[BUG] Upstream package bug prevents OpenCL from compiling on Intel Arc

首先检查容器如何安装软件包，然后在 Intel Arc 系统上使用 darktable -d opencl 重现报告的问题。检查列出的 OpenCL、Intel GPU、Vulkan 和 unzip 依赖项，以及 AI 安装命令。当 Intel Arc OpenCL 和 Darktable AI 功能无需手动安装软件包、删除缓存或更新系统即可正常工作时，即表示完成。

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

Container does not work out of the box with Intel descrete GPU's (Intel Arc Alchemist)

Expected Behavior

The container should support ARC out of the box. Including Darktable's AI features.

Steps To Reproduce

Hello!
I believe we need a package update for this container.

I have spent a few days trying to chase OpenCL failing to compile for 'darktable -d opencl'. I've had to make a few tweaks:

• I've had to install the following packages on container launch:
  igsc|intel-compute-runtime|clinfo|intel-gpu-tools|intel-media-driver|ocl-icd|opencl-headers|vulkan-intel
• It's required to close darktable, update the system with pacman -Syu
• It's required to delete ~/.cache/darktable/cached_*
• It's required to launch "darktable -d opencl" to then recompile Intel OpenCL for Darktable.

Additionally, to use AI features, the container requires the unzip package when using Darktable's one liner AI GPU script found here: https://docs.darktable.org/usermanual/development/en/special-topics/ai/gpu-acceleration/

The oneliner to install the AI features is:
curl -fsSL https://raw.githubusercontent.com/darktable-org/darktable/HEAD/tools/ai/install-ort-gpu.sh | bash

I have a few other variables set, but I don't know if they're required or not. Wayland grinds to a halt if using multiGPU. On a single card, it seems to be fine.

I'm happy to assist, but I'm a novice to building docker containers. But I can work through issues with them.

Thank you for the work that your team does!

Environment

- OS: Unraid
- How docker service was installed: Unraid

CPU architecture

x86-64

Docker creation

Unraid

Container logs

.