We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 88bc845 commit eb7ea6bCopy full SHA for eb7ea6b
include/cute/tensor_sg.hpp
@@ -115,7 +115,11 @@ CUTE_HOST_DEVICE
115
constexpr auto
116
make_subgroup_tensor(Layout<Shape,Stride> const& sg_layout)
117
{
118
- return make_subgroup_tensor(make_fragment_like<T>(sg_layout(0,_)), sg_layout);
+ using _SG = intel::_SGSize;
119
+ auto ilayout = make_layout(make_shape(_SG{}, size(sg_layout) / _SG{}),
120
+ make_stride(_1{}, _16{}));
121
+ auto sv_layout = sg_layout.compose(ilayout);
122
+ return make_subgroup_tensor(make_fragment_like<T>(sv_layout(0,_)), sv_layout);
123
}
124
125
template <typename T, class... Args>
0 commit comments