Brad Larson

More will be forthcoming regarding this. There's still going to be good reasons to use OpenGL / OpenGL ES to target older devices, Linux, and other targets, but yeah, the...

So this is what I was hinting at: https://github.com/BradLarson/GPUImage3 I'm working with Janie Clayton to translate all of GPUImage 2 to Metal, building it as an OpenGL-free project to avoid...

@mblaughton - The case of supporting either is something we're debating right now. We want to have a pure Metal version of the framework, to avoid OpenGL deprecation warnings and...

@omarojo - Yes, the shaders will be in .metal files using the Metal Shading Language. Some of the operations may also have parallel paths that can use built-in Metal Performance...

I described some thoughts about this in the design review [here](https://docs.google.com/document/d/1eIFrogM7HYn-hIMvrVcoOGjC15KCxDEyn60GNX1_DJI/edit#heading=h.2kwalj5sazl4), so I'll pull out the bullet points I'd listed there: - **Compilation**. As a basic requirement, make sure that...

@Shashi456 - For performance benchmarks, we are building out public and internal infrastructure that will run those independently of the normal unit tests. The [Benchmarks target](https://github.com/tensorflow/swift-models/tree/master/Benchmarks) is one aspect of...

This is a goal, and it's one reason why we've been working on a robust checkpoint loader. It's also a reason why we've created a Google Cloud Storage bucket for...

@rahulbhalley - Sorry for the delay. In the time since my last update above, we had implemented the remaining checkpoint reading and writing functionality as basic components. PR #631 ties...

Sorry about the lacking documentation, that's something I was hoping to get to with overall improvements to the use of pretrained checkpoints. For now, the best example is in the...