Commit: dfd6055eb0798c8cba245d47cd4a3fe51270a52a
Author: Mai Lavelle
Date: Wed Feb 22 07:41:18 2017 -0500
Branches: master
https://developer.blender.org/rBdfd6055eb0798c8cba245d47cd4a3fe51270a52a
Cycles: Add more atomic operations
===================================================================
M intern/cycles/util/util_atomic.h
===================================================================
diff --git a/intern/cycles/util/util_atomic.h b/intern/cycles/util/util_atomic.h
index 433e41fbbb..6c52117ef9 100644
--- a/intern/cycles/util/util_atomic.h
+++ b/intern/cycles/util/util_atomic.h
@@ -32,6 +32,13 @@ ATOMIC_INLINE void atomic_update_max_z(size_t
*maximum_value, size_t value)
}
}
+#define atomic_add_and_fetch_float(p, x) atomic_add_and_fetch_fl((p), (x))
+
+#define atomic_fetch_and_inc_uint32(p) atomic_fetch_and_add_uint32((p), 1)
+
+#define CCL_LOCAL_MEM_FENCE 0
+#define ccl_barrier(flags) (void)0
+
#else /* __KERNEL_GPU__ */
#ifdef __KERNEL_OPENCL__
@@ -39,7 +46,7 @@ ATOMIC_INLINE void atomic_update_max_z(size_t *maximum_value,
size_t value)
/* Float atomics implementation credits:
*
http://suhorukov.blogspot.in/2011/12/opencl-11-atomic-operations-on-floating.html
*/
-ccl_device_inline void atomic_add_and_fetch_float(volatile ccl_global float
*source,
+ccl_device_inline float atomic_add_and_fetch_float(volatile ccl_global float
*source,
const float operand)
{
union {
@@ -56,10 +63,29 @@ ccl_device_inline void atomic_add_and_fetch_float(volatile
ccl_global float *sou
} while(atomic_cmpxchg((volatile ccl_global unsigned int *)source,
prev_value.int_value,
new_value.int_value) != prev_value.int_value);
+ return new_value.float_value;
}
+#define atomic_fetch_and_add_uint32(p, x) atomic_add((p), (x))
+#define atomic_fetch_and_inc_uint32(p) atomic_inc((p))
+
+#define CCL_LOCAL_MEM_FENCE CLK_LOCAL_MEM_FENCE
+#define ccl_barrier(flags) barrier(flags)
+
#endif /* __KERNEL_OPENCL__ */
+#ifdef __KERNEL_CUDA__
+
+#define atomic_add_and_fetch_float(p, x) (atomicAdd((float*)(p), (float)(x)) +
(float)(x))
+
+#define atomic_fetch_and_add_uint32(p, x) atomicAdd((unsigned int*)(p),
(unsigned int)(x))
+#define atomic_fetch_and_inc_uint32(p) atomic_fetch_and_add_uint32((p), 1)
+
+#define CCL_LOCAL_MEM_FENCE
+#define ccl_barrier(flags) __syncthreads()
+
+#endif /* __KERNEL_CUDA__ */
+
#endif /* __KERNEL_GPU__ */
#endif /* __UTIL_ATOMIC_H__ */
_______________________________________________
Bf-blender-cvs mailing list
Bf-blender-cvs@blender.org
https://lists.blender.org/mailman/listinfo/bf-blender-cvs
