From 51ad07d1fbffd41c758eba172aa77ebba3204924 Mon Sep 17 00:00:00 2001 From: Yong He Date: Sun, 23 Feb 2025 10:31:05 -0800 Subject: Improve performance when compiling small shaders. (#6396) Improve performance when compiling small shaders. Avoid copying witness table entries that are not getting used during linking. Avoid copying auto-diff related decorations and derivative functions during linking, if the user modules doesn't use autodiff. Cache operator overload resolution results on global session, so each new Session doesn't need to repetitively run through overload resolution from scratch. --- docs/design/stdlib-intrinsics.md | 6 ------ 1 file changed, 6 deletions(-) (limited to 'docs/design') diff --git a/docs/design/stdlib-intrinsics.md b/docs/design/stdlib-intrinsics.md index a9369138d..2ea50cd54 100644 --- a/docs/design/stdlib-intrinsics.md +++ b/docs/design/stdlib-intrinsics.md @@ -114,12 +114,6 @@ Sections of the `expansion` string that are to be replaced are prefixed by the ` * $XH - Ray tracing hit object attribute * $P - Type-based prefix as used for CUDA and C++ targets (I8 for int8_t, F32 - float etc) -## __specialized_for_target(target) - -Specialized for target allows defining an implementation *body* for a particular target. The target is the same as is used for [__target_intrinsic](#target-intrinsic). - -A declaration can consist of multiple definitions with bodies (for each target) using, `specialized_for_target`, as well as having `target_intrinsic` if that is applicable for a target. - ## __attributeTarget(astClassName) For an attribute, specifies the AST class (and derived class) the attribute can be applied to. -- cgit v1.2.3