Package Details: xmr-stak-git r1457.67cf951-1

Git Clone URL: (read-only, click to copy)
Package Base: xmr-stak-git
Description: Unified All-in-one Monero miner
Upstream URL:
Licenses: GPL3
Conflicts: xmr-stak
Provides: xmr-stak
Submitter: huiyiqun
Maintainer: a36233
Last Packager: a36233
Votes: 16
Popularity: 0.000000
First Submitted: 2017-11-22 03:03 (UTC)
Last Updated: 2020-04-20 10:13 (UTC)

Latest Comments

a36233 commented on 2019-03-28 16:45 (UTC)

please change cmake to: cmake -DXMR-STAK_COMPILE=generic

in order to compile on different machine from the one running it

pentix commented on 2018-07-09 10:38 (UTC)

Building gcc6 from the AUR currently failes due to svn connection problems. I fixed the build by adding 'gcc7' to the makedepends and adjusting the paths of CC and CXX env vars.

severach commented on 2018-05-23 02:41 (UTC)


huiyiqun commented on 2018-04-16 00:55 (UTC)

@OriginCode This is a git version instead of a stable version, it will be updated when commit is pushed to master branch of github. NEVER marked it as output of dated when new version is released. It's always latest.

If you'd like to update the package and you are using yaourt, you can:

yaourt -Syua --devel

OriginCode commented on 2018-04-15 12:41 (UTC)

Pls update the pkgver.

mivaho commented on 2018-04-02 14:40 (UTC)

Any plans on updating?

RemoteAdmin commented on 2018-02-14 17:51 (UTC)

@huiyiqun I'm suggesting to split the package. One with and one without the no-donate patch. Either both inside of one PKGBUILD (which would double compiling) or two separate.

It would make installing this package a lot easier, as you don't have to remember to choose at each rebuild

huiyiqun commented on 2018-01-02 02:13 (UTC)

@dougvj Actually, gcc6 is depended by cuda.

dougvj commented on 2017-12-31 23:49 (UTC)

Please add gcc6 to build dependencies

wangbx_arch commented on 2017-12-16 06:08 (UTC)

Got the following error: /tmp/yaourt-tmp-[user]/aur-xmr-stak-git/src/xmr-stak/xmrstak/backend/nvidia/nvcc_code/ fatal error: device_functions.hpp: No such file or directory

Renaming should be able to fix it, similar to:


GRBurst commented on 2017-12-11 22:43 (UTC)

@dominicm As you mentioned, the line starts with an "#", meaning that it is commented out ;-) To enable the patch, you have to remove the leading # and correct the path the path to ../no-donate.patch

huiyiqun commented on 2017-12-08 04:53 (UTC)

If this patch is enabled, then you can see

Configurable dev donation level is set to 0.0%

when you running it.

If you'd like to enable it automatically, you could try customizepkg.

huiyiqun commented on 2017-12-08 04:51 (UTC)

@noraj cuda require gcc6 yet.

If you'd like to use gcc7 and do not need cuda, you should try instead.

noraj commented on 2017-12-08 00:42 (UTC)

Why forcing to use gcc6 and not gcc7 ?

dominicm commented on 2017-12-07 16:01 (UTC)

@GRBurst I though it was included by default, I even saw a boost in hashrate vs git install not to mention that the thank you note dissapeared from web UI. How can I verify if it's compiled with the patch? Also the line is #patch -p1 -i ../../no-donate.patch so unless I am missing something you also need to uncomment it.

dominicm commented on 2017-12-07 15:57 (UTC)

@arcmags Why not include a systemd service with this package? This way it could run as its own dedicated system user and set the proper config location as default. If there is need to run as user, systemd user services can be included too or just run an the executable from /bin like now.

GRBurst commented on 2017-12-07 09:53 (UTC)

Quick info: In order to use the no-donate.patch, you have to change the path from "../../no-donate.patch" to "../no-donate.patch" in the pkgconfig.

mags commented on 2017-12-06 22:23 (UTC)

@dominicm I agree it's not an ideal design, but that's how upstream does it. You could fork a repo, make changes, and build your own package; or you could write a simple wrapper script to switch to some chosen config directory with each launch...

dominicm commented on 2017-12-06 16:48 (UTC) (edited on 2017-12-06 17:06 (UTC) by dominicm)

That seems like a really bad design decision to me. Current directory will obviously change causing issues so why not use the standard config file directories like all the other packages do? I really do not understand why you wouldn't put the config file in the config directory like /etc or seeing as it isn't a system service, ~/.config seems like a good place.

huiyiqun commented on 2017-12-05 12:46 (UTC)

Actually, this package save config to CURRENT DIRECTORY and try to load config from CURRENT DIRECTORY. In other word, if you do not want to answer the questions every time, you should execute xmr-stak at same working directory.

dominicm commented on 2017-12-05 12:42 (UTC)

This package doesn't appear to save the config file anywhere as it asks the same config questions every launch. Am I missing something here or is the package just incomplete? Where is the config file saved? I know I can specify the config location but it makes the initial setup wizard useless then.

mags commented on 2017-11-30 00:24 (UTC) (edited on 2017-11-30 20:59 (UTC) by mags)

To compile this on my minimal installation I had to install ocl-icd. It appears to not be installed by any current dependency.

huiyiqun commented on 2017-11-26 08:32 (UTC)

You can use following repo instead:

whompy commented on 2017-11-25 22:28 (UTC)

@huiyiqun It's not too hard to remove CUDA support. I build with: cmake . -DCUDA_ENABLE=OFF and remove the install line for the CUDA libraries. It could probably be handled with a define near the top of the PKGBUILD, but I'm lazy.

huiyiqun commented on 2017-11-22 16:25 (UTC)

@rigon it seems libclc is not required. I have added opencl-headers. Thanks. I have no idea how to add an option to build with/without support of CUDA and OpenCL except that building multiple packages. Any suggestion here?

rigon commented on 2017-11-22 12:41 (UTC)

Please add dependencies libclc and opencl-headers, they are required to build with OpenCL. Consider also add options to build with/without support of CUDA and OpenCL.

Smasher816 commented on 2017-11-22 07:17 (UTC)

Good timing. I was in the middle of making this PKGBUILD and about to upload it. You might change the description to mention that it contains both cpu and gpu (nvidia+amd) miners