AMD gpu detected as Intel

[puttup]

lspci indicates I have an AMD gpu, while ramalama info detects Intel:

$ lspci | grep -i 'VGA compatible controller'
09:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Navi 22 [Radeon RX 6700/6700 XT/6750 XT / 6800M/6850M XT] (rev c1)

    "GPUs": {
        "Detected GPUs": [
            {
                "Env": "ONEAPI_DEVICE_SELECTOR",
                "GPU": "Intel GPU",
                "VRAM": "12272 MiB"
            }
        ],
        "INFO": [
            {
                "INFO": "`nvidia-smi` not found. No NVIDIA GPU detected or drivers missing.",
                "Vendor": "NVIDIA"
            },
            {
                "INFO": "No AMD GPU detected or drivers missing.",
                "Vendor": "AMD"
            }
        ]
    }

Fedora 41
ramalama 0.6.0

[rhatdan]

I am seeing similar issues with

$ lspci | grep VGA
00:02.0 VGA compatible controller: Intel Corporation Raptor Lake-S UHD Graphics (rev 04)
01:00.0 VGA compatible controller: NVIDIA Corporation AD103GLM [RTX 5000 Ada Generation Laptop GPU] (rev a1)

    "GPUs": {
        "Detected GPUs": [
            {
                "Details": "00:02.0 VGA compatible controller: Intel Corporation Raptor Lake-S UHD Graphics (rev 04)",
                "Env": "ONEAPI_DEVICE_SELECTOR",
                "GPU": "Intel GPU",
                "VRAM": "Unknown"
            }
        ],
        "INFO": [
            {
                "INFO": "No NVIDIA GPU detected or drivers missing.",
                "Vendor": "NVIDIA"
            },
            {
                "INFO": "No AMD GPU detected or drivers missing.",
                "Vendor": "AMD"
            }
        ]
    },

[rhatdan]

@cgruver PTAL

Is this caused by:
#749

[rhatdan]

@ericcurtin PTAL

[ericcurtin]

This GPUDetector code's only functionality is displaying GPU details. It doesn't have any real functionality in RamaLama.

I use nvtop as my generic tool to display all GPU info, I'm not even sure we should have extended RamaLama to do things like this.

[cgruver]

@ericcurtin @rhatdan Instead of this functionality, should we just modify ramalama to report what it would do with a --dry-run or something?

The GPU detector code has a risk of reporting different info, since it uses different logic from the detector logic in common.py

[rhatdan]

I think we should unify these to be able to report the same thing.