github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-23 14:13:13 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #13591] Issue: AMD Radeon Pro 5300M not detected on macOS 15.2 - GPU count remains 0 #86637

New Issue
Open
opened 2026-05-10 03:48:05 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-05-10 03:48:05 -05:00
Owner
Copy Link

Originally created by @andrew-tawfeek on GitHub (Dec 31, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/13591

What is the issue?

(This below report was generated with Claude's help after trying to debug this CPU/GPU issue for an hour or so with no avail. I can't find something similar ion the issue reports so it felt best to open one up.)

System Information:

  • OS: macOS 15.2 (Build 25B78)
  • Ollama Version: 0.13.5
  • CPU: Intel Core i7-9750H
  • GPU: AMD Radeon Pro 5300M (4GB VRAM)
  • RAM: 16GB
  • Xcode Command Line Tools: Installed

Problem:
Ollama fails to detect AMD Radeon Pro 5300M GPU and falls back to CPU-only inference. GPU discovery completes but finds zero GPUs despite Metal 3 being supported.

GPU Verification:

$ system_profiler SPDisplaysDataType | grep -A 10 "AMD Radeon"
AMD Radeon Pro 5300M:
  Chipset Model: AMD Radeon Pro 5300M
  Type: GPU
  Bus: PCIe
  VRAM (Total): 4 GB
  Metal Support: Metal 3

Debug Output:

time=2025-12-30T23:32:22.220-08:00 level=INFO source=runner.go:67 msg="discovering available GPUs..."
time=2025-12-30T23:32:22.350-08:00 level=DEBUG source=runner.go:40 msg="GPU bootstrap discovery took" duration=129.337766ms
time=2025-12-30T23:32:22.350-08:00 level=INFO source=types.go:60 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="16.0 GiB" available="5.7 GiB"
time=2025-12-30T23:32:39.670-08:00 level=DEBUG source=sched.go:194 msg="updating default concurrency" gpu_count=0

Expected Behavior:
Ollama should detect the AMD Radeon Pro 5300M via Metal and use it for GPU acceleration.

Actual Behavior:

  • GPU discovery runs but finds 0 GPUs
  • All inference runs on CPU at 100%
  • ollama ps shows 100% CPU / 0% GPU
  • Total VRAM reported as 0 B

Steps to Reproduce:

  1. Install Ollama 0.13.5 on macOS 15.2 with AMD discrete GPU
  2. Run OLLAMA_DEBUG=1 ollama serve
  3. In another terminal: ollama run qwen2.5:7b
  4. Check ollama ps - shows 100% CPU usage, no GPU

Additional Context:

  • Metal framework is present and functional
  • Both Intel UHD 630 (integrated) and AMD Radeon Pro 5300M (discrete) report Metal 3 support
  • Automatic Graphics Switching is enabled
  • Same models work with GPU on Ubuntu with NVIDIA GPU

Workarounds Attempted:

  • ✗ GGML_METAL=1 - no effect
  • ✗ OLLAMA_NUM_GPU=1 - no effect
  • ✗ sudo pmset -a gpuswitch 1 - no effect
  • ✗ Reinstalling Ollama - no effect

Is Metal GPU acceleration expected to work with AMD discrete GPUs on macOS, or only with Apple Silicon?

Originally created by @andrew-tawfeek on GitHub (Dec 31, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/13591 ### What is the issue? (This below report was generated with Claude's help after trying to debug this CPU/GPU issue for an hour or so with no avail. I can't find something similar ion the issue reports so it felt best to open one up.) **System Information:** - OS: macOS 15.2 (Build 25B78) - Ollama Version: 0.13.5 - CPU: Intel Core i7-9750H - GPU: AMD Radeon Pro 5300M (4GB VRAM) - RAM: 16GB - Xcode Command Line Tools: Installed **Problem:** Ollama fails to detect AMD Radeon Pro 5300M GPU and falls back to CPU-only inference. GPU discovery completes but finds zero GPUs despite Metal 3 being supported. **GPU Verification:** ```bash $ system_profiler SPDisplaysDataType | grep -A 10 "AMD Radeon" AMD Radeon Pro 5300M: Chipset Model: AMD Radeon Pro 5300M Type: GPU Bus: PCIe VRAM (Total): 4 GB Metal Support: Metal 3 ``` **Debug Output:** ``` time=2025-12-30T23:32:22.220-08:00 level=INFO source=runner.go:67 msg="discovering available GPUs..." time=2025-12-30T23:32:22.350-08:00 level=DEBUG source=runner.go:40 msg="GPU bootstrap discovery took" duration=129.337766ms time=2025-12-30T23:32:22.350-08:00 level=INFO source=types.go:60 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="16.0 GiB" available="5.7 GiB" time=2025-12-30T23:32:39.670-08:00 level=DEBUG source=sched.go:194 msg="updating default concurrency" gpu_count=0 ``` **Expected Behavior:** Ollama should detect the AMD Radeon Pro 5300M via Metal and use it for GPU acceleration. **Actual Behavior:** - GPU discovery runs but finds 0 GPUs - All inference runs on CPU at 100% - `ollama ps` shows `100% CPU / 0% GPU` - Total VRAM reported as `0 B` **Steps to Reproduce:** 1. Install Ollama 0.13.5 on macOS 15.2 with AMD discrete GPU 2. Run `OLLAMA_DEBUG=1 ollama serve` 3. In another terminal: `ollama run qwen2.5:7b` 4. Check `ollama ps` - shows 100% CPU usage, no GPU **Additional Context:** - Metal framework is present and functional - Both Intel UHD 630 (integrated) and AMD Radeon Pro 5300M (discrete) report Metal 3 support - Automatic Graphics Switching is enabled - Same models work with GPU on Ubuntu with NVIDIA GPU **Workarounds Attempted:** - ✗ `GGML_METAL=1` - no effect - ✗ `OLLAMA_NUM_GPU=1` - no effect - ✗ `sudo pmset -a gpuswitch 1` - no effect - ✗ Reinstalling Ollama - no effect Is Metal GPU acceleration expected to work with AMD discrete GPUs on macOS, or only with Apple Silicon?
GiteaMirror added the bug label 2026-05-10 03:48:05 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#86637
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?