Misc. bug: Crashed on Intel Mac with AMD GPU.

[halechan]

Name and Version

./build/bin/Debug/llama-cli --version
register_backend: registered backend Metal (1 devices)
register_device: registered device Metal (AMD Radeon HD GFX10 Family Unknown Prototype)
register_backend: registered backend BLAS (1 devices)
register_device: registered device BLAS (Accelerate)
register_backend: registered backend CPU (1 devices)
register_device: registered device CPU (Intel(R) Core(TM) i7-14700K)
version: 4726 (6dde178)
built with Apple clang version 16.0.0 (clang-1600.0.26.4) for x86_64-apple-darwin24.1.0

Operating systems

Mac

Which llama.cpp modules do you know to be affected?

libllama (core library)

Command line

llama-cli -m ../DeepSeek-R1-Distill-Qwen-1.5B-Q4_K_M.gguf

Problem description & steps to reproduce

crashed with message:

-[MTLDebugDevice newResidencySetWithDescriptor:error:]:2768: failed assertion `device does not support residency sets.'

First Bad Commit

No response

Relevant log output

[halechan]

Refer to halechan@3cecb21 and modify ggml/src/ggml-metal/ggml-metal.m to fix this issue.