Return to Level1Techs.com

The AMDGPU Info/News Thread

linux
helpdesk

#304

True AMDGPU is meant to be the future kernel 4.8, 4.9 and on.


#305

yeah that makes sense. ok, im gonna do that and document it every step. i really wanna get this working and have a guide ready so people can install that driver more easily.


#306

that is the biggest problem. there are a lot of different things you have to install before you can mess with the amd driver. the correct kernel is one of them and in that special case some say so some say other. 4.4 vs. 4.8-??.

but we'll get there.


#307

When the open stack becomes fully featured it will not even be an issue any more. Newer kernel newer driver and it might not even need the PRO to get good performance. And i am sure that AMD will eventually streamline the PRO driver when that becomes fully featured.


#308

Didn't apple use the M275X in something recently? I thought they had used it... I managed to run El Capitan on my laptop and in my mac pro it uses AMDGPU to run my 370 when its in there. But on my y40-70 it doesn't know what it is.... Then again I guess it IS basically a 250X in a laptop so I don't expect too much from it....

Hmmm maybe thin needs some digging if I am going to do this.


#309

maybe i missed it but what are you going to do again? hackintosh with amd?

and btw: newer macs will use amd graphics inside the intel chip.


#310

Yeah OSX El Cap or Sierra on Lenovo y40-70
No KB or mouse control so thats to be done first, then I have to do wifi. Then sound and GPU last.


#311

a little upgrade since over the holidays and new years and such I took some time off of Linux and PC stuff in general.

I reinstalled the AMD GPU-pro driver und did every fix, still no improvements.

I'm gonna try it this again with kernel 4.7 but I don't think that that will work.

Could we make a list what is needed to support the driver?

You need a 'new' kernel 4.7 - 4.X
a card that is compatible,
Ubuntu, Arch, Mint, seem to work (mostly).

I sorta feel the urge to move to arch and try again because I don't see anyway I can get it to work on ubuntu. while writing this post I searched for hours again for some fixes and I have tried everything from reinstalling, different kernel versions video group user, dkms-reconfiguring (still gives errors in 4.8 & 4.7), driver install with overwriting etc.

I really don't mind the whole fiddling but I don't get where the fudge the error is and how to attack that error. that is very unsatisfactory.

the last straw I'm thinking about right now is something with the x11|xorg|xserver thing. is there some option that works against the driver or stops it from operating?

edit:

[email protected]:~$ lspci -k | grep -EA2 'VGA|3D'
01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Device 67df (rev c7)
	Subsystem: XFX Pine Group Inc. Device 9480
	Kernel driver in use: amdgpu

and

Screenfetch
GPU: AMD Radeon (TM) RX 480 Graphics

so the driver is in use and the card is detected but i still cannot fix tearing i cannot set xrandr to 75hz (crashes and needs a reboot) and i cannot play my beloved rocket league


#312
 sudo dpkg-reconfigure amdgpu-pro-dkms

------------------------------
Deleting module version: 16.40-348864
completely from the DKMS tree.
------------------------------
Done.
Loading new amdgpu-pro-16.40-348864 DKMS files...
First Installation: checking all kernels...
Building for 4.7.0-040700-generic and 4.8.0-040800-generic
Building for architecture x86_64
Building initial module for 4.7.0-040700-generic
ERROR (dkms apport): kernel package linux-headers-4.7.0-040700-generic is not supported
Error! Bad return status for module build on kernel: 4.7.0-040700-generic (x86_64)
Consult /var/lib/dkms/amdgpu-pro/16.40-348864/build/make.log for more information.

doenst work.


#313

.....I would be using Kernel 4.9 :\


#314

ok, then im gonna do it again with 4.9

so im running 4.9 now and in the installation of the driver

Building only for 4.9.0-040900-generic
Building for architecture x86_64
Building initial module for 4.9.0-040900-generic
ERROR (dkms apport): kernel package linux-headers-4.9.0-040900-generic is not supported
Error! Bad return status for module build on kernel: 4.9.0-040900-generic (x86_64)
Consult /var/lib/dkms/amdgpu-pro/16.50-362463/build/make.log for more information.

so maybe reconfigure?

[email protected]:~/Downloads/amdgpu-pro-16.50-362463$ sudo dpkg-reconfigure amdgpu-pro-dkms

------------------------------
Deleting module version: 16.50-362463
completely from the DKMS tree.
------------------------------
Done.
Loading new amdgpu-pro-16.50-362463 DKMS files...
First Installation: checking all kernels...
Building only for 4.9.0-040900-generic
Building for architecture x86_64
Building initial module for 4.9.0-040900-generic
ERROR (dkms apport): kernel package linux-headers-4.9.0-040900-generic is not supported
Error! Bad return status for module build on kernel: 4.9.0-040900-generic (x86_64)
Consult /var/lib/dkms/amdgpu-pro/16.50-362463/build/make.log for more information.

build log dump:

