[SYCL][CUDA] Add no-fast-math to tests that rely on it. (#9889)

Same as #9419. This updates a few tests that were missed where the
`fast-math` flag affects at least the cuda backend. These tests assume
`no-fast-math` precision.

---------

Signed-off-by: JackAKirk <[email protected]>
Co-authored-by: JackAKirk <[email protected]>

Author: JackAKirk

sycl/test-e2e/BFloat16/bfloat16_builtins.cpp

@@ -1,5 +1,6 @@
+// DEFINE: %{mathflags} = %if cl_options %{/clang:-fno-fast-math%} %else %{-fno-fast-math%}
 // REQUIRES: aspect-ext_oneapi_bfloat16_math_functions
-// RUN: %clangxx -fsycl -fsycl-targets=%{sycl_triple} %if any-device-is-cuda %{ -Xsycl-target-backend --cuda-gpu-arch=sm_80 %} %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%{sycl_triple} %if any-device-is-cuda %{ -Xsycl-target-backend --cuda-gpu-arch=sm_80 %} %s -o %t.out %{mathflags}
 // RUN: %{run} %t.out
 // Currently the feature isn't supported on FPGA.
 // UNSUPPORTED: accelerator

sycl/test-e2e/DeviceLib/built-ins/scalar_relational.cpp

@@ -1,4 +1,5 @@
-// RUN: %{build} -fsycl-device-code-split=per_kernel -o %t.out
+// DEFINE: %{mathflags} = %if cl_options %{/clang:-fno-fast-math%} %else %{-fno-fast-math%}
+// RUN: %{build} -fsycl-device-code-split=per_kernel -o %t.out %{mathflags}
 // RUN: %{run} %t.out
 
 #include <sycl/sycl.hpp>

sycl/test-e2e/DeviceLib/built-ins/vector_relational.cpp

@@ -1,6 +1,7 @@
 // FIXME unsupported on windows (opencl and level-zero) until fix of libdevice
 // UNSUPPORTED: windows && (opencl || level_zero)
-// RUN: %{build} -o %t.out
+// DEFINE: %{mathflags} = %if cl_options %{/clang:-fno-fast-math%} %else %{-fno-fast-math%}
+// RUN: %{build} -o %t.out %{mathflags}
 // RUN: %{run} %t.out
 
 #include <sycl/sycl.hpp>