11ch version can found blocks  [GPU 0] T=-1C A=-1% E=0 primes=0.101864 fermat=424887/sec cpd=0.45/day
[GPU 1] T=-1C A=-1% E=0 primes=0.101945 fermat=425209/sec cpd=0.45/day
[GPU 2] T=-1C A=-1% E=0 primes=0.101918 fermat=424364/sec cpd=0.45/day
(ST/INV/DUP): 47x 9ch(1/0/0) 1x 10ch(0/0/0) 1x 11ch(0/0/0)
(ST/INV/DUP): 7x 9ch(0/0/0) 1x 11ch(0/0/0)
[GPU 0] crashed!
[GPU 1] T=-1C A=-1% E=0 primes=0.103769 fermat=295072/sec cpd=0.38/day
[GPU 2] crashed!
[GPU 3] T=-1C A=-1% E=0 primes=0.103595 fermat=278581/sec cpd=0.35/day
[GPU 4] T=-1C A=-1% E=3 primes=0.103670 fermat=301876/sec cpd=0.39/day
(ST/INV/DUP): 7x 9ch(0/0/0) 1x 11ch(0/0/0)
It's last two blocks on pool (534730, 534747). | card | 10ch-old | 10ch-new | 11ch-old | 11ch-new | | 7970(1100/1500) | 4.17 | 4.76 | 0.31-0.32 | 0.31-0.32 | | 7950(1085/1500) | 3.68 | 4.13 | 0.26-0.27 | 0.26-0.27 |
May be i doing something wrong, but accoring to 10-15 min running, new and old 11ch versions have same speed. Same with Radeon R9 290X.. wait a next version Claymore also prepares 11ch version madprimeminer worked well with 13.12 the new kernel does not?
OpenCL compiler from 13.x has a bug, 14.x required for patched kernel. |
|
|
|
i have 68CPD power
the profit calculator says i suppose to mine 55xpm a day with current diff, and through the last day i mined about 40xpm.
11ch 32,27 times less then 10ch (must be 10-12, profit calculator also use same coefficient), original miner not ready for 11ch, I'm working on it. |
|
|
|
Mined only about 2.8 xpm in 13 hours with previous 11ch version ~1.16 cpd (~18.25cpd on new 10ch version) with previous 11ch version. this is normal?
I collect statistics about 11ch version and I think 10ch remains more profitable.. until next optimized 11ch release and difficulty=10,96. Can you post your CPD with previous and current version? |
|
|
|
New patch: https://www.dropbox.com/s/brco83jy9g7mnkx/patch_11_v2.tar.gz (More optimized Fermat test, used 352-bit operations instead of 384-bit) 10ch version, 3xRadeon R9 290X 1040/1250: [GPU 0] T=-1C A=-1% E=0 primes=0.102384 fermat=647998/sec cpd=7.09/day
[GPU 1] T=-1C A=-1% E=0 primes=0.101982 fermat=650073/sec cpd=6.83/day
[GPU 2] T=-1C A=-1% E=0 primes=0.102027 fermat=648890/sec cpd=6.85/day
7 chains/day on R9 290X ! 11ch version: [GPU 0] T=-1C A=-1% E=0 primes=0.102074 fermat=422945/sec cpd=0.46/day
[GPU 1] T=-1C A=-1% E=0 primes=0.101867 fermat=423985/sec cpd=0.45/day
[GPU 2] T=-1C A=-1% E=0 primes=0.101903 fermat=423321/sec cpd=0.45/day
I don't predict what version is more profitable for pool mining (now - may be 10ch, it produces more shares), but difficulty grows very quickly... Are the 10-chains found by the new 11-chain way utilized?
Yes, 11ch version also produces 10ch shares, but less than original version. |
|
|
|
pool now says, 1.19cpd with3x290 is that normal?
Yes, now it's normal, 3xRadeon 290X give 1,33cpd. Now, with current difficulty each 17-18 10-number chain is block, miner searches 11-number chains only 15 times slower.
Hello! what about Nvidia ? Is there a win7 miner ? I have 4 gtx 750 Ti. Can i mine som XPM with those?? Thanks!
NVidia not supported now, sorry. |
|
|
|
How to use this patch?
And is there a different kernel or other for 290 compared to 280x?
Didnt get my 290`s running.
Replace kernel.bin, its same for all AMD GPU's. Catalyst 14.x required. Is anybody working on improving MadMax his code even further to get the same performance as claymore?
I'll try to optimize miner in next 7 days (I already know two optimizations, need time to implement and debug). |
|
|
|
Pool can be unstable now, moving to shares 9ch-11ch from 8ch-10ch. Did you compile an improve version of the madMAx miner? If yes, is it solo and/or pool?
