!5537 Fix some tiny bugs for converter in windows enviroment and conv ops

Merge pull request !5537 from liuwenhao/r0.7

build.bat

@@ -20,14 +20,14 @@ IF NOT EXIST "%BASEPATH%/build" (
     md "build"
 )
 
-cd "%BASEPATH%/build"
+cd %BASEPATH%/build
 set BUILD_PATH=%CD%
 
 IF NOT EXIST "%BUILD_PATH%/mindspore" (
     md "mindspore"
 )
 
-cd "%CD%/mindspore"
+cd %CD%/mindspore
 
 IF "%1%" == "lite" (
     call :gene_gtest
@@ -56,7 +56,7 @@ IF "%1%" == "lite" (
         echo "build fail."
         goto run_fail
     ) ELSE (
-        cd "%BASEPATH%/output"
+        cd %BASEPATH%/output
         rd /s /q _CPack_Packages
     )
 ) ELSE (
@@ -78,22 +78,22 @@ IF "%1%" == "lite" (
     )
 )
 
-cd "%BASEPATH%"
+cd %BASEPATH%
 
 goto run_eof
 
 :run_cmake
-    cd "%BUILD_PATH%/mindspore"
+    cd %BUILD_PATH%/mindspore
     cmake -DBUILD_DEVICE=on -DBUILD_CONVERTER=on -DPLATFORM_ARM64=off -DSUPPORT_TRAIN=off ^
     -DCMAKE_BUILD_TYPE=Release -DSUPPORT_GPU=off -DBUILD_MINDDATA=off -DOFFLINE_COMPILE=off ^
     -G "CodeBlocks - MinGW Makefiles" "%BASEPATH%/mindspore/lite"
 GOTO:EOF
 
 :gene_gtest
-    cd "%BASEPATH%/third_party"
+    cd %BASEPATH%/third_party
     IF EXIST googletest rd /s /q googletest
     git submodule update --init --recursive googletest
-    cd "%BUILD_PATH%/mindspore"
+    cd %BUILD_PATH%/mindspore
 GOTO:EOF
 
 :gene_protobuf
@@ -111,22 +111,22 @@ GOTO:EOF
 
 :gene_flatbuffer
     SET FLATC="%BASEPATH%/build/mindspore/_deps/flatbuffers-src/_build/flatc"
-    SET FLAT_DIR="%BASEPATH%/mindspore/lite/schema"
+    SET FLAT_DIR=%BASEPATH%/mindspore/lite/schema
     cd %FLAT_DIR%
     IF EXIST inner rd /s /q inner
     md inner
 
     %FLATC% -c -b *.fbs
-    %FLATC% -c -b --reflect-types --gen-mutable --reflect-names --gen-object-api -o %FLAT_DIR%/inner *.fbs
+    %FLATC% -c -b --reflect-types --gen-mutable --reflect-names --gen-object-api -o "%FLAT_DIR%/inner" *.fbs
 
-    SET FLAT_DIR="%BASEPATH%/mindspore/lite/tools/converter/parser/tflite"
+    SET FLAT_DIR=%BASEPATH%/mindspore/lite/tools/converter/parser/tflite
     cd %FLAT_DIR%
-    %FLATC% -c -b --reflect-types --gen-mutable --reflect-names --gen-object-api -o %FLAT_DIR% *.fbs
-    cd "%BUILD_PATH%/mindspore"
+    %FLATC% -c -b --reflect-types --gen-mutable --reflect-names --gen-object-api -o "%FLAT_DIR%" *.fbs
+    cd %BUILD_PATH%/mindspore
 GOTO:EOF
 
