Mac X64 broken (not really surprised)

[peardox]

Not that bothered really but FYI

Really old Macbook Pro (Later 2013)

Built as

cmake -B build -DBINDINGS_FLAT=ON -DBUILD_SHARED_LIBS=ON -DGGML_METAL=ON -DGGML_BACKEND_DL=ON -DGGML_CPU_ALL_VARIANTS=ON

Note : Using an unmerged PR

./build/bin/whisper-bench -m ~/models/ggml-base.en.bin
+++ BINDINGS_FLAT +++
load_backend: loaded BLAS backend from /Users/simon/src/whisper.cpp/./build/bin/libggml-blas.so
load_backend: loaded Metal backend from /Users/simon/src/whisper.cpp/./build/bin/libggml-metal.so
load_backend: loaded CPU backend from /Users/simon/src/whisper.cpp/./build/bin/libggml-cpu-haswell.so
whisper_init_from_file_with_params_no_state: loading model from '/Users/simon/models/ggml-base.en.bin'
whisper_init_with_params_no_state: use gpu = 1
whisper_init_with_params_no_state: flash attn = 0
whisper_init_with_params_no_state: gpu_device = 0
whisper_init_with_params_no_state: dtw = 0
whisper_init_with_params_no_state: devices = 3
whisper_init_with_params_no_state: backends = 3
whisper_model_load: loading model
whisper_model_load: n_vocab = 51864
whisper_model_load: n_audio_ctx = 1500
whisper_model_load: n_audio_state = 512
whisper_model_load: n_audio_head = 8
whisper_model_load: n_audio_layer = 6
whisper_model_load: n_text_ctx = 448
whisper_model_load: n_text_state = 512
whisper_model_load: n_text_head = 8
whisper_model_load: n_text_layer = 6
whisper_model_load: n_mels = 80
whisper_model_load: ftype = 1
whisper_model_load: qntvr = 0
whisper_model_load: type = 2 (base)
whisper_model_load: adding 1607 extra tokens
whisper_model_load: n_langs = 99
whisper_model_load: Metal total size = 147.37 MB
whisper_model_load: model size = 147.37 MB
whisper_backend_init_gpu: using Metal backend
ggml_metal_init: allocating
ggml_metal_init: found device: Intel Iris Graphics
ggml_metal_init: picking default device: Intel Iris Graphics
ggml_metal_load_library: using embedded metal library
ggml_metal_init: GPU name: Intel Iris Graphics
ggml_metal_init: GPU family: MTLGPUFamilyCommon2 (3002)
ggml_metal_init: simdgroup reduction = false
ggml_metal_init: simdgroup matrix mul. = false
ggml_metal_init: has residency sets = false
ggml_metal_init: has bfloat = false
ggml_metal_init: use bfloat = false
ggml_metal_init: hasUnifiedMemory = true
ggml_metal_init: recommendedMaxWorkingSetSize = 1610.61 MB
ggml_metal_init: skipping kernel_soft_max_f16 (not supported)
ggml_metal_init: skipping kernel_soft_max_f16_4 (not supported)
ggml_metal_init: skipping kernel_soft_max_f32 (not supported)
ggml_metal_init: skipping kernel_soft_max_f32_4 (not supported)
ggml_metal_init: skipping kernel_get_rows_bf16 (not supported)
ggml_metal_init: skipping kernel_rms_norm (not supported)
ggml_metal_init: skipping kernel_l2_norm (not supported)
ggml_metal_init: skipping kernel_group_norm (not supported)
ggml_metal_init: error: load pipeline error: Error Domain=CompilerError Code=2 "Compiler encountered an internal error" UserInfo={NSLocalizedDescription=Compiler encountered an internal error}
ggml_backend_metal_device_init: error: failed to allocate context
whisper_backend_init_gpu: failed to initialize Metal backend
whisper_backend_init: using BLAS backend
whisper_init_state: kv self size = 6.29 MB
whisper_init_state: kv cross size = 18.87 MB
whisper_init_state: kv pad size = 3.15 MB
whisper_init_state: compute buffer (conv) = 16.26 MB
whisper_init_state: compute buffer (encode) = 85.86 MB
whisper_init_state: compute buffer (cross) = 4.65 MB
whisper_init_state: compute buffer (decode) = 96.35 MB

system_info: n_threads = 4 / 4 | WHISPER : COREML = 0 | OPENVINO = 0 | Metal : EMBED_LIBRARY = 1 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | ACCELERATE = 1 | AARCH64_REPACK = 1 |

whisper_print_timings: load time = 263.02 ms
whisper_print_timings: fallbacks = 0 p / 0 h
whisper_print_timings: mel time = 0.00 ms
whisper_print_timings: sample time = 0.00 ms / 1 runs ( 0.00 ms per run)
whisper_print_timings: encode time = 2357.60 ms / 1 runs ( 2357.60 ms per run)
whisper_print_timings: decode time = 1939.51 ms / 256 runs ( 7.58 ms per run)
whisper_print_timings: batchd time = 1191.79 ms / 320 runs ( 3.72 ms per run)
whisper_print_timings: prompt time = 10549.91 ms / 4096 runs ( 2.58 ms per run)
whisper_print_timings: total time = 16040.46 ms

I guess building without metal will help

...

Nope - can't seem to disable metal - aw well, remove libggml-metal?

Yep...

whisper_backend_init: using BLAS backend
whisper_init_state: kv self size = 6.29 MB
whisper_init_state: kv cross size = 18.87 MB
whisper_init_state: kv pad size = 3.15 MB
whisper_init_state: compute buffer (conv) = 16.26 MB
whisper_init_state: compute buffer (encode) = 85.86 MB
whisper_init_state: compute buffer (cross) = 4.65 MB
whisper_init_state: compute buffer (decode) = 96.35 MB

system_info: n_threads = 4 / 4 | WHISPER : COREML = 0 | OPENVINO = 0 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | ACCELERATE = 1 | AARCH64_REPACK = 1 |

whisper_print_timings: load time = 155.85 ms
whisper_print_timings: fallbacks = 0 p / 0 h
whisper_print_timings: mel time = 0.00 ms
whisper_print_timings: sample time = 0.00 ms / 1 runs ( 0.00 ms per run)
whisper_print_timings: encode time = 2270.11 ms / 1 runs ( 2270.11 ms per run)
whisper_print_timings: decode time = 1889.95 ms / 256 runs ( 7.38 ms per run)
whisper_print_timings: batchd time = 1188.71 ms / 320 runs ( 3.71 ms per run)
whisper_print_timings: prompt time = 10139.99 ms / 4096 runs ( 2.48 ms per run)
whisper_print_timings: total time = 15490.28 ms