Bitcoin Forum
March 28, 2024, 08:21:07 PM *
News: Latest Bitcoin Core release: 26.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1] 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 ... 1135 »
  Print  
Author Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX]  (Read 3426839 times)
cbuchner1 (OP)
Hero Member
*****
Offline Offline

Activity: 756
Merit: 502


View Profile
April 04, 2013, 05:01:50 PM
Last edit: June 20, 2014, 03:46:49 PM by cbuchner1
Merited by OgNasty (2)
 #1




(cudaminer+ccminer logo by chapmanjw)

Here's my new pet project. I started this during the easter holidays in 2013 and this uses CUDA to squeeze up to 200%  more performance out of nVidia cards - as compared to OpenCL mining applications. Grin ccMiner is a collaborative project by me and a co author also named Christian (call us the C&C Hash Factory, if you will). ccMiner is less polished, but often features new coins+algorithms close to launch date (usually with a clear mining advantage for nVidia).

More information can be found on our dedicated nVidia mining forum http://www.cudaminers.net/

cudaMiner Algorithms:
  • scrypt
  • scrypt-jane
  • scrypt-N
  • keccak
  • blake (github version only)

Installation requirements:
- a recent nVidia driver supporting at least CUDA 5.5
- Visual Studio 2010 SP1 redistributable (redist). Install when MSVCR100.DLL is not found.
  http://www.microsoft.com/de-de/download/details.aspx?id=8328  (x86)
  http://www.microsoft.com/de-de/download/details.aspx?id=13523  (x64)

cudaMiner latest full release:

cudaminer-2014-02-28.zip [32+64bit version] (7.0 MB) speed-up for YAC (compute 3.0 or later), keccak (compute 3.5 or later)
SHA256 sum: 214df7efa386fa4895bf529e72f26463e5469432e6d8d239ee6653809bf072e5
MAC-OS X compiled binaries are found here: http://www.johnchapman.net/cudaminer/  (a third party site maintained by John Chapman)

ccMiner Algorithms:
  • HeavyCoin
  • MjollnirCoin
  • Fugue
  • Groestl
  • Myriad-Groestl
  • Diamond-Groestl
  • JackpotCoin
  • Quark
  • AnimeCoin
  • TalkCoin
  • X11/DarkCoin
  • X13/MaruCoin

ccMiner latest releases:

https://github.com/cbuchner1/ccminer/releases


Have a look at the Google Docs Spreadsheet for configuration and performance data. Please enter new data using this form. We now also have a spreadsheet for scrypt-jane, e.g. Yacoin and the corresponding Data Entry Form. Performance data for scrypt-jane is preliminary, as the feature is under development!

Here is another (somewhat outdated and chaotic) Google Docs spreadsheet with some performance figures and associated configuration settings.

Please carefully inspect the README.txt file before use. Usage is pretty much identical to pooler's cpuminer


previous cudaMiner releases:

cudaminer-2014-02-18.zip [32+64bit version] (7.1 MB) also runs on Maxwell
SHA256 sum: 066e8ffff0de6a3a2d814be4e7fb1f9c59ac6d5378f6710912ccdba61eee00bb

cudaminer-2014-02-09.zip [32+64bit version] (7.1 MB)
SHA256 sum: ef37c97562d98cb95a7a243d8bb378250d9f067b62db239435e6a9975a83a3b8

cudaminer-2014-02-04.zip [32+64bit version] (6.6 MB)
SHA256 sum: 4a38026f662dd06d84f2fe0e5f5098189888de8c0e9947bc24b3cedc4b5f0fdd

cudaminer-2014-02-02.zip [32+64bit version] (6.6 MB)
SHA256 sum: bad402d908862995ec67fd3dddb48a82883176ac7f20cef1f228462f88b04fbb

cudaminer-2013-12-18.zip [32+64bit version] (7.2 MB)
SH256 sum: 4d505804c80bd78fa1c661f74cc5d0e39f92a86f0507abc8ff2aa2b50ffba44b
ATTENTION! Fermi based devices like GTX 560, 570, 580, 590 seem to run quite hot with this release.

