You are not logged in.

#1 2017-12-05 15:17:07

cipherkit
Member
Registered: 2017-12-05
Posts: 1

AMD Radeon Pro WX 7100 not recognized as an openCL platform

Hello, I am new to Arch so I am still getting used to the hands on technical details. I haven't given up yet.

What I am trying to do is get my brand new AMD Radeon Pro WX 7100 GPU to do some GPUPU work. I am using BOINC as a test, I figure if I can get the GPU doing work on BOINC then I can move on to other programs. I decided to use Gnome Display Manager as I was given a choice and just picked it. I can change it if need be. I am also using the mate desktop environment.

So far I have installed so many openCL, AMDGPU and AMDGPU-PRO packages that I have lost track of what is on the system. I am looking through my history so here is what I have installed or attempted to install:

Attempted to install the catalyst driver from:https://wiki.archlinux.org/index.php/AM … repository
But I ran into a dependency issue with X-server specifically that x-server<1.18 I couldn't figure out how to downgrade so I did some research and found that the open source drivers were probably adequate so long as I could get openCL running.


pacman'd the following:
lib32-mesa
xf86-mesa-amdgpu
xf86-video-amdgpu
vulkan-radeon
mesa-vdpau
xorg-server
xf86-input-libinput
xorg-server-common
xorg-server-xwayland
libxfont
mesa
lib32-mesa-vdpau
ocl-icd
amdgpu-pro

AUR:
amdgpu-pro-installer
libopencl
linux-amd-staging-drm-next-git
linux-amd-staging-git
opencl-amd

I have many libraries so it may just be a matter of configuration at this point.

I think the AMDGPU-PRO drivers are not being loaded, see "dmesg | grep amd" below.

Basic diagnostics is something that I am having trouble with being new to this. So I will do my best to include the relevant info as well as update  the commands that others suggest.

My specific questions are:
1. What sort of configuration files exist such that I can puzzle out a proper config?
2. What other sort of diagnostic commands should I run?
3. Is it necessary to run X-server instead of GDM?

First goal: Get this to report that there is one platform
"clinfo"

Number of platforms                               0

Second goal: Get BOINC to recognize the graphics card as something it can use for work.

Boinc currently complains that:

Tue 05 Dec 2017 09:06:43 AM MST |  | [coproc] launching child process at /usr/bin/boinc_client
Tue 05 Dec 2017 09:06:43 AM MST |  | [coproc] relative to directory /
Tue 05 Dec 2017 09:06:43 AM MST |  | [coproc] with data directory /var/lib/boinc
Tue 05 Dec 2017 09:06:43 AM MST |  | [coproc] NVIDIA: libcuda.so: cannot open shared object file: No such file or directory
Tue 05 Dec 2017 09:06:43 AM MST |  | [coproc] ATI: libaticalrt.so: cannot open shared object file: No such file or directory
Tue 05 Dec 2017 09:06:43 AM MST |  | [coproc] clGetPlatformIDs() failed to return any OpenCL platforms
Tue 05 Dec 2017 09:06:43 AM MST |  | No usable GPUs found

"dmesg | grep vga" shows this

[    1.360038] pci 0000:01:00.0: vgaarb: setting as boot VGA device
[    1.360038] pci 0000:01:00.0: vgaarb: VGA device added: decodes=io+mem,owns=io+mem,locks=none
[    1.360038] pci 0000:01:00.0: vgaarb: bridge control possible
[    1.360038] vgaarb: loaded
[    8.367216] snd_hda_intel 0000:01:00.1: Handle vga_switcheroo audio client

"lshw -c video":

*-display                 
       description: VGA compatible controller
       product: Ellesmere [Radeon Pro WX 7100]
       vendor: Advanced Micro Devices, Inc. [AMD/ATI]
       physical id: 0
       bus info: pci@0000:01:00.0
       version: 00
       width: 64 bits
       clock: 33MHz
       capabilities: pm pciexpress msi vga_controller bus_master cap_list rom
       configuration: driver=amdgpu latency=0
       resources: irq:31 memory:e0000000-efffffff memory:f0000000-f01fffff ioport:e000(size=256) memory:fbe00000-fbe3ffff memory:c0000-dffff

"lspci -v | grep VGA"

