glalonde

Do you also stop in your tracks and write blog posts when you see suicide hotline signs on the GGB?

This is genuinely insightful.

"extrapolation" literally implies outside the extents of current knowledge.

*Overstated

"Optimize the kernel (in KernelBuilder.build_kernel) as much as possible in the available time, as measured by test_kernel_cycles on a frozen separate copy of the simulator." from perf_takehome.py

which makes sense in a portfolio, but a company only has one CEO in their basket. But I suppose if CEO elections represent stock holders with many companies, then perhaps it may still be optimized like a portfolio with multiple CEOs.

Is that true and is that sustainable? My understanding has been that only the relatively low level libraries/kernels are written in cuda and all the magical algorithms are in python using various ML libaries. It's like how intel BLAS isn't much of a moat -- there are several open source implementations and you can mix and match.

How is CUDA so sticky when most ML devs aren't writing CUDA but something several layers of abstraction above it? Why can't intel, AMD, google w/e come along and write an adapter for that lowest level to TF, pytorch or whatever is the framework of the day?

it had a gps

Atlas is also electric. An electric motor tied to a gear pump presumably.

have to appreciate the irony in worrying about the bitrate of simulated lo-fi crackling

internal internal combustion engine engine

Presumably early waymo. Ian probably worked for bot & dolly which was bought by Google.

The weird thing about this article is that it's only comparing annotation performance, which is important, but not what you should ultimately care about. If you trained a visual model using annotated lidar for ground truth, then you might expect better performance from the model than from human annotations of the image alone, and certainly better than a model trained on those annotations.

Yeah I'm kinda surprised by this. You would think there would be some industry support for emulating graphics/gpgpu pipelines to make cloud CI cheaper and simpler, but maybe it's just a much harder problem than I imagine -- to the point where mocking opengl or using dedicated hardware makes more sense.