ilemhadri

Thanks @Syllo I just checked out the branch it looks good to me I like how the units on the x-axis adapts to the user's screen size.

good for merging to master?

>@ilemhadri can you confirm that this patch fixes the problem for you? I don't have a way of testing this locally, but @jakevdp might be able to check!

@seberg I would vote in favor of just merging!

in particular, this happens inside `lax.scan` as well.

Unfortunately, that does not seem to work. Re-running the trace with ```python @jax.jit def compute(M): Z = M@M with jax.named_scope("InsideAnnotation"): with jax.profiler.TraceAnnotation("InsideAnnotation"): for _ in range(3): Z = jnp.sqrt(Z@Z) _,...

ok, so this time i'm running the following snippet (using `named_scope` only) on a V100 GPU. Unfortunately, I don't observe meaningful changes to the profiling. [another screenshot here](https://i.imgur.com/yqKwjM1.png). I can...

@mattjj is there any blocker to merging this PR now?

>Do you have bottleneck installed? Could you try xr.set_options(use_bottleneck=False)? `use_bottleneck=False` seems to solve it on the latest version of `xarray`, in the sense that the negative values are now exactly...

> You'll have to uninstall bottleneck then - the `use_bottleneck` option was added more recently. @mathause Does xarray use bottleneck in other places than rolling operations? That'd help me assess...