================
@@ -6,32 +6,35 @@
 //
 
//===----------------------------------------------------------------------===//
 
+#include <clc/atomic/clc_atomic_fetch_add.h>
 #include <clc/opencl/atomic/atom_add.h>
-#include <clc/opencl/atomic/atomic_add.h>
+
+// Non-volatile overloads are for backward compatibility with OpenCL 1.0.
----------------
arsenm wrote:

I don't remember the history here, but isn't this backwards? Even if the 
language is volatile qualified we probably shouldn't be emitting IR with 
volatile markers 

https://github.com/llvm/llvm-project/pull/168325
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to