DKMS make.log for amdgpu-pro-16.50-362463 for kernel 4.9.0-040900-generic (x86_64)
Di 3. Jan 20:31:47 CET 2017
make: Entering directory '/usr/src/linux-headers-4.9.0-040900-generic'
  LD      /var/lib/dkms/amdgpu-pro/16.50-362463/build/built-in.o
  LD      /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/built-in.o
  CC [M]  /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/amdgpu_drv.o
In file included from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:5:0,
                 from <command-line>:0:
/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_ttm.h: In function ‘kcl_ttm_bo_reserve’:
/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_ttm.h:116:9: error: too many arguments to function ‘ttm_bo_reserve’
  return ttm_bo_reserve(bo, interruptible, no_wait, false, ticket);
         ^
In file included from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_ttm.h:6:0,
                 from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:5,
                 from <command-line>:0:
./include/drm/ttm/ttm_bo_driver.h:861:19: note: declared here
 static inline int ttm_bo_reserve(struct ttm_buffer_object *bo,
                   ^
In file included from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:5:0,
                 from <command-line>:0:
/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_ttm.h: In function ‘kcl_ttm_bo_move_accel_cleanup’:
/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_ttm.h:153:11: error: incompatible type for argument 4 of ‘ttm_bo_move_accel_cleanup’
    evict, no_wait_gpu, new_mem);
           ^
In file included from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_ttm.h:6:0,
                 from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:5,
                 from <command-line>:0:
./include/drm/ttm/ttm_bo_driver.h:1027:12: note: expected ‘struct ttm_mem_reg *’ but argument is of type ‘bool {aka _Bool}’
 extern int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo,
            ^
In file included from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:5:0,
                 from <command-line>:0:
/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_ttm.h:152:9: error: too many arguments to function ‘ttm_bo_move_accel_cleanup’
  return ttm_bo_move_accel_cleanup(bo, fence,
         ^
In file included from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_ttm.h:6:0,
                 from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:5,
                 from <command-line>:0:
./include/drm/ttm/ttm_bo_driver.h:1027:12: note: declared here
 extern int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo,
            ^
In file included from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:9:0,
                 from <command-line>:0:
/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_mm.h: In function ‘kcl_get_user_pages’:
/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_mm.h:12:55: warning: passing argument 4 of ‘get_user_pages’ makes pointer from integer without a cast [-Wint-conversion]
         return get_user_pages(start, nr_pages, write, force, pages, vmas);
                                                       ^
In file included from ./include/linux/scatterlist.h:7:0,
                 from ./include/linux/dma-mapping.h:10,
                 from ./include/drm/drmP.h:37,
                 from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_drm.h:5,
                 from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:4,
                 from <command-line>:0:
./include/linux/mm.h:1278:6: note: expected ‘struct page **’ but argument is of type ‘int’
 long get_user_pages(unsigned long start, unsigned long nr_pages,
      ^
In file included from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:9:0,
                 from <command-line>:0:
/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_mm.h:12:62: error: passing argument 5 of ‘get_user_pages’ from incompatible pointer type [-Werror=incompatible-pointer-types]
         return get_user_pages(start, nr_pages, write, force, pages, vmas);
                                                              ^
In file included from ./include/linux/scatterlist.h:7:0,
                 from ./include/linux/dma-mapping.h:10,
                 from ./include/drm/drmP.h:37,
                 from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_drm.h:5,
                 from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:4,
                 from <command-line>:0:
./include/linux/mm.h:1278:6: note: expected ‘struct vm_area_struct **’ but argument is of type ‘struct page **’
 long get_user_pages(unsigned long start, unsigned long nr_pages,
      ^
In file included from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:9:0,
                 from <command-line>:0:
/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_mm.h:12:16: error: too many arguments to function ‘get_user_pages’
         return get_user_pages(start, nr_pages, write, force, pages, vmas);
                ^
In file included from ./include/linux/scatterlist.h:7:0,
                 from ./include/linux/dma-mapping.h:10,
                 from ./include/drm/drmP.h:37,
                 from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/include/kcl/kcl_drm.h:5,
                 from /var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/../backport/backport.h:4,
                 from <command-line>:0:
./include/linux/mm.h:1278:6: note: declared here
 long get_user_pages(unsigned long start, unsigned long nr_pages,
      ^
cc1: some warnings being treated as errors
scripts/Makefile.build:293: recipe for target '/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/amdgpu_drv.o' failed
make[2]: *** [/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu/amdgpu_drv.o] Error 1
scripts/Makefile.build:544: recipe for target '/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu' failed
make[1]: *** [/var/lib/dkms/amdgpu-pro/16.50-362463/build/amd/amdgpu] Error 2
Makefile:1496: recipe for target '_module_/var/lib/dkms/amdgpu-pro/16.50-362463/build' failed
make: *** [_module_/var/lib/dkms/amdgpu-pro/16.50-362463/build] Error 2
make: Leaving directory '/usr/src/linux-headers-4.9.0-040900-generic'

ok, im pretty sure i need to fix this dkms apport error. it seems that when i install the driver or do anything driver related it fails to do something with the kernel giving this exact error

i uninstalled 16.50 and installed 16.40 and the same

Building initial module for 4.9.0-040900-generic
ERROR (dkms apport): kernel package linux-headers-4.9.0-040900-generic is not supported
Error! Bad return status for module build on kernel: 4.9.0-040900-generic (x86_64)
Consult /var/lib/dkms/amdgpu-pro/16.40-348864/build/make.log for more information.

occurs.

so im gonna try and find out what the hell dkms apport is and why it doenst like kernel 4.4-4.9

so i found a command line to repair the dkms modules

* dkms: running auto installation service for kernel 4.9.0-040900-generic                       
Kernel preparation unnecessary for this kernel.  Skipping...

