I recently call an HP ZBook Firefly 14 G10 A my own. It's the successor of an older Thinkpad with Intel UHD GPU that has become a bit slow for my usecase. But it runs OpenCL in darktable smoothly. Both run Kubuntu 23.10 on Wayland.
My System has currently 16GB of DDR5 RAM with a clock speed of 5600 Mhz (will upgrade to 32GB soon), the Ryzen 7 Pro 7840HS, with 8 Zen 4 cores. GPU is the built-in Radeon 780M RDNA3 GPU. In UEFI Firmware settings I've chosen "Gaming optimized" for VRAM. I'm really amazed by this device. It runs XP12 in FHD on lower settings with 30-40 FPS which is quite impressive regarding that this is just mobile iGPU.
I just edited some RAWs in darktable and wondered about the extreme CPU usage. So, I took a look at the "processing" settings where I shockingly recognised that OpenCL wasn't enabled and greyed out.
By searching on the internet about this issue I found the command "darktable-cltest" which returns me the following:
0,0179 [dt_get_sysresource_level] switched to 1 as `default'
0,0180 total mem: 13740MB
0,0180 mipmap cache: 1717MB
0,0180 available mem: 6870MB
0,0180 singlebuff: 107MB
0,0180 OpenCL tune mem: OFF
0,0180 OpenCL pinned: OFF
[opencl_init] opencl related configuration options:
[opencl_init] opencl: ON
[opencl_init] opencl_scheduling_profile: 'default'
[opencl_init] opencl_library: 'default path'
[opencl_init] opencl_device_priority: '*/!0,*/*/*/!0,*'
[opencl_init] opencl_mandatory_timeout: 400
[opencl_init] opencl library 'libOpenCL' found on your system and loaded
[opencl_init] found 4 platforms
[opencl_init] platform 2 is not FULL_PROFILE
[opencl_init] no devices found for Mesa/X.org (vendor) - rusticl (name)
[opencl_init] found 2 devices
[dt_opencl_device_init]
DEVICE: 0: 'AMD Radeon Graphics (gfx1103_r1, LLVM 15.0.7, DRM 3.54, 6.5.0-14-generic)'
PLATFORM NAME & VENDOR: Clover, Mesa
CANONICAL NAME: cloveramdradeongraphics
DRIVER VERSION: 23.2.1-1ubuntu3.1
DEVICE VERSION: OpenCL 1.1 Mesa 23.2.1-1ubuntu3.1
DEVICE_TYPE: GPU
*** insufficient device version ***
[dt_opencl_device_init]
DEVICE: 1: 'AMD Ryzen 7 PRO 7840HS w/ Radeon 780M Graphics '
PLATFORM NAME & VENDOR: Intel(R) OpenCL, Intel(R) Corporation
CANONICAL NAME: intelropenclamdryzen7pro7840hswradeon780mgraphics
DRIVER VERSION: 2023.16.12.0.12_195853.xmain-hotfix
DEVICE VERSION: OpenCL 3.0 (Build 0)
DEVICE_TYPE: CPU
GLOBAL MEM SIZE: 13741 MB
MAX MEM ALLOC: 6870 MB
MAX IMAGE SIZE: 16384 x 16384
MAX WORK GROUP SIZE: 8192
MAX WORK ITEM DIMENSIONS: 3
MAX WORK ITEM SIZES: [ 8192 8192 8192 ]
ASYNC PIXELPIPE: NO
PINNED MEMORY TRANSFER: NO
MEMORY TUNING: NO
FORCED HEADROOM: 400
AVOID ATOMICS: NO
MICRO NAP: 1000
ROUNDUP WIDTH: 16
ROUNDUP HEIGHT: 16
CHECK EVENT HANDLES: 128
TILING ADVANTAGE: 0.000
DEFAULT DEVICE: NO
*** marked as disabled ***
[opencl_init] no suitable devices found.
[opencl_init] FINALLY: opencl is NOT AVAILABLE and NOT ENABLED.
Okay, it tells me that it has actually found theoretically relevant devices one of whom is my GPU and the other one seems to be a software renderer on the CPU. Both are assessed as ineligible but, why? Of course, the second "device" isn't properly a device and perhaps using a COU to software render OpenCL will not give a decent boost in performance.
But what's about the GPU? It bothers me a lot that the information printed over the GPU is that sparse. There are so many informations missing.
I mean I have neither exotic hardware nor an exotic OS. It would be strange if I was alone with this problem. Whats the issue with OpenCL on Ryzen APUs so far? What packages do we need?
