Merge branch 'develop' into workspace_size

src/targets/gpu/include/migraphx/gpu/lrn.hpp

@@ -26,7 +26,7 @@
 
 #include <migraphx/shape.hpp>
 #include <migraphx/reflect.hpp>
-#include <migraphx/gpu/miopen.hpp>
+#include <migraphx/gpu/context.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

src/targets/gpu/include/migraphx/gpu/prefuse_ops.hpp

@@ -25,7 +25,7 @@
 #define MIGRAPHX_GUARD_GPU_PREFUSE_OPS_HPP
 
 #include <migraphx/config.hpp>
-#include <migraphx/gpu/context.hpp>
+#include <string>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

src/targets/gpu/include/migraphx/gpu/reverse.hpp

@@ -27,7 +27,7 @@
 #include <migraphx/argument.hpp>
 #include <migraphx/reflect.hpp>
 #include <migraphx/op/reverse.hpp>
-#include <migraphx/gpu/miopen.hpp>
+#include <migraphx/gpu/context.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

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

@@ -24,22 +24,10 @@
 #ifndef MIGRAPHX_GUARD_RTGLIB_SOFTMAX_HPP
 #define MIGRAPHX_GUARD_RTGLIB_SOFTMAX_HPP
 
-#include <migraphx/gpu/lowering.hpp>
-#include <migraphx/manage_ptr.hpp>
-#include <migraphx/instruction.hpp>
 #include <migraphx/op/softmax.hpp>
-#include <migraphx/generate.hpp>
-#include <migraphx/shape_for_each.hpp>
-#include <migraphx/config.hpp>
-#include <migraphx/gpu/miopen.hpp>
-#include <migraphx/gpu/hip.hpp>
-#include <migraphx/dfor.hpp>
-#include <migraphx/gpu/device/contiguous.hpp>
-#include <migraphx/gpu/device/add.hpp>
-#include <migraphx/iterator_for.hpp>
-#include <migraphx/gpu/rocblas.hpp>
+#include <migraphx/shape.hpp>
+#include <migraphx/reflect.hpp>
 #include <migraphx/gpu/context.hpp>
-#include <utility>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

src/targets/gpu/include/migraphx/gpu/sync_device.hpp

@@ -25,8 +25,6 @@
 #define MIGRAPHX_GUARD_RTGLIB_GPU_SYNC_DEVICE_HPP
 
 #include <string>
-#include <migraphx/instruction_ref.hpp>
-#include <migraphx/gpu/context.hpp>
 #include <migraphx/config.hpp>
 
 namespace migraphx {

src/targets/gpu/include/migraphx/gpu/topk.hpp

@@ -27,7 +27,7 @@
 #include <migraphx/argument.hpp>
 #include <migraphx/reflect.hpp>
 #include <migraphx/op/topk.hpp>
-#include <migraphx/gpu/miopen.hpp>
+#include <migraphx/gpu/context.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

src/targets/gpu/jit/gathernd.cpp

@@ -27,13 +27,6 @@
 
 #include <migraphx/gpu/compile_hip_code_object.hpp>
 #include <migraphx/gpu/compile_hip.hpp>
-#include <migraphx/ranges.hpp>
-#include <migraphx/reduce_dims.hpp>
-#include <migraphx/stringutils.hpp>
-#include <migraphx/dead_code_elimination.hpp>
-#include <migraphx/eliminate_common_subexpression.hpp>
-#include <migraphx/module.hpp>
-#include <migraphx/pass_manager.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

src/targets/gpu/jit/layernorm.cpp

@@ -27,8 +27,6 @@
 #include <migraphx/gpu/compile_hip.hpp>
 #include <migraphx/gpu/compile_gen.hpp>
 
-#include <migraphx/cpp_generator.hpp>
-#include <migraphx/ranges.hpp>
 #include <migraphx/reduce_dims.hpp>
 #include <migraphx/stringutils.hpp>
 

src/targets/gpu/jit/pointwise.cpp

@@ -26,16 +26,7 @@
 #include <migraphx/gpu/compile_hip_code_object.hpp>
 #include <migraphx/gpu/compile_hip.hpp>
 #include <migraphx/gpu/compile_gen.hpp>
-
-#include <migraphx/cpp_generator.hpp>
-#include <migraphx/ranges.hpp>
 #include <migraphx/reduce_dims.hpp>
-#include <migraphx/permutation.hpp>
-#include <migraphx/stringutils.hpp>
-#include <migraphx/dead_code_elimination.hpp>
-#include <migraphx/eliminate_common_subexpression.hpp>
-#include <migraphx/module.hpp>
-#include <migraphx/pass_manager.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

