The run efficiency of main target built based on CMakeList and main built based on Makefile differ greatly

[chenqianhe]

main built based on make run

whisper_init_from_file: loading model from 'ggml-large.bin'
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: f16           = 1
whisper_model_load: type          = 5
whisper_model_load: mem required  = 4641.00 MB (+   71.00 MB per decoder)
whisper_model_load: kv self size  =   70.00 MB
whisper_model_load: kv cross size =  234.38 MB
whisper_model_load: adding 1608 extra tokens
whisper_model_load: model ctx     = 2950.97 MB
whisper_model_load: model size    = 2950.66 MB

system_info: n_threads = 4 / 8 | AVX = 0 | AVX2 = 0 | AVX512 = 0 | FMA = 0 | NEON = 1 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 1 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | VSX = 0 | 

main: processing 'samples_jfk.wav' (176000 samples, 11.0 sec), 4 threads, 1 processors, lang = zh, task = transcribe, timestamps = 1 ...


[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_timings:     fallbacks =   0 p /   0 h
whisper_print_timings:     load time =  2146.14 ms
whisper_print_timings:      mel time =    19.26 ms
whisper_print_timings:   sample time =    11.52 ms /    27 runs (    0.43 ms per run)
whisper_print_timings:   encode time =  6265.27 ms /     1 runs ( 6265.27 ms per run)
whisper_print_timings:   decode time =  1646.73 ms /    27 runs (   60.99 ms per run)
whisper_print_timings:    total time = 10142.16 ms

main built based on cmake run

whisper_init_from_file: loading model from '../../ggml-large.bin'
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: f16           = 1
whisper_model_load: type          = 5
whisper_model_load: mem required  = 4641.00 MB (+   71.00 MB per decoder)
whisper_model_load: kv self size  =   70.00 MB
whisper_model_load: kv cross size =  234.38 MB
whisper_model_load: adding 1608 extra tokens
whisper_model_load: model ctx     = 2950.97 MB
whisper_model_load: model size    = 2950.66 MB

system_info: n_threads = 4 / 8 | AVX = 0 | AVX2 = 0 | AVX512 = 0 | FMA = 0 | NEON = 1 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 1 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | VSX = 0 | 

main: processing '../../samples_jfk.wav' (176000 samples, 11.0 sec), 4 threads, 1 processors, lang = zh, task = transcribe, timestamps = 1 ...


[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_timings:     fallbacks =   0 p /   0 h
whisper_print_timings:     load time = 27118.13 ms
whisper_print_timings:      mel time =   107.82 ms
whisper_print_timings:   sample time =    46.74 ms /    27 runs (    1.73 ms per run)
whisper_print_timings:   encode time = 63709.25 ms /     1 runs (63709.25 ms per run)
whisper_print_timings:   decode time =  8359.73 ms /    27 runs (  309.62 ms per run)
whisper_print_timings:    total time = 99418.73 ms

I used the same device to get the above results.

I wonder why cmake's main is much slower. Is there something wrong with me

[chenqianhe]

My goal is to optimize #260 ; I am implementing the addon of node, which can call the Whisper inference implemented by cpp. But it depends on cmake.

This is another case of addon that I have implemented. It is crucial that I can complete CMakeList.txt

[ggerganov]

Most likely you have CMake build in Debug.
Try the following:

rm CMakeCache.txt
cmake -DCMAKE_BUILD_TYPE=Release ../../
make

[chenqianhe]

Most likely you have CMake build in Debug. Try the following:

rm CMakeCache.txt
cmake -DCMAKE_BUILD_TYPE=Release ../../
make

Thank you very much!
It is really caused by this problem.