Running the pre_build script:

Building module:
cleaning build area....
make KERNELRELEASE=4.9.0-040900-generic -C /lib/modules/4.9.0-040900-generic/build M=/var/lib/dkms/amdgpu-pro/16.40-348864/build....(bad exit status: 2)
ERROR (dkms apport): kernel package linux-headers-4.9.0-040900-generic is not supported
Error! Bad return status for module build on kernel: 4.9.0-040900-generic (x86_64)
Consult /var/lib/dkms/amdgpu-pro/16.40-348864/build/make.log for more information.

same error still but it is detected.

so i did this:

~# /usr/sbin/dkms autoinstall

Kernel preparation unnecessary for this kernel.  Skipping...

Running the pre_build script:

    Building module:
    cleaning build area....
    make KERNELRELEASE=4.9.0-040900-generic -C /lib/modules/4.9.0-040900-generic/build M=/var/lib/dkms/amdgpu-pro/16.40-348864/build....(bad exit status: 2)
    ERROR (dkms apport): kernel package linux-headers-4.9.0-040900-generic is not supported
    Error! Bad return status for module build on kernel: 4.9.0-040900-generic (x86_64)
    Consult /var/lib/dkms/amdgpu-pro/16.40-348864/build/make.log for more information.

bad exit status 2

[email protected]:~$ sudo apt-get install amdgpu-pro
Reading package lists... Done
Building dependency tree       
Reading state information... Done
amdgpu-pro is already the newest version (16.50-362463).
0 upgraded, 0 newly installed, 0 to remove and 2 not upgraded.

[email protected]:~$ sudo apt-get install amdgpu-pro-lib32 
Reading package lists... Done
Building dependency tree       
Reading state information... Done
amdgpu-pro-lib32 is already the newest version (16.50-362463).
0 upgraded, 0 newly installed, 0 to remove and 2 not upgraded.

[email protected]:~$ sudo apt-get install amd64-microcode 
Reading package lists... Done
Building dependency tree       
Reading state information... Done
amd64-microcode is already the newest version (2.20160316.1).
0 upgraded, 0 newly installed, 0 to remove and 2 not upgraded.

[email protected]:~$ sudo apt-get install amd-pro-dkms
Reading package lists... Done
Building dependency tree       
Reading state information... Done
E: Unable to locate package amd-pro-dkms

#315

**i fixed it i think.

get kernel 4.4
install 16.50
sudo dpkg-reconfigure amdgpu-pro-dkms
sudo -i for root
echo -e "high" > /sys/class/drm/card0/device/power_dpm_force_performance_level

but i still get 30-40 fps in rocket league everything maxed out at 3440x1440 60fps/hz thats a bit disappointing**


#316

Anyone else getting system crash after joining a community csgo server with custom map?
My computer fully locks up and I have to force hardware reboot.


#317

The game has been broken since the last update 3 days ago. Matchmaking has been down, CEVO and FACEIT don't always connect. Its a shitty game I don't get why people play it.


#318

Well even with latest gallium/mesa certain games are currently broken for me (e.g., Tyranny). I get random green screen flicker that doesn't happen if I'm running my integrated graphics.


#319

Apparently there are issues with Mesa and the latest Unreal Engine 4 as well. Which is frustrating because I enjoy starting that up from time to time just to derp around. Looking at you Tim Sweeney! 0_0


#320

I just ordered an NVidia card instead. I just don't find having to mess with this stuff interesting. Maybe once the dust settles I'll try AMD again but for now they've lost me.


#321

http://wccftech.com/amd-confirms-ryzen-cpus-unlocked-overclocking/
We summarized the key takeaways and announcements in several points :

All AMD Ryzen CPUs will be multiplier unlocked and ready for overclocking right out of the box.There are no locked Ryzen CPUs.
AM4 X370, X300 and B350 chipsets all support overclocking.
Ryzen CPUs will come in different configurations with varying core counts.The highest core count parts will have eight cores and support sixteen threads.
Ryzen will launch closer to the middle of Q1 or early Q1 rather than the end of Q1. AMD wouldn’t give us a specific date but sometime in February is a good bet.
AM4 compatible aftermarket air and liquid cooling solutions will be available on day one.

So much better then penalizing overclockers


#322

YEAH INTEL GOOOOD
kek


#323

Having competition has to suck