MacBook Pro M3 Crashes
aaronedell opened this issue · 5 comments
When i try
--engine WHISPER_LARGE \
--dataset LIBRI_SPEECH_TEST_OTHER \
--dataset-folder /Users/aedell/Documents/GitHub/speech-to-text-benchmark/datasets/LibriSpeech/test-other \my Macbook Pro's memory spikes and then the machine locks up. According to ChatGPT, the issue is;
Concurrent Execution: The script uses ProcessPoolExecutor for parallel processing. If the dataset is large and the number of processes (num_workers) is not optimally set, this could lead to high memory usage as each process might consume a significant amount of memory.
Dataset and Engine Initialization: Depending on how the Dataset and Engine classes are implemented (not visible in the provided script), loading large datasets or initializing multiple instances of speech-to-text engines could consume a lot of memory.
Memory Leaks: If there are any memory leaks within the Engine or Dataset classes (e.g., not properly releasing resources), repeated calls in a loop could exacerbate memory consumption over time.
Any suggestions?
Do you get the same issue when running other engines or datasets? Or is it only this combination of arguments?
I tried it with LibriSpeech/test-clean and got the same result. Here is a screengrab of my activity monitor before the crash.
However, when I try with WHISPER_SMALL I don't have the issue.
I see. Whisper large is a big model and takes a lot of RAM to run. By default we run a thread for each CPU count, and it has a copy of the model on each thread. Try reducing the amount of threads used in the benchmark using the --num-workers argument? Try something like 7 or 8? I imagine then it will run fine.
Closing due to inactivity