src/targets/gpu/jit/reduce.cpp

@@ -26,15 +26,7 @@
 #include <migraphx/gpu/compile_hip_code_object.hpp>
 #include <migraphx/gpu/compile_hip.hpp>
 #include <migraphx/gpu/compile_gen.hpp>
-
-#include <migraphx/cpp_generator.hpp>
-#include <migraphx/ranges.hpp>
 #include <migraphx/reduce_dims.hpp>
-#include <migraphx/stringutils.hpp>
-#include <migraphx/dead_code_elimination.hpp>
-#include <migraphx/eliminate_common_subexpression.hpp>
-#include <migraphx/module.hpp>
-#include <migraphx/pass_manager.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

src/targets/gpu/jit/roialign.cpp

@@ -24,16 +24,7 @@
 #include <migraphx/gpu/compiler.hpp>
 #include <migraphx/gpu/compile_hip_code_object.hpp>
 #include <migraphx/gpu/context.hpp>
-
 #include <migraphx/gpu/compile_hip.hpp>
-#include <migraphx/cpp_generator.hpp>
-#include <migraphx/ranges.hpp>
-#include <migraphx/reduce_dims.hpp>
-#include <migraphx/stringutils.hpp>
-#include <migraphx/dead_code_elimination.hpp>
-#include <migraphx/eliminate_common_subexpression.hpp>
-#include <migraphx/module.hpp>
-#include <migraphx/pass_manager.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

src/targets/gpu/jit/scatternd.cpp

@@ -24,16 +24,8 @@
 #include <migraphx/gpu/compiler.hpp>
 #include <migraphx/make_op.hpp>
 #include <migraphx/gpu/context.hpp>
-
 #include <migraphx/gpu/compile_hip_code_object.hpp>
 #include <migraphx/gpu/compile_hip.hpp>
-#include <migraphx/ranges.hpp>
-#include <migraphx/reduce_dims.hpp>
-#include <migraphx/stringutils.hpp>
-#include <migraphx/dead_code_elimination.hpp>
-#include <migraphx/eliminate_common_subexpression.hpp>
-#include <migraphx/module.hpp>
-#include <migraphx/pass_manager.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

src/targets/gpu/jit/softmax.cpp

@@ -26,15 +26,7 @@
 #include <migraphx/gpu/compile_hip_code_object.hpp>
 #include <migraphx/gpu/compile_hip.hpp>
 #include <migraphx/gpu/compile_gen.hpp>
-
-#include <migraphx/cpp_generator.hpp>
-#include <migraphx/ranges.hpp>
 #include <migraphx/reduce_dims.hpp>
-#include <migraphx/stringutils.hpp>
-#include <migraphx/dead_code_elimination.hpp>
-#include <migraphx/eliminate_common_subexpression.hpp>
-#include <migraphx/module.hpp>
-#include <migraphx/pass_manager.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {

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

@@ -163,7 +163,7 @@ constexpr Iterator1 search(Iterator1 first, Iterator1 last, Iterator2 s_first, I
             {
                 return last;
             }
-            if(!(*it == *s_it))
+            if(not(*it == *s_it))
             {
                 break;
             }

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

@@ -153,7 +153,7 @@ struct array
         return true;
     }
 
-    friend constexpr bool operator!=(const array& x, const array& y) { return !(x == y); }
+    friend constexpr bool operator!=(const array& x, const array& y) { return not(x == y); }
     // This uses the product order rather than lexical order
     friend constexpr bool operator<(const array& x, const array& y)
     {

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

@@ -73,10 +73,10 @@ MIGRAPHX_INTEGRAL_CONSTANT_BINARY_OP(!=)
 MIGRAPHX_INTEGRAL_CONSTANT_BINARY_OP(&)
 MIGRAPHX_INTEGRAL_CONSTANT_BINARY_OP(^)
 MIGRAPHX_INTEGRAL_CONSTANT_BINARY_OP(|)
-MIGRAPHX_INTEGRAL_CONSTANT_BINARY_OP(&&)
-MIGRAPHX_INTEGRAL_CONSTANT_BINARY_OP(||)
+MIGRAPHX_INTEGRAL_CONSTANT_BINARY_OP(and)
+MIGRAPHX_INTEGRAL_CONSTANT_BINARY_OP(or)
 
