From c4cf399755ebacc19f81798a90d1cedaa98ef56a Mon Sep 17 00:00:00 2001
From: Xavier Hallade <me@ph0b.com>
Date: Thu, 27 Mar 2025 17:18:30 +0100
Subject: [PATCH] Cycles: oneAPI: Re-enable -ffast-math

The initial limitation preventing from using -ffast-math, worked around
in 09df1f4cafb996802af4c89ab8af0630f750d599, got fixed upstream in LLVM
and the fix is part of current DPC++ compiler:
https://github.com/intel/llvm/commit/63ecd2a72523fa591aacf54d310478aabcd30d08

We're now able to go back to using -ffast-math, which helps simplifying
the set of compiler flags.

No performance nor conformance change is expected from this change (most
of the gain is achieved already with the use of -cl-fast-relaxed-math
since 284b89a0a35a2dfbf9bc54669e382f446e7cccc5) and this has been
verified on Arc B580 under Windows.
---
 intern/cycles/kernel/CMakeLists.txt | 7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)

diff --git a/intern/cycles/kernel/CMakeLists.txt b/intern/cycles/kernel/CMakeLists.txt
index 5791ba4a354..1543e7e6fc5 100644
--- a/intern/cycles/kernel/CMakeLists.txt
+++ b/intern/cycles/kernel/CMakeLists.txt
@@ -968,12 +968,7 @@ if(WITH_CYCLES_DEVICE_ONEAPI)
     -shared
     -DWITH_ONEAPI
     -O2
-    -fno-fast-math
-    -ffp-contract=fast
-    -fassociative-math
-    -freciprocal-math
-    -fno-signed-zeros
-    -ffinite-math-only
+    -ffast-math
     -D__KERNEL_LOCAL_ATOMIC_SORT__
     -o"${cycles_kernel_oneapi_lib}"
     -I"${CMAKE_CURRENT_SOURCE_DIR}/.."