 :run_fail
-    cd "%BASEPATH%"
+    cd %BASEPATH%
     set errorlevel=1
 
 :run_eof

mindspore/lite/nnacl/assembly/opt/MatmulDpInt8.S

@@ -421,13 +421,37 @@ End3:
   smax v17.4s, v17.4s, v7.4s
   smax v18.4s, v18.4s, v7.4s
   smax v19.4s, v19.4s, v7.4s
-
-  // Apply the act_min bound
+  smax v20.4s, v20.4s, v7.4s
+  smax v21.4s, v21.4s, v7.4s
+  smax v22.4s, v22.4s, v7.4s
+  smax v23.4s, v23.4s, v7.4s
+  smax v24.4s, v24.4s, v7.4s
+  smax v25.4s, v25.4s, v7.4s
+  smax v26.4s, v26.4s, v7.4s
+  smax v27.4s, v27.4s, v7.4s
+  smax v28.4s, v28.4s, v7.4s
+  smax v29.4s, v29.4s, v7.4s
+  smax v30.4s, v30.4s, v7.4s
+  smax v31.4s, v31.4s, v7.4s
+
+  // Apply the act_max bound
   dup v6.4s, w9
   smin v16.4s, v16.4s, v6.4s
   smin v17.4s, v17.4s, v6.4s
   smin v18.4s, v18.4s, v6.4s
   smin v19.4s, v19.4s, v6.4s
+  smin v20.4s, v20.4s, v6.4s
+  smin v21.4s, v21.4s, v6.4s
+  smin v22.4s, v22.4s, v6.4s
+  smin v23.4s, v23.4s, v6.4s
+  smin v24.4s, v24.4s, v6.4s
+  smin v25.4s, v25.4s, v6.4s
+  smin v26.4s, v26.4s, v6.4s
+  smin v27.4s, v27.4s, v6.4s
+  smin v28.4s, v28.4s, v6.4s
+  smin v29.4s, v29.4s, v6.4s
+  smin v30.4s, v30.4s, v6.4s
+  smin v31.4s, v31.4s, v6.4s
 
   // int32 -> int16
   sqxtn v0.4h, v16.4s

mindspore/lite/nnacl/opt_op_handler.c

@@ -57,6 +57,6 @@ void MatMulRInt8_optimize_handler(const int8_t *a, const int8_t *b, int8_t *dst,
                                   int32_t *right_shift, int32_t *multiplier, int32_t output_zp, int32_t mini,
                                   int32_t maxi, bool per_channel) {
   return MatmulInt8DpNeon64(a, b, dst, UP_ROUND(row, 8), UP_ROUND(col, 8), deep_4, input_sum, bias, mini, maxi,
-                            output_zp, multiplier[0], left_shift[0], right_shift[0], row, col, col);
+                            output_zp, multiplier[0], left_shift[0], right_shift[0], row, col, stride);
 }
 #endif

mindspore/lite/src/runtime/kernel/arm/base/split_base.cc

@@ -28,7 +28,11 @@ using mindspore::lite::RET_OK;
 using mindspore::schema::PrimitiveType_Split;
 
 namespace mindspore::kernel {
-int SplitBaseCPUKernel::Init() { return RET_OK; }
+int SplitBaseCPUKernel::Init() {
+  auto split_dim = param->split_dim_;
+  param->split_dim_ = split_dim >= 0 ? split_dim : in_tensors_.front()->shape().size() + split_dim;
+  return RET_OK;
+}
 
 int SplitBaseCPUKernel::ReSize() {
   auto in_tensor = in_tensors_.front();

mindspore/lite/src/runtime/kernel/arm/int8/convolution_1x1_int8.cc

@@ -38,7 +38,7 @@ Convolution1x1Int8CPUKernel::~Convolution1x1Int8CPUKernel() {
     matmul_param_ = nullptr;
   }
   if (packed_weight_ != nullptr) {
-    delete packed_weight_;
+    free(packed_weight_);
     packed_weight_ = nullptr;
   }
   FreeResizeBuf();

mindspore/lite/src/runtime/kernel/arm/int8/convolution_depthwise_int8.cc

@@ -60,6 +60,7 @@ int ConvolutionDepthwiseInt8CPUKernel::InitWeightBias() {
   for (int i = 0; i < weight_tensor->ElementsNum(); i++) {
     packed_weight_[i] = (int16_t)(tmp_weight[i] - weight_zp);
   }
+  free(tmp_weight);
 
   bias_data_ = reinterpret_cast<int32_t *>(malloc(channel * sizeof(int32_t)));
   if (bias_data_ == nullptr) {

mindspore/lite/src/runtime/kernel/arm/int8/convolution_int8.cc

@@ -402,7 +402,7 @@ kernel::LiteKernel *CpuConvInt8KernelCreator(const std::vector<lite::tensor::Ten
   if (kernel_h == 3 && kernel_w == 3 && stride_h == 1 && stride_w == 1 && dilation_h == 1 && dilation_w == 1) {
     kernel = new (std::nothrow) kernel::ConvolutionInt8CPUKernel(opParameter, inputs, outputs, ctx, primitive);
   } else if (kernel_h == 1 && kernel_w == 1 && filter_quant_size == 1) {
-    kernel = new (std::nothrow) kernel::ConvolutionInt8CPUKernel(opParameter, inputs, outputs, ctx, primitive);
+    kernel = new (std::nothrow) kernel::Convolution1x1Int8CPUKernel(opParameter, inputs, outputs, ctx, primitive);
   } else {
     kernel = new (std::nothrow) kernel::ConvolutionInt8CPUKernel(opParameter, inputs, outputs, ctx, primitive);
   }