-MIGRAPHX_INTEGRAL_CONSTANT_UNARY_OP(!)
+MIGRAPHX_INTEGRAL_CONSTANT_UNARY_OP(not )
 MIGRAPHX_INTEGRAL_CONSTANT_UNARY_OP(~)
 MIGRAPHX_INTEGRAL_CONSTANT_UNARY_OP(+)
 MIGRAPHX_INTEGRAL_CONSTANT_UNARY_OP(-)

src/targets/gpu/lowering.cpp

@@ -27,42 +27,24 @@
 #include <migraphx/instruction.hpp>
 #include <migraphx/make_op.hpp>
 
-#include <migraphx/op/abs.hpp>
-#include <migraphx/op/batch_norm_inference.hpp>
 #include <migraphx/op/convolution.hpp>
 #include <migraphx/op/deconvolution.hpp>
 #include <migraphx/op/dot.hpp>
-#include <migraphx/op/elu.hpp>
 #include <migraphx/op/if_op.hpp>
-#include <migraphx/op/leaky_relu.hpp>
-#include <migraphx/op/lrn.hpp>
-#include <migraphx/op/pooling.hpp>
 #include <migraphx/op/reshape.hpp>
 #include <migraphx/op/quant_convolution.hpp>
 #include <migraphx/op/quant_dot.hpp>
 
-#include <migraphx/gpu/abs.hpp>
 #include <migraphx/gpu/batch_norm_inference.hpp>
 #include <migraphx/gpu/context.hpp>
 #include <migraphx/gpu/convolution.hpp>
 #include <migraphx/gpu/deconvolution.hpp>
 #include <migraphx/gpu/device_name.hpp>
-#include <migraphx/gpu/elu.hpp>
-#include <migraphx/gpu/equal.hpp>
 #include <migraphx/gpu/gemm.hpp>
-#include <migraphx/gpu/greater.hpp>
 #include <migraphx/gpu/int8_conv_pack.hpp>
-#include <migraphx/gpu/leaky_relu.hpp>
-#include <migraphx/gpu/less.hpp>
-#include <migraphx/gpu/logical_and.hpp>
-#include <migraphx/gpu/logical_or.hpp>
-#include <migraphx/gpu/logical_xor.hpp>
-#include <migraphx/gpu/lrn.hpp>
 #include <migraphx/gpu/miopen.hpp>
 #include <migraphx/gpu/quant_convolution.hpp>
 #include <migraphx/gpu/rocblas.hpp>
-#include <migraphx/gpu/unary_not.hpp>
-#include <migraphx/gpu/where.hpp>
 #include <migraphx/gpu/compiler.hpp>
 #include <migraphx/iterator_for.hpp>
 #include <migraphx/program.hpp>
@@ -341,7 +323,7 @@ struct miopen_apply
             catch(migraphx::exception&)
             {
                 // In case no solver supports the default format, retry using the other format.
-                compile_quant_conv_with_format(!int8_x4_format);
+                compile_quant_conv_with_format(not int8_x4_format);
             }
 
             auto args      = ins->inputs();

src/targets/gpu/mlir.cpp

@@ -78,7 +78,7 @@ struct mlir_handle
 
         friend bool operator==(ptr x, ptr y) { return x.get_value() == y.get_value(); }
 
-        friend bool operator!=(ptr x, ptr y) { return !(x == y); }
+        friend bool operator!=(ptr x, ptr y) { return not(x == y); }
         T obj{};
     };
 
@@ -503,7 +503,7 @@ struct mlir_program
                 pp =
                     problem_params{ins->get_operator(), to_shapes(ins->inputs()), ins->get_shape()};
                 std::string tuned = get_tune_params();
-                if(!tuned.empty())
+                if(not tuned.empty())
                     ops.add_attributes({{"perf_config", tuned}});
                 // check if HW supports xdlops
                 if(contains(get_xdlops_archs(), target_name))

src/targets/gpu/pack_int8_args.cpp

@@ -154,7 +154,7 @@ void pack_int8_args::apply(module& m) const
 
             bool transa = inputs[0]->get_shape().transposed();
             bool transb = inputs[1]->get_shape().transposed();
-            if(!transb)
+            if(not transb)
             {
                 auto packed_b = m.insert_instruction(
                     ins, make_op("hip::allocate", {{"shape", to_value(inputs[1]->get_shape())}}));

src/targets/gpu/prefuse_ops.cpp

@@ -23,6 +23,7 @@
  */
 #include <migraphx/gpu/prefuse_ops.hpp>
 #include <migraphx/match/layernorm.hpp>
+#include <migraphx/check_shapes.hpp>
 #include <migraphx/make_op.hpp>
 #include <migraphx/register_op.hpp>