Merge branch 'develop' into fix_for_multiconfig_generators

src/targets/gpu/kernels/include/migraphx/kernels/softmax.hpp

@@ -43,7 +43,7 @@ __device__ void softmax(Input input1, Output output)
         auto exp_in = r.inner([&](auto x) { return migraphx::exp(x - c); })(input);
         auto batch_sum =
             r.reduce(op::sum{}, 0, [](auto x) { return migraphx::convert<float>(x); })(exp_in);
-        r.inner([&](auto& y, auto x) { y = x / batch_sum; })(output, exp_in);
+        r.inner([&](auto& y, auto x) { y = implicit_conversion(x / batch_sum); })(output, exp_in);
     });
 }
 

src/targets/gpu/kernels/include/migraphx/kernels/tensor_view.hpp

@@ -27,6 +27,7 @@
 #include <migraphx/kernels/shape.hpp>
 #include <migraphx/kernels/debug.hpp>
 #include <migraphx/kernels/iota_iterator.hpp>
+#include <migraphx/kernels/float8.hpp>
 
 namespace migraphx {