p47068915
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[MS][LITE][Develop] conv1x1 int8 bug : input buf malloc using allocator

This commit is contained in:
ling 2020-09-01 17:53:57 +08:00
parent 8f106d685a
commit 7781f70486
1 changed files with 19 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

45
mindspore/lite/src/runtime/kernel/arm/int8/convolution_1x1_int8.cc
View File

@ -170,36 +170,28 @@ int Convolution1x1Int8CPUKernel::InitParam() {
matmul_param_->deep_4_ = UP_ROUND(matmul_param_->deep_, C4NUM);
matmul_param_->deep_16_ = UP_ROUND(matmul_param_->deep_, C16NUM);
int row_pack_count = 0;
int col_pack_count = 0;
if (support_optimize_) {
row_pack_count = C8NUM;
col_pack_count = C8NUM;
} else {
row_pack_count = C4NUM;
col_pack_count = C4NUM;
}
/* init input sum size */
if (support_optimize_) {
if (conv_quant_arg_->per_channel_ & FILTER_PER_CHANNEL) {
input_sum_size = UP_ROUND(conv_param_->output_channel_, C8NUM) * UP_ROUND(matmul_param_->row_, C8NUM);
} else {
input_sum_size = UP_ROUND(matmul_param_->row_, C8NUM);
}
if (conv_quant_arg_->per_channel_ & FILTER_PER_CHANNEL) {
input_sum_size = UP_ROUND(matmul_param_->col_, col_pack_count) * UP_ROUND(matmul_param_->row_, row_pack_count);
} else {
if (conv_quant_arg_->per_channel_ & FILTER_PER_CHANNEL) {
input_sum_size = UP_ROUND(conv_param_->output_channel_, C4NUM) * UP_ROUND(matmul_param_->row_, C4NUM);
} else {
input_sum_size = UP_ROUND(matmul_param_->row_, C4NUM);
}
input_sum_size = UP_ROUND(matmul_param_->row_, row_pack_count);
}
if (support_optimize_) {
thread_count_ = MSMIN(op_parameter_->thread_num_, UP_DIV(matmul_param_->col_, C8NUM));
thread_stride_ = UP_DIV(UP_DIV(matmul_param_->col_, C8NUM), thread_count_);
} else {
thread_count_ = MSMIN(op_parameter_->thread_num_, UP_DIV(matmul_param_->col_, C4NUM));
thread_stride_ = UP_DIV(UP_DIV(matmul_param_->col_, C4NUM), thread_count_);
}
thread_count_ = MSMIN(op_parameter_->thread_num_, UP_DIV(matmul_param_->col_, row_pack_count));
thread_stride_ = UP_DIV(UP_DIV(matmul_param_->col_, row_pack_count), thread_count_);
if (support_optimize_) {
thread_count_hw_ = MSMIN(op_parameter_->thread_num_, UP_DIV(matmul_param_->row_, C8NUM));
thread_stride_hw_ = UP_DIV(UP_DIV(matmul_param_->row_, C8NUM), thread_count_hw_);
} else {
thread_count_hw_ = MSMIN(op_parameter_->thread_num_, UP_DIV(matmul_param_->row_, C4NUM));
thread_stride_hw_ = UP_DIV(UP_DIV(matmul_param_->row_, C4NUM), thread_count_hw_);
}
thread_count_hw_ = MSMIN(op_parameter_->thread_num_, UP_DIV(matmul_param_->row_, col_pack_count));
thread_stride_hw_ = UP_DIV(UP_DIV(matmul_param_->row_, col_pack_count), thread_count_hw_);
if (pre_trans_input_) {
input_ptr_ = reinterpret_cast<int8_t *>(malloc(matmul_param_->row_ * matmul_param_->deep_ * sizeof(int8_t)));
@ -296,7 +288,7 @@ int Convolution1x1Int8Impl(void *cdata, int task_id) {
}
int Convolution1x1Int8CPUKernel::InitRunBuf() {
input_sum_ = reinterpret_cast<int32_t *>(malloc(input_sum_size * sizeof(int32_t)));
input_sum_ = reinterpret_cast<int32_t *>(ctx_->allocator->Malloc(input_sum_size * sizeof(int32_t)));
if (input_sum_ == nullptr) {
MS_LOG(ERROR) << "malloc input_sum_ failed.";
return RET_ERROR;
@ -334,6 +326,7 @@ int Convolution1x1Int8CPUKernel::Run() {
int error_code = InitRunBuf();
if (error_code != RET_OK) {
MS_LOG(ERROR) << "conv1x1 int8 InitRunBuf error_code[" << error_code << "]";
FreeRunBuf();
return RET_ERROR;
}