cudaminer-2013-12-10.zip [32+64bit version] (10.4 MB)
SHA256 sum: f6a9b1cfcd35867978589c2f36aaef45a16d0f57494777cb14a93366222c195a

cudaminer-2013-12-07.zip [32+64bit version] (9.5 MB)
SHA256 sum: 76dcddcf6d85cbd1ebe4acbb24497bfdae0f3ca9999694c4b152917f4559263a

cudaminer-2013-12-01.zip [32+64bit version] (10.4 MB)
SHA256 sum: dfb4f3a74e534132d397e45aae2f71933a013f557c3be4299e11759c6590b2be

cudaminer-2013-11-20.zip [Update: 32+64bit version] (11.4 MB)
SHA256 sum: 2db068884d0d5683e1b379cf8b4808f55a43b2612547757ce98a8bfe8d2fa0d4

cudaminer-2013-11-15.zip (5.1 MB)
SHA256 sum: 4d4821b0539c24b8882d00caa388e6f7aa8efb2480206e4c9dc2bc95532e3837

cudaminer-2013-11-14.zip (5.4 MB)
SHA256 sum: 5a81f97e183533683373849d73fc30b0b4d287cddb83ef327b0baba006b07c4f

cudaminer-2013-11-01.zip (5.0 MB)
SHA256 sum: 27564fdbc4c41b9d6994a03f8fd2e0a14a1d4a64f0da216b06e8810b604e4ab9

cudaminer-2013-10-10.zip (4.9 MB)
SHA256 sum: 7938965a046b84734daa4332327313b48d198d808a0cf85cb3f0a27e65260c4c

cudaminer-2013-07-13.zip (3.0 MB)
SHA256 sum: d14792ffc8fb5fc910b442d802c480f4f478fa18e2fc95736f525f97a0a9ad52

cudaminer-2013-04-30.zip (3.2 MB)
SHA256 sum: 2d81b52e1051a4f724e75b0e84e231293809437aa060d21b2fd3b8bfc5b711f2

 Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin
If you find this useful, please donate a little. How about the first litecoin you successfully mine with this app? LKS1WDKGED647msBQfLBHV3Ls8sveGncnm      This is my "motivation address".
 Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin Grin

Required dependencies for building on Windows:
pthreads: http://sourceware.org/pthreads-win32/
OpenSSL-Win32: http://slproweb.com/download/Win32OpenSSL-1_0_1e.exe
curl-7.29.0: http://curl.haxx.se/download/curl-7.29.0.zip
or in precompiled form for Visual Studio 2010 SP1 cudaminer_vc2010_prerequisites.7z (49.3 MB)

Linux compilation is also possible now:
chmod +x configure autogen.sh     (.zip does not preserve the x bit)
./autogen.sh && ./configure && make

Better grab the sourcecode from github, as the .zip file contains has Windows style line endings in all ASCII files which you would have to convert first.

Christian
1711657267
Hero Member
*
Offline Offline

Posts: 1711657267

View Profile Personal Message (Offline)

Ignore
1711657267
Reply with quote  #2

1711657267
Report to moderator
1711657267
Hero Member
*
Offline Offline

Posts: 1711657267

View Profile Personal Message (Offline)

Ignore
1711657267
Reply with quote  #2

1711657267
Report to moderator
TalkImg was created especially for hosting images on bitcointalk.org: try it next time you want to post an image
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
1711657267
Hero Member
*
Offline Offline

Posts: 1711657267

View Profile Personal Message (Offline)

Ignore
1711657267
Reply with quote  #2

1711657267
Report to moderator
1711657267
Hero Member
*
Offline Offline

Posts: 1711657267

View Profile Personal Message (Offline)

Ignore
1711657267
Reply with quote  #2

1711657267
Report to moderator
tacotime
Legendary
*
Offline Offline

Activity: 1484
Merit: 1005



View Profile
April 04, 2013, 05:06:02 PM
 #2

Fails to compile on Linux

Code:
cuda-miner.cpp: In function ‘int main(int, char**)’:
cuda-miner.cpp:1029:31: error: ‘_strdup’ was not declared in this scope

Non-standard C++ functions.  I'll try to rewrite it a bit for unix.

Code:
XMR: 44GBHzv6ZyQdJkjqZje6KLZ3xSyN1hBSFAnLP6EAqJtCRVzMzZmeXTC2AHKDS9aEDTRKmo6a6o9r9j86pYfhCWDkKjbtcns
DarkHyudrA
Legendary
*
Offline Offline

Activity: 1386
Merit: 1000


English <-> Portuguese translations


View Profile
April 04, 2013, 05:08:20 PM
 #3

Fails to compile on Linux

Maybe because it says on the OP and in the title that this is for Windows?

