|Low-Level Thinking in High-Level Shading Languages
|Track / Format:
Did you know free users get access to 30% of content from the last 2 years?
|As programmable shading has grown increasingly sophisticated, an industry-wide problem is happening: shader authors are losing touch with the underlying hardware. This lecture argues that low-level thinking is still a relevant practice, and shows examples of why this is the case. The connection between high-level shading constructs and the resulting hardware instructions that are generated will be discussed, as well as what the compiler is able to optimize and what the shader author must express explicitly in the shader. Numerous low-level optimization tricks are covered, which illustrate the benefits of a close-to-the-metal mindset while writing shaders in high-level languages.