Emery Berger

### 90x speedup https://github.com/plasma-umass/scalene/issues/58#issuecomment-720252786 presents the following code: ```Python for i in range(n_features): for n in range(n_samples): subgrad[i] += (- y[n] * X[n][i]) if y[n] * (np.dot(X[n], w) + b)...

### 18x speedup over original; 1.8x speedup over manual optimization https://github.com/plasma-umass/scalene/issues/58#issuecomment-812953671 presents the following code (with some edits to put the bottleneck code into a function `do_it()`: ```Python column_names_example =...

Thanks for the note. I added info to the docs that you can use this website: https://plasma-umass.org/scalene-gui/, which hosts the web UI (you can't actually run it locally without spawning...

You can now run `scalene --viewer`. In addition, Scalene now packages the profile into both a `profile.html` and `profile.json` file; the former is self-contained, and can be opened in any...

Done. Should be fixed with https://github.com/plasma-umass/scalene/commit/dec3550f8d9f921229ae5e230b104d8e87afe9ea (please install from the repo to test: `python3 -m pip install git+https://github.com/plasma-umass/scalene`).

Just a note: it is already possible to use Scalene for the specified use-case of finding who is waiting on a synchronization operation via a command-line option. As long as...

Have you tried the version in the repository? Please provide some more details - thanks.

I'm not clear why you are increasing the sampling rate, but it makes perfect sense that this distorts the results. Each sample takes a fixed amount of time to process....

Can you explain why you need absolute execution time numbers? I don't see why you need more than the utilization numbers. FWIW the latest version in the repo now adds...

Belatedly: this _should_ now be fixed. You will need to run your code with the latest version of Scalene.