English <-> Brazilian Portuguese translations
cbuchner1 (OP)
Hero Member
*****
Offline Offline

Activity: 756
Merit: 502


View Profile
April 04, 2013, 05:12:22 PM
 #4

If you compile on Linux, make sure that on the nvcc command line you specify -arch sm_11 and -maxrregcount=124. Otherwise performance could be poor.

do not let it generate sm_2x/sm_3x code, it leads to high stack utilization (due to register pressure) in the Kepler optimized kernel. By targeting sm_11 you can avoid this, as the JIT compiler in CUDA produces better results than NVCC under register pressure.

tacotime
Legendary
*
Offline Offline

Activity: 1484
Merit: 1005



View Profile
April 04, 2013, 05:14:21 PM
 #5

Fails to compile on Linux
Maybe because it says on the OP and in the title that this is for Windows?

If coded correctly for compatibility, C++ programs are usually cross compatible among unix and windows.  Anyway, compiling on unix is proving problematic, I'm giving up for now.

Code:
XMR: 44GBHzv6ZyQdJkjqZje6KLZ3xSyN1hBSFAnLP6EAqJtCRVzMzZmeXTC2AHKDS9aEDTRKmo6a6o9r9j86pYfhCWDkKjbtcns
DarkHyudrA
Legendary
*
Offline Offline

Activity: 1386
Merit: 1000


English <-> Portuguese translations


View Profile
April 04, 2013, 05:17:33 PM
 #6

Fails to compile on Linux
Maybe because it says on the OP and in the title that this is for Windows?

If coded correctly for compatibility, C++ programs are usually cross compatible among unix and windows.  Anyway, compiling on unix is proving problematic, I'm giving up for now.

It's not a matter of wrong o right, it's a matter of what methods are used to be compatible.

English <-> Brazilian Portuguese translations
dentldir
Sr. Member
****
Offline Offline

Activity: 333
Merit: 250



View Profile
April 04, 2013, 05:21:20 PM
 #7

Confirmed a 660ti on 314.22 drivers getting around 112KH/s.  Driver crashes but recovers when exiting with ctrl-c.  Might just be this machine tho.  Good stuff man.


1DentLdiRMv3dpmpmqWsQev8BUaty9vN3v
cbuchner1 (OP)
Hero Member
*****
Offline Offline

Activity: 756
Merit: 502


View Profile
April 04, 2013, 05:25:00 PM
 #8

Confirmed a 660ti on 314.22 drivers getting around 112KH/s.  Driver crashes but recovers when exiting with ctrl-c.  Might just be this machine tho.  Good stuff man.

with some overclocking, you might get 120kH/sec out of this.

I also have occasional crashes when using ctrl-c. I guess I need to do proper error checking on all the CUDA API calls. It's on my TODO list.

tacotime
Legendary
*
Offline Offline

Activity: 1484
Merit: 1005



View Profile
April 04, 2013, 05:25:42 PM
 #9

It's not a matter of wrong o right, it's a matter of what methods are used to be compatible.

Yeah.  It's not a huge deal, I'm happy to see a nice CUDA miner.  Good job OP!  I will try to add it to GUIminer-scrypt when I get a chance

Code:
XMR: 44GBHzv6ZyQdJkjqZje6KLZ3xSyN1hBSFAnLP6EAqJtCRVzMzZmeXTC2AHKDS9aEDTRKmo6a6o9r9j86pYfhCWDkKjbtcns
cbuchner1 (OP)
Hero Member
*****
Offline Offline

Activity: 756
Merit: 502


View Profile
April 04, 2013, 05:32:12 PM
 #10

Yeah.  It's not a huge deal, I'm happy to see a nice CUDA miner.  Good job OP!  I will try to add it to GUIminer-scrypt when I get a chance

I can see if I can add extra #ifdefs to support both Linux and Windows compilation. I've done this in previous projects. This time I was lazy and preferred to get this out ASAP.
Notanon
Sr. Member
****
Offline Offline

Activity: 388
Merit: 250



View Profile
April 04, 2013, 05:44:34 PM
Last edit: April 04, 2013, 06:18:13 PM by Notanon
 #11

How do I set it to pool mine?

Edit: Never mind, got it running. Bit too used to entering the inputs in CGminer and it's 2am where I am at the moment.

Anyway, crashed both times when it tried to do the first lot of work (happened 10 secs after it got the first batch of data). I'm using a GT 220M, which doesn't seem to have much in the way of support with CUDA, so will try another machine tomorrow and see what happens.
cbuchner1 (OP)
Hero Member
*****
Offline Offline

Activity: 756
Merit: 502


View Profile
April 04, 2013, 06:00:29 PM
 #12

How do I set it to pool mine?

Every decent pool should offer some instructions how to set up pooler-cpuminer.

