Skip to content

Commit 6ecf720

Browse files
committed
more information on no_dims
1 parent 656ed70 commit 6ecf720

File tree

1 file changed

+8
-4
lines changed

1 file changed

+8
-4
lines changed

clang/include/clang/Basic/AttrDocs.td

Lines changed: 8 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -7565,10 +7565,14 @@ The ``__cluster_dims__`` and `__no_cluster__`` attributes are mutually exclusive
75657565
def CUDANoClusterAttrDoc : Documentation {
75667566
let Category = DocCatDecl;
75677567
let Content = [{
7568-
In CUDA/HIP programming, the LLVM/Clang-exclusive ``__no_cluster__`` attribute can be applied to
7569-
a kernel function to indicate that the thread block cluster feature will not be enabled at both
7570-
compile time and kernel launch time. The ``__cluster_dims__`` and `__no_cluster__`` attributes
7571-
are mutually exclusive.
7568+
In CUDA/HIP programming, a kernel function can still be launched with the cluster feature
7569+
enabled at runtime, even without the ``__cluster_dims__`` attribute. The LLVM/Clang-exclusive
7570+
``__no_cluster__`` attribute can be applied to a kernel function to explicitly indicate that
7571+
the cluster feature will not be enabled either at compile time or at kernel launch time. This
7572+
allows the compiler to apply certain optimizations without assuming that clustering could be
7573+
enabled at runtime. It is undefined behavior to launch a kernel with the ``__no_cluster__``
7574+
attribute if the cluster feature is enabled at runtime. The ``__cluster_dims__`` and
7575+
``__no_cluster__`` attributes are mutually exclusive.
75727576

75737577
.. code::
75747578

0 commit comments

Comments
 (0)