https://github.com/rjodinchr created
https://github.com/llvm/llvm-project/pull/135002
clspv is already handling generation of fp16. This implementation is preventing
clspv from making the best choice to use an emulation on top of fp32-fma, or
the native fp16-fma, depending on the command-line arguments.
>From 30559f68db88ce1319a29bee5b295195a81b5934 Mon Sep 17 00:00:00 2001
From: Romaric Jodin <rjo...@google.com>
Date: Wed, 9 Apr 2025 15:17:17 +0200
Subject: [PATCH] libclc: clspv: fma: remove fp16 implementation
clspv is already handling generation of fp16. This implementation is
preventing clspv from making the best choice to use an emulation on
top of fp32-fma, or the native fp16-fma, depending on the command-line
arguments.
---
libclc/clspv/lib/math/fma.cl | 10 ----------
1 file changed, 10 deletions(-)
diff --git a/libclc/clspv/lib/math/fma.cl b/libclc/clspv/lib/math/fma.cl
index 9b4710122bfad..3c67315d53f89 100644
--- a/libclc/clspv/lib/math/fma.cl
+++ b/libclc/clspv/lib/math/fma.cl
@@ -12,13 +12,3 @@
_CLC_DEFINE_TERNARY_BUILTIN(float, fma, __clc_sw_fma, float, float, float)
-#ifdef cl_khr_fp16
-
-#pragma OPENCL EXTENSION cl_khr_fp16 : enable
-
-_CLC_DEF _CLC_OVERLOAD half fma(half a, half b, half c) {
- return (half)mad((float)a, (float)b, (float)c);
-}
-_CLC_TERNARY_VECTORIZE(_CLC_DEF _CLC_OVERLOAD, half, fma, half, half, half)
-
-#endif
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