In my case (Burnside's pool) this looks like:
cudaminer.exe -o http://ltc.kattare.com:9332 -O myname.1:mypassword
wndrbr3d
Hero Member
*****
Offline Offline

Activity: 914
Merit: 500


View Profile
April 04, 2013, 07:22:42 PM
 #13

This is really exciting! I'm stoked to see CUDA finally getting a fair shake on crypto currencies vs. trying to cram OpenCL onto an nVidia card.
TheBigYak
Newbie
*
Offline Offline

Activity: 32
Merit: 0


View Profile
April 04, 2013, 08:55:10 PM
 #14

I get the following output before the program crashes with an Access Violation error:

Command line: cudaminer.exe --url http://localhost:8332/ --userpass user:pass -s 5 -D

       *** CudaMiner for nVidia GPUs by Christian Buchner ***
                     This is version 2013-04-04 (alpha)
        based on pooler-cpuminer 2.2.3 (c) 2010 Jeff Garzik, 2012 pooler
               Cuda additions Copyright 2013 Christian Buchner
           My donation address: LKS1WDKGED647msBQfLBHV3Ls8sveGncnm

[2013-04-04 15:50:55] 1 miner threads started, using 'scrypt' algorithm.
[2013-04-04 15:50:55] Long-polling activated for http://localhost:8332/lp
[2013-04-04 15:50:55] DEBUG: got new work in 6 ms
[2013-04-04 15:50:55] GPU #0: 3024000 khash/sec at 42 blocks, 3 warps


I've tried to activate debug mode, but only get that one listed above.  I've got a build environment on this box, but it's not configured for CUDA development at all.  Any ideas what the problem might be?  If not, I'll see about getting the CUDA SDK installed and building from source so I have PDBs.

cbuchner1 (OP)
Hero Member
*****
Offline Offline

Activity: 756
Merit: 502


View Profile
April 04, 2013, 10:28:47 PM
 #15

How much VRAM is on your Card? What model is it? Do you have Drivers above 304.something? (required for CUDA 5.x)

Christian

trogdorjw73
Hero Member
*****
Offline Offline

Activity: 482
Merit: 500


View Profile WWW
April 04, 2013, 11:25:53 PM
 #16

I installed this on a couple laptops. Some do better than others, naturally.

GT 630M DDR3: ~16KHash/s
GT 640M LE GDDR5: ~17.7KHash/s
GT 650M GDDR5: ~24KHash/s

Combined, the three laptops above are doing 58KHash on GPUs, with another 90KHash/s on the CPUs (two quad-core i6, one dual-core i5 ULV). Not sure if it's all that effective on the GPUs to mine right now, but total power draw is about 150W I'd guess which means I'm making $4 per day and spending around $0.40, so it's definitely a net win. :-D Now to get my GTX 580 system running again...even if it uses 350W it should still get me 150-200KHash/s, so it's a definite win!

cbuchner1 (OP)
Hero Member
*****
Offline Offline

Activity: 756
Merit: 502


View Profile
April 05, 2013, 12:02:56 AM
Last edit: April 05, 2013, 12:13:06 AM by cbuchner1
 #17

I'd guess which means I'm making $4 per day and spending around $0.40, so it's definitely a net win. :-D

Assuming constant LTC value and mining difficulty, you will be earning $1314 per year, but you cannot really use all these laptops while they are mining. But maybe within a year, you can afford a new laptop to actually use, lol Wink

The best GPU I've seen in a laptop so far was a GTX 680M with 1344 CUDA cores. About identical to a 660Ti desktop card, but clocked somewhat lover. You could expect about 100 kHash/sec from it.
cbuchner1 (OP)
Hero Member
*****
Offline Offline

Activity: 756
Merit: 502


View Profile
April 05, 2013, 12:10:17 AM
 #18

Are there any switches to use to up khash/s on these nvidia cards with your program?

No switches yet, this is a really early release. The program has some built in sweet-spot calculations for kernel launch parameters. I intend to add a manual override switch soon

Occasionally this determined sweet spot woult consume more VRAM than your card is equpped with (or it would exceed the memory allocation limitations imposed by the WDDM driver model on Win Vista,7,8). This is why some of you are seeing crashes.
jeezy
Legendary
*
Offline Offline

Activity: 1237
Merit: 1010



View Profile
April 05, 2013, 08:52:29 AM
 #19

cbuchner1: This might be of interest https://bitcointalk.org/index.php?topic=163750.0
goemon888
Sr. Member
****
Offline Offline

Activity: 435
Merit: 260


View Profile WWW
April 05, 2013, 09:13:04 AM
 #20

msi nvidia 570 92 khash/s
Pages: [1] 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 ... 1135 »
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!