clang format

src/targets/gpu/device/add.cpp

@@ -11,27 +11,28 @@ namespace device {
 __global__ void add_kernel(__half* a, __half* b, __half* r, int n)
 {
     int tid = blockIdx.x * blockDim.x + threadIdx.x;
-    if (tid < n)
+    if(tid < n)
     {
-        r[tid] = a[tid] + b[tid%768];
+        r[tid] = a[tid] + b[tid % 768];
     }
 }
 
 void add(hipStream_t stream, const argument& result, const argument& arg1, const argument& arg2)
 {
     auto s2 = arg2.get_shape();
-    if (s2.element_space() == 768 and s2.type() == shape::half_type)
+    if(s2.element_space() == 768 and s2.type() == shape::half_type)
     {
-        auto elem_num = s2.elements();
+        auto elem_num  = s2.elements();
         int block_size = 1024;
-        int block_num = (elem_num + block_size - 1) / block_size;
+        int block_num  = (elem_num + block_size - 1) / block_size;
         add_kernel<<<block_num, block_size>>>(reinterpret_cast<__half*>(arg1.data()),
-                                              reinterpret_cast<__half*>(arg2.data()), 
-                                              reinterpret_cast<__half*>(result.data()), elem_num);
+                                              reinterpret_cast<__half*>(arg2.data()),
+                                              reinterpret_cast<__half*>(result.data()),
+                                              elem_num);
     }
     else
     {
-        nary(stream, result, arg1, arg2)([](auto x, auto y) __device__ { return x + y; });        
+        nary(stream, result, arg1, arg2)([](auto x, auto y) __device__ { return x + y; });
     }
 }
 

src/targets/gpu/device/contiguous.cpp

@@ -14,7 +14,7 @@ void contiguous_nonstandard(hipStream_t stream, const argument& result, const ar
     visit_all(result, arg)([&](auto output_v, auto input_v) {
         hip_visit_views(output_v, input_v, s)([&](auto output, auto input, auto standard_shape) {
             gs_launch(stream, s.elements())([=](auto i) __device__ {
-                auto idx = standard_shape.multi(i);
+                auto idx    = standard_shape.multi(i);
                 output[idx] = input[idx];
             });
             // mi_gs_launch(stream,
@@ -34,8 +34,8 @@ void contiguous_packed(hipStream_t stream, const argument& result, const argumen
     //         auto* output      = device_cast(output_v.data());
     //         const __half2* input2 = reinterpret_cast<__half2*>(input_v.data());
     //         __half2* output2 = reinterpret_cast<__half2*>(output_v.data());
-    //         gs_launch(stream, nelements / 2)([=](auto i) __device__ { 
-    //             output2[i] = input2[i]; 
+    //         gs_launch(stream, nelements / 2)([=](auto i) __device__ {
+    //             output2[i] = input2[i];
     //             if (i == 0 and (nelements % 2) == 1)
     //             {
     //                 output[nelements - 1] = input[nelements - 1];
@@ -45,11 +45,11 @@ void contiguous_packed(hipStream_t stream, const argument& result, const argumen
     // }
     // else
     // {
-        visit_all(result, arg)([&](auto output_v, auto input_v) {
-            const auto* input = device_cast(input_v.data());
-            auto* output      = device_cast(output_v.data());
-            gs_launch(stream, nelements)([=](auto i) __device__ { output[i] = input[i]; });
-        });        
+    visit_all(result, arg)([&](auto output_v, auto input_v) {
+        const auto* input = device_cast(input_v.data());
+        auto* output      = device_cast(output_v.data());
+        gs_launch(stream, nelements)([=](auto i) __device__ { output[i] = input[i]; });
+    });
     // }
 }
 

src/targets/gpu/device/include/migraphx/gpu/device/multi_index.hpp

@@ -57,8 +57,8 @@ inline auto mi_nglobal(const hip_shape<N>& s, index_int nlocal)
 {
     assert(s.standard);
     assert(s.elements() > 0);
-    index_int n       = s.elements();
-    index_int groups  = (n + nlocal - 1) / nlocal;
+    index_int n      = s.elements();
+    index_int groups = (n + nlocal - 1) / nlocal;
     // change the max group num to 1 Million
     index_int nglobal = std::min<index_int>((1 << 20), groups) * nlocal;
 

src/targets/gpu/device/mul.cpp

@@ -11,28 +11,28 @@ namespace device {
 __global__ void mul_kernel(__half* a, __half* b, __half* r, int n)
 {
     int tid = blockIdx.x * blockDim.x + threadIdx.x;
-    if (tid < n)
+    if(tid < n)
     {
-        r[tid] = a[tid] * b[tid%768];
+        r[tid] = a[tid] * b[tid % 768];
     }
 }
 
-
 void mul(hipStream_t stream, const argument& result, const argument& arg1, const argument& arg2)
 {
     auto s2 = arg2.get_shape();
-    if (s2.element_space() == 768 and s2.type() == shape::half_type)
+    if(s2.element_space() == 768 and s2.type() == shape::half_type)
     {
-        auto elem_num = s2.elements();
+        auto elem_num  = s2.elements();
         int block_size = 1024;
-        int block_num = (elem_num + block_size - 1) / block_size;
+        int block_num  = (elem_num + block_size - 1) / block_size;
         mul_kernel<<<block_num, block_size>>>(reinterpret_cast<__half*>(arg1.data()),
-                                              reinterpret_cast<__half*>(arg2.data()), 
-                                              reinterpret_cast<__half*>(result.data()), elem_num);
+                                              reinterpret_cast<__half*>(arg2.data()),
+                                              reinterpret_cast<__half*>(result.data()),
+                                              elem_num);
     }
     else
     {
-        nary(stream, result, arg1, arg2)([](auto x, auto y) __device__ { return x * y; });        
+        nary(stream, result, arg1, arg2)([](auto x, auto y) __device__ { return x * y; });
     }
 }