Use fp16 for convolution on navi

[pfultz2]

Motivation

Since navi doesnt have fp32 wmmas, we can use fp16 instead to get better performance.

Technical Details

This only converts the wmma to fp16 but still outputs fp32. To improve accuracy beyond just vanilla fp16 it extends the weights with residuals.

Changelog Category

Add a CHANGELOG.md entry for any option other than Not Applicable

• • Added: New functionality.
• • Changed: Changes to existing functionality.
• • Removed: Functionality or support that has been removed. (Compared to a previous release)
• • Optimized: Component performance that has been optimized or improved.
• • Resolved Issues: Known issues from a previous version that have been resolved.
• • Not Applicable: This PR is not to be included in the changelog.