debugging

driver/device_implicit_gemm_convolution_2_chwn_cyxk_khwn.hpp

@@ -191,7 +191,7 @@ void device_implicit_gemm_convolution_2_chwn_cyxk_khwn(InDesc,
     constexpr index_t WeiBlockCopyDataPerRead = 4;
 
     constexpr index_t BlockSize = 256;
-#elif 1
+#elif 0
     // 1x1, 14x14, Vega 20, disable lds_double_buffer, enable register double buffer
     constexpr index_t BPerBlock = 64;
     constexpr index_t KPerBlock = 128;
@@ -266,7 +266,7 @@ void device_implicit_gemm_convolution_2_chwn_cyxk_khwn(InDesc,
     for(index_t i = 0; i < nrepeat; ++i)
     {
         constexpr auto gridwise_conv =
-#if 0
+#if 1
             GridwiseConvolutionImplicitGemm_v2_chwn_cyxk_khwn
 #else
             GridwiseConvolutionImplicitGemm_v2_chwn_cyxk_khwn_lds_double_buffer

src/include/blockwise_2d_tensor_op.hip.hpp

@@ -1,6 +1,6 @@
 #pragma once
+#include "common.hip.hpp"
 #include "ConstantTensorDescriptor.hip.hpp"
-#include "inline_asm.hpp"
 
 template <index_t BlockSize, class Float, class DstDesc, class F>
 __device__ void

src/include/blockwise_gemm.hip.hpp

@@ -1,4 +1,5 @@
 #pragma once
+#include "common.hip.hpp"
 #include "threadwise_gemm.hip.hpp"
 
 // if following number are power of 2, index calculation shall be greatly reduced:

src/include/common.hip.hpp

@@ -5,6 +5,10 @@
 #include "Array.hip.hpp"
 #include "functional.hip.hpp"
 
+#if DEVICE_BACKEDN_HIP
+#include "inline_asm.hpp"
+#endif
+
 __device__ index_t get_thread_local_1d_id() { return threadIdx.x; }
 
 __device__ index_t get_block_1d_id() { return blockIdx.x; }

src/include/data_type.hip.hpp

@@ -15,7 +15,11 @@ struct vector_type<float, 1>
 template <>
 struct vector_type<float, 2>
 {
+#if 1
     typedef float MemoryType __attribute__((ext_vector_type(2)));
+#else
+    using MemoryType = float2;
+#endif
 
     __host__ __device__ static MemoryType Pack(float s0, float s1)
     {
@@ -34,7 +38,11 @@ struct vector_type<float, 2>
 template <>
 struct vector_type<float, 4>
 {
+#if 1
     typedef float MemoryType __attribute__((ext_vector_type(4)));
+#else
+    using MemoryType = float4;
+#endif
 };
 
 #if 0

src/include/gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn.hip.hpp

@@ -222,21 +222,6 @@ struct GridwiseConvolutionImplicitGemm_v2_chwn_cyxk_khwn
 
             blockwise_in_copy.RunStoreRegisterClipboard(p_in_register_clipboard, p_in_block);
             blockwise_wei_copy.RunStoreRegisterClipboard(p_wei_register_clipboard, p_wei_block);
-#elif 1
-            Float4 tmp_in, tmp_wei;
-            Float4* glb_in_p =
-                (Float4*)(p_in_global_block_offset + blockwise_in_copy.mSrcMyThreadOffset);
-            Float4* loc_in_p = (Float4*)(p_in_block + blockwise_in_copy.mDstMyThreadOffset);
-
-            Float4* glb_wei_p =
-                (Float4*)(p_wei_global_block_offset + blockwise_wei_copy.mSrcMyThreadOffset);
-            Float4* loc_wei_p = (Float4*)(p_wei_block + blockwise_wei_copy.mDstMyThreadOffset);
-
-            global_load(tmp_in, glb_in_p);
-            global_load(tmp_wei, glb_wei_p);
-            vmcnt(0);
-            ds_write_b128(tmp_in, loc_in_p);
-            ds_write_b128(tmp_wei, loc_wei_p);
 #endif
 
             __syncthreads();
@@ -247,11 +232,11 @@ struct GridwiseConvolutionImplicitGemm_v2_chwn_cyxk_khwn
             {
                 for(index_t x = 0; x < X; ++x)
                 {
-#if 0
+#if 1
                     blockwise_gemm.Run
 #elif 0
                     blockwise_gemm.Run_RegisterDoubleBuffer
-#elif 1
+#elif 0
                     blockwise_gemm.Run_asm
 #endif
                     (p_wei_block + wei_cyxk_block_desc.Get1dIndex(0, y, x, 0),

src/include/threadwise_gemm.hip.hpp

@@ -1,7 +1,5 @@
 #pragma once
 
-#include "inline_asm.hpp"
-
 template <class Float, class SrcMatrix, class DstMatrix, index_t NRow, index_t NCol>
 __device__ void threadwise_matrix_copy(SrcMatrix,
                                        const Float* __restrict__ p_src,