Enable bfloat16 math tests on all backends (intel/llvm-test-suite#1495)

Signed-off-by: jinge90 <ge.jin@intel.com>
Co-authored-by: Alexey Bader <alexey.bader@intel.com>

SYCL/BFloat16/bfloat16_builtins.cpp

@@ -1,11 +1,11 @@
-// REQUIRES: cuda
-//
-// Currently this test fails to compile for backends other than cuda.
-// Other backends could use this test when bfloat16 math function support is
-// added.
+// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %if cuda %{ -Xsycl-target-backend --cuda-gpu-arch=sm_80 %} %s -o %t.out
+
+// Currently the feature isn't supported on FPGA.
+
+// RUN: %CPU_RUN_PLACEHOLDER %t.out
+// RUN: %GPU_RUN_PLACEHOLDER %t.out
+// TODO: add test execution on FPGA
 //
-// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out -Xsycl-target-backend --cuda-gpu-arch=sm_80
-// RUN: %t.out
 #include <sycl/sycl.hpp>
 
 #include <cmath>