01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Ellesmere [Radeon Pro WX 7100] (prog-if 00 [VGA controller])

"dmesg | grep amd" shows that AMDGPU is loaded I think I need AMDGPU-PRO

[    6.596395] [drm] amdgpu kernel modesetting enabled.
[    6.769434] fb: switching to amdgpudrmfb from VESA VGA
[    6.906460] amdgpu 0000:01:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0xffff
[    6.926313] amdgpu 0000:01:00.0: VRAM: 8192M 0x0000000000000000 - 0x00000001FFFFFFFF (8192M used)
[    6.926316] amdgpu 0000:01:00.0: GTT: 8192M 0x0000000200000000 - 0x00000003FFFFFFFF
[    6.926469] [drm] amdgpu: 8192M of VRAM memory ready
[    6.926471] [drm] amdgpu: 8192M of GTT memory ready.
[    6.927712] amdgpu 0000:01:00.0: amdgpu: using MSI.
[    6.927725] [drm] amdgpu: irq initialized.
[    7.049755] amdgpu: [powerplay] amdgpu: powerplay sw initialized
[    7.149269] amdgpu 0000:01:00.0: fence driver on ring 0 use gpu addr 0x0000000200000008, cpu addr 0xffff9e65799a5008
[    7.149302] amdgpu 0000:01:00.0: fence driver on ring 1 use gpu addr 0x0000000200000018, cpu addr 0xffff9e65799a5018
[    7.149335] amdgpu 0000:01:00.0: fence driver on ring 2 use gpu addr 0x0000000200000028, cpu addr 0xffff9e65799a5028
[    7.149364] amdgpu 0000:01:00.0: fence driver on ring 3 use gpu addr 0x0000000200000038, cpu addr 0xffff9e65799a5038
[    7.149392] amdgpu 0000:01:00.0: fence driver on ring 4 use gpu addr 0x0000000200000048, cpu addr 0xffff9e65799a5048
[    7.149419] amdgpu 0000:01:00.0: fence driver on ring 5 use gpu addr 0x0000000200000058, cpu addr 0xffff9e65799a5058
[    7.149446] amdgpu 0000:01:00.0: fence driver on ring 6 use gpu addr 0x0000000200000068, cpu addr 0xffff9e65799a5068
[    7.149473] amdgpu 0000:01:00.0: fence driver on ring 7 use gpu addr 0x0000000200000078, cpu addr 0xffff9e65799a5078
[    7.149498] amdgpu 0000:01:00.0: fence driver on ring 8 use gpu addr 0x0000000200000088, cpu addr 0xffff9e65799a5088
[    7.149512] amdgpu 0000:01:00.0: fence driver on ring 9 use gpu addr 0x000000020000009c, cpu addr 0xffff9e65799a509c
[    7.160336] amdgpu 0000:01:00.0: fence driver on ring 10 use gpu addr 0x00000002000000ac, cpu addr 0xffff9e65799a50ac
[    7.160827] amdgpu 0000:01:00.0: fence driver on ring 11 use gpu addr 0x00000002000000bc, cpu addr 0xffff9e65799a50bc
[    7.162928] amdgpu 0000:01:00.0: fence driver on ring 12 use gpu addr 0x000000000122d420, cpu addr 0xffffba530445a420
[    7.163550] amdgpu 0000:01:00.0: fence driver on ring 13 use gpu addr 0x00000002000000dc, cpu addr 0xffff9e65799a50dc
[    7.163576] amdgpu 0000:01:00.0: fence driver on ring 14 use gpu addr 0x00000002000000ec, cpu addr 0xffff9e65799a50ec
[    7.206122] amdgpu: [powerplay] [AVFS] Something is broken. See log!
[    7.208003] amdgpu: [powerplay] Can't find requested voltage id in vdd_dep_on_sclk table!
[    8.025426] fbcon: amdgpudrmfb (fb0) is primary device
[    8.343757] amdgpu 0000:01:00.0: fb0: amdgpudrmfb frame buffer device
[    8.367011] [drm] Initialized amdgpu 3.18.0 20150101 for 0000:01:00.0 on minor 0

Last edited by cipherkit (2017-12-05 16:11:29)

Offline

Board footer

Powered by FluxBB