Vulkan support broken at v1.8.0

[iAdanos]

Hey, just built whisper.cpp v1.8.0 with Vulkan support and it does not detect by AMD GPU (ROCM driver, integrated graphics).

When I do switch back and build v1.7.6 - it works correctly.

[Janghou]

Confirmed

whisper_init_from_file_with_params_no_state: loading model from 'models/ggml-base.en.bin'
whisper_init_with_params_no_state: use gpu = 1
whisper_init_with_params_no_state: flash attn = 1
whisper_init_with_params_no_state: gpu_device = 0
whisper_init_with_params_no_state: dtw = 0
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = AMD Radeon Graphics (RADV GFX1152) (radv) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 64 | shared memory:
...
whisper_backend_init_gpu: no GPU found

[DarkSwan86]

Hi, on my RDNA2 system (Radeon 6800) on Fedora KDE 42 up-to-date (kernel 6.16.10, Mesa 25.1.9) the 1.8.0 (build from source with the Vulkan support) works fine:

XXXXXXXXX@fedora:~/XXXXXXXXX/test/whispercpp/whisper.cpp$ ./build/bin/whisper-cli -f samples/jfk.wav -m models/ggml-large-v1.bin -pc -pp -l auto
whisper_init_from_file_with_params_no_state: loading model from 'models/ggml-large-v1.bin'
whisper_init_with_params_no_state: use gpu = 1
whisper_init_with_params_no_state: flash attn = 1
whisper_init_with_params_no_state: gpu_device = 0
whisper_init_with_params_no_state: dtw = 0
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = AMD Radeon RX 6800 (RADV NAVI21) (radv) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 65536 | int dot: 1 | matrix cores: none
whisper_init_with_params_no_state: devices = 2
whisper_init_with_params_no_state: backends = 2
whisper_model_load: loading model
whisper_model_load: n_vocab = 51865
whisper_model_load: n_audio_ctx = 1500
whisper_model_load: n_audio_state = 1280
whisper_model_load: n_audio_head = 20
whisper_model_load: n_audio_layer = 32
whisper_model_load: n_text_ctx = 448
whisper_model_load: n_text_state = 1280
whisper_model_load: n_text_head = 20
whisper_model_load: n_text_layer = 32
whisper_model_load: n_mels = 80
whisper_model_load: ftype = 1
whisper_model_load: qntvr = 0
whisper_model_load: type = 5 (large)
whisper_model_load: adding 1608 extra tokens
whisper_model_load: n_langs = 99
whisper_model_load: Vulkan0 total size = 3093.99 MB
whisper_model_load: model size = 3093.99 MB
whisper_backend_init_gpu: using Vulkan0 backend
whisper_init_state: kv self size = 83.89 MB
whisper_init_state: kv cross size = 251.66 MB
whisper_init_state: kv pad size = 7.86 MB
whisper_init_state: compute buffer (conv) = 35.67 MB
whisper_init_state: compute buffer (encode) = 55.35 MB
whisper_init_state: compute buffer (cross) = 9.27 MB
whisper_init_state: compute buffer (decode) = 100.04 MB

system_info: n_threads = 4 / 32 | WHISPER : COREML = 0 | OPENVINO = 0 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | OPENMP = 1 | REPACK = 1 |

main: processing 'samples/jfk.wav' (176000 samples, 11.0 sec), 4 threads, 1 processors, 5 beams + best of 5, lang = auto, task = transcribe, timestamps = 1 ...
main: color scheme: red (low confidence), yellow (medium), green (high confidence)

whisper_full_with_state: auto-detected language: en (p = 0.988389)

[00:00:00.000 --> 00:00:11.000] And so my fellow Americans, ask not what your country can do for you, ask what you can do for your country.
whisper_print_progress_callback: progress = 100%

whisper_print_timings: load time = 1126.48 ms
whisper_print_timings: fallbacks = 0 p / 0 h
whisper_print_timings: mel time = 9.96 ms
whisper_print_timings: sample time = 34.34 ms / 142 runs ( 0.24 ms per run)
whisper_print_timings: encode time = 625.72 ms / 2 runs ( 312.86 ms per run)
whisper_print_timings: decode time = 10.01 ms / 1 runs ( 10.01 ms per run)
whisper_print_timings: batchd time = 316.71 ms / 140 runs ( 2.26 ms per run)
whisper_print_timings: prompt time = 0.00 ms / 1 runs ( 0.00 ms per run)
whisper_print_timings: total time = 2136.65 ms

It might be a driver bug of the RDNA 3.5?
What version of Mesa are you using?
Did you also try the latest stable 25.2.4 or even the master?

Maybe you can report it to the mesa bugtracker, just to be safe.

[BruBetIta]

Same situation here, vulkan working with 1.7.6 not working with 1.8.0

AMD Radeon Graphics (RADV RENOIR) (radv) | uma: 1 | fp16: 1 | warp size: 64 | shared memory: 65536 | int dot: 0 | matrix

mesa-25.1.7