat.cl, together with the sha256.cl and sieve.cl into the gpu dir (sha256 and sieve .cl inside xpmclient gpu src)
3) if you are using Catalyst 13.12, you need Catalyst 14's libamdocl64.so. replace or load it using LD_LIBRARY_PATH

[primer10]

Peeps, if you want to make use of existing madmax's client with the improved fermat.cl, you can do the following:

1) mkdir ../gpu (while in the madmax dir where xpmclient is located)
2) copy the new fermat.cl, together with the sha256.cl and sieve.cl into the gpu dir (sha256 and sieve .cl inside xpmclient gpu src)
3) if you are using Catalyst 13.12, you need Catalyst 14's libamdocl64.so. replace or load it using LD_LIBRARY_PATH

Btw, once .cl is compiled into .bin, you do not need to stick with the above steps, including the need to install new Catalyst drivers

I have kernel.bin for 280x and 7950, if anyone is interested.

CREDIT: to eXtremal for improving the fermat.cl, the use of Catalyst 14's libamdocl64.so, among other things Thanks!

[wayla]

/usr/local/include/czmq_prelude.h:498:13: error: conflicting declaration ‘typede
f int SOCKET’
 typedef int SOCKET;
             ^
In file included from netbase.h:11:0,
                 from util.h:33,
                 from bignum.h:12,
                 from main.h:10,
                 from main.hpp:13,
                 from primeserver.h:13,
                 from pool.h:13,
                 from pool.cpp:10:
compat.h:29:15: error: ‘SOCKET’ has a previous declaration as ‘typedef u_int SOC
KET’
 typedef u_int SOCKET;
Any ideas?

I make temporary patch:

Code:

diff --git a/primeserver/src/compat.h b/primeserver/src/compat.h
index 7062216..c407587 100644
--- a/primeserver/src/compat.h
+++ b/primeserver/src/compat.h
@@ -5,6 +5,8 @@
 #ifndef _BITCOIN_COMPAT_H
 #define _BITCOIN_COMPAT_H 1
 
+#include <czmq.h>
+
 #ifdef WIN32
 #define _WIN32_WINNT 0x0501
 #define WIN32_LEAN_AND_MEAN 1
@@ -26,7 +28,7 @@
 #include <ifaddrs.h>
 #endif
 
-typedef u_int SOCKET;
+//typedef u_int SOCKET;
 #ifdef WIN32
 #define MSG_NOSIGNAL        0
 #define MSG_DONTWAIT        0
@@ -42,7 +44,7 @@ typedef int socklen_t;
 #define WSAEINPROGRESS      EINPROGRESS
 #define WSAEADDRINUSE       EADDRINUSE
 #define WSAENOTSOCK         EBADF
-#define INVALID_SOCKET      (SOCKET)(~0)
+//#define INVALID_SOCKET      (SOCKET)(~0)
 #define SOCKET_ERROR        -1
 #endif
 
@@ -58,7 +60,7 @@ inline int myclosesocket(SOCKET& hSocket)
     hSocket = INVALID_SOCKET;
     return ret;
 }
-#define closesocket(s)      myclosesocket(s)
+//#define closesocket(s)      myclosesocket(s)
 
 
 #endif

Thanks, but how to use this?

[primer10]

@eXtremal

Would it be possible to make changes to xpmclminer instead? I find it easier to mod xpmclminer for pool mining than madmax's.

[bctx]

Is it just coincidental, or did Claymore just grab this open source miner, recompile it, and post it on his version 8.0 miner? The CPD is right about what madMax promised(4.8-4.9 CPD at 1150); Caymore's miner was previously running about 2.33 CPD (version 7.0).

I love the fact that madMax open sourced his miner, and I do plan to donate, but if someone else is going to be charging for his work, I think he should give it an extra days worth of work and post the miner up for a pool (with or without a fee). Just a thought.

[primer10]

Is it just coincidental, or did Claymore just grab this open source miner, recompile it, and post it on his version 8.0 miner? The CPD is right about what madMax promised(4.8-4.9 CPD at 1150); Caymore's miner was previously running about 2.33 CPD (version 7.0).

I love the fact that madMax open sourced his miner, and I do plan to donate, but if someone else is going to be charging for his work, I think he should give it an extra days worth of work and post the miner up for a pool (with or without a fee). Just a thought.

As much as I hate to see Windows-only, ypool-only, 10% dev fee (dev fee is fine, 10% is too much. Also, Claymore and primeGPU, have not kept to their words to releasing their source when 2 GPU miners were open sourced. Now why should ppl be held ransom to these 2 buggers?) xpm mafia devs, I do not think it is easy to just grab the code, recompile and post as his 8.0 miner. My suspicion is that Claymore did not release his fastest miner outright at the start.