Yes, download standard MadMax miner and patch Patch for miner with target 11ch: <deleted> (Fermat test patch included). Current results: [GPU 0] T=-1C A=-1% E=0 primes=0.101868 fermat=410633/sec cpd=0.43/day
[GPU 1] T=-1C A=-1% E=0 primes=0.102032 fermat=413339/sec cpd=0.45/day
[GPU 2] T=-1C A=-1% E=0 primes=0.101829 fermat=412303/sec cpd=0.43/day
|
|
|
|
overclocking cards helps more cpd? Yes, my 3xRadeon R9 290X @ 1040/1250MHz (memory overclocking has no effect unlike scrypt mining): [GPU 0] T=-1C A=-1% E=0 primes=0.102172 fermat=624595/sec cpd=6.69/day
[GPU 1] T=-1C A=-1% E=0 primes=0.102141 fermat=625096/sec cpd=6.68/day
[GPU 2] T=-1C A=-1% E=0 primes=0.102151 fermat=624178/sec cpd=6.67/day
also it says balance:1.6xpm requested:1xpm. Whats that requested xpm mean? 1 xpm ready to payout.. payouts starts after two days mining. Yes, the modular exponentiation code that was released with this miner is a good start for others to use, but very slow.
I'm wrong about 400% (miner shows 500-600k Fermat tests per second, but miner do some other work, not only Fermat tests ), really 100-120%. |
|
|
|
How often pool pays ? I have allready 55 xpm in requested.
Pool can't pay now, all mined coins unconfirmed.. after 20-24 hours pool can access first mined coins. |
|
|
|
pool down?
Pool not stable (you must know ), today autorestarting script would be created. |
|
|
|
OpenCL error: -11 at ..\xpmclient.cpp:888
Try with latest drivers, I use catalyst 14.3 for create this binary. |
|
|
|
Replacing the old kernel.bin to your version does not work for 280X card  Output ? Pool stats: Workers 107
Latency 149
CPD 354.224 chains/day
107 workers? Hmm... |
|
|
|
You have exceeded the limit of 5 personal messages per hour. Is this only linux or windows too? Both.. |
|
|
|
Miner is here https://bitcointalk.org/index.php?topic=598542.0Patch: https://www.dropbox.com/s/ukfszdh4y17ndkr/xpmminer_patch.tar.gz (replace kernel.bin) For private pool address - send PM (payment after 11XPM mined, if I right). Claymore already take code of sieve algorithm and release new version with 7.0CPD on Radeon R9 290X This miner with patch from xpmclminer shows only 6.6CPD now. Oh I thought you were gonna change xpmclminer instead of madMax's miner... looks like I got it wrong ..
In this case, is it not good enough to use madmax's compiled miner since only .cl file is changed? In other words, no need to build client?
I will change xpmclminer for speedup and supporting madmax's pool. It's more good because xpmclminer support Radeon HD5xxx/6xxx and NVidia GPUs. |
|
|
|
There will be a cold day in hell before I manage to do that.....  Or another variant, I can send address of my pool and you join it |
|
|
|
is this in usable form for ypool or beeeer or solo? or am I in a hurry?
Only private pool mining available, you must build pool and client from source. |
|
|
|
Patch: https://www.dropbox.com/s/ukfszdh4y17ndkr/xpmminer_patch.tar.gzBefore (one gpu crashed, it's unstable): (ST/INV/DUP): 142x 8ch(4/0/0) 15x 9ch(0/0/0) 6x 10ch(0/0/0)
Work received: height=535123 diff=10.903917 latency=91ms
[GPU 0] T=-1C A=-1% E=1 primes=0.103852 fermat=407245/sec cpd=5.13/day
[GPU 1] crashed!
[GPU 2] T=-1C A=-1% E=2 primes=0.103874 fermat=421410/sec cpd=5.32/day
[GPU 3] T=-1C A=-1% E=0 primes=0.103761 fermat=409549/sec cpd=5.12/day
[GPU 4] T=-1C A=-1% E=0 primes=0.103880 fermat=408327/sec cpd=5.16/day
after: Work received: height=535131 diff=10.903908 latency=87ms
[GPU 0] T=-1C A=-1% E=0 primes=0.103765 fermat=443118/sec cpd=5.54/day
[GPU 1] T=-1C A=-1% E=3 primes=0.103738 fermat=443034/sec cpd=5.53/day
[GPU 2] T=-1C A=-1% E=0 primes=0.103840 fermat=453146/sec cpd=5.71/day
[GPU 3] T=-1C A=-1% E=0 primes=0.103844 fermat=442998/sec cpd=5.58/day
[GPU 4] T=-1C A=-1% E=0 primes=0.103471 fermat=443800/sec cpd=5.39/day
(ST/INV/DUP): 4x 8ch(0/0/0) 2x 9ch(0/0/0) 1x 10ch(0/0/0)
5xRadeon R9 290 (non-X), primorial 12. It's not final result |
|
|
|
/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: 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
|
|
|
|
This is on a 290X correct? Yes. I changed weave depth and got this results: [GPU 0] T=-1C A=-1% E=0 primes=0.102074 fermat=589708/sec cpd=6.26/day
(ST/INV/DUP): 9x 8ch(0/0/0) 1x 10ch(0/0/0)
Fermat tests number increased, CPD - no.. need more time to found right configuration. I will release something after 7CPD reaching on R9 290X. |
|
|
|
|
Show Posts