EDIT: Now remember this impt point, Claymore (so did primeGPU) protected his miner from casual reverse engineering using vmprotect, it is trivial to just "dumb down" his code and you would be non the wiser ..

[bctx]

Is it just coincidental, or did Claymore just grab this open source miner, recompile it, and post it on his version 8.0 miner? The CPD is right about what madMax promised(4.8-4.9 CPD at 1150); Caymore's miner was previously running about 2.33 CPD (version 7.0).

I love the fact that madMax open sourced his miner, and I do plan to donate, but if someone else is going to be charging for his work, I think he should give it an extra days worth of work and post the miner up for a pool (with or without a fee). Just a thought.

As much as I hate to see Windows-only, ypool-only, 10% dev fee (dev fee is fine, 10% is too much. Also, Claymore and primeGPU, have not kept to their words to releasing their source when 2 GPU miners were open sourced. Now why should ppl be held random to these 2 buggers?) xpm mafia devs, I do not think it is easy to just grab the code, recompile and post as his 8.0 miner. My suspicion is that Claymore did not release his fastest miner outright at the start.

He might not have released his fastest miner, but Claymore specialized in optimizing his miner for 290s. To get double the CPD from a 280x using his miner, just a day after this open source miner was released is a little fishy. I wouldn't mind if he improved on it and charged 5% -- but 10%?

[SpeedDemon13]

Is it just coincidental, or did Claymore just grab this open source miner, recompile it, and post it on his version 8.0 miner? The CPD is right about what madMax promised(4.8-4.9 CPD at 1150); Caymore's miner was previously running about 2.33 CPD (version 7.0).

I love the fact that madMax open sourced his miner, and I do plan to donate, but if someone else is going to be charging for his work, I think he should give it an extra days worth of work and post the miner up for a pool (with or without a fee). Just a thought.

As much as I hate to see Windows-only, ypool-only, 10% dev fee (dev fee is fine, 10% is too much. Also, Claymore and primeGPU, have not kept to their words to releasing their source when 2 GPU miners were open sourced. Now why should ppl be held random to these 2 buggers?) xpm mafia devs, I do not think it is easy to just grab the code, recompile and post as his 8.0 miner. My suspicion is that Claymore did not release his fastest miner outright at the start.

He might not have released his fastest miner, but Claymore specialized in optimizing his miner for 290s. To get double the CPD from a 280x using his miner, just a day after this open source miner was released is a little fishy. I wouldn't mind if he improved on it and charged 5% -- but 10%?

I have a hunch that he releases the latest version after the open source version of madMAx, then took what ever code in that and merged it with his. That is just my speculation on it.

[primer10]

Damn ...

If you look at v7 and below, Claymore has always used K<num>.dat where the .dat file is the compiled OpenCL bin. If you see v8, it is replaced by a Data.bin that looks encrypted... Something to hide ?? What do you guys think??

[XPM-guru]

Claymore and ypool, shady stuff.

[ffwong]

Damn ...

If you look at v7 and below, Claymore has always used K<num>.dat where the .dat file is the compiled OpenCL bin. If you see v8, it is replaced by a Data.bin that looks encrypted... Something to hide ?? What do you guys think??

The only way to compete is to make the compilation of the source more easily by preparing ready-to-go various makefile/project files for VC++, mingw, cmake, etc.

It would be necessary to have compiled binaries (better with MD5 checksum) making available most to non-programmer miners.

May be some low-rate pools are needed too. Otherwise, people still go to ypool after successful compilation.

[primer10]

Damn ...

If you look at v7 and below, Claymore has always used K<num>.dat where the .dat file is the compiled OpenCL bin. If you see v8, it is replaced by a Data.bin that looks encrypted... Something to hide ?? What do you guys think??

The only way to compete is to make the compilation of the source more easily by preparing ready-to-go various makefile/project files for VC++, mingw, cmake, etc.

It would be necessary to have compiled binaries (better with MD5 checksum) making available most to non-programmer miners.

May be some low-rate pools are needed too. Otherwise, people still go to ypool after successful compilation.

You are right. Claymore has replied in his thread. There's gonna be shady ppl and shady business anywhere. Ppl just have to decide what's best for them.

[B.T.Coin]

I have kernel.bin for 280x and 7950, if anyone is interested.

Very interested if it's compiled on 13.12
Can you please post a link.