Skip to content

[BugFix][OpenCL] Fix concat image impl when axis is not 1. test=develop #4241

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 8 commits into from
Sep 7, 2020
Merged

[BugFix][OpenCL] Fix concat image impl when axis is not 1. test=develop #4241

Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
37b01fa
[BugFix][OpenCL] Fix concat image impl when concat axis is not 1
zhaoyang-star Sep 2, 2020
d794dec
fix code when axis == 1. test=develop
zhaoyang-star Sep 2, 2020
d3cf5e1
fix code when axis == -1. test=develop
zhaoyang-star Sep 2, 2020
7913180
fix code, test=develop
zhaoyang-star Sep 2, 2020
705d4f6
format code, test=develop
zhaoyang-star Sep 2, 2020
a64133c
fix illegal access when print debug info. test=develop
zhaoyang-star Sep 2, 2020
656cb45
fix typo
zhaoyang-star Sep 3, 2020
53a965a
test=develop
zhaoyang-star Sep 3, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
46 changes: 14 additions & 32 deletions lite/kernels/opencl/concat_image_compute.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,15 +38,13 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
void PrepareForRun() override {
auto& context = ctx_->As<OpenCLContext>();
concat_param_ = param_.get_mutable<param_t>();
axis_ = concat_param_->axis;
if (-1 == axis_) {
axis_ = concat_param_->x[0]->dims().size() - 1;
}

auto inputs = concat_param_->x;
auto axis_ = concat_param_->axis;
auto output_tensor_dims = concat_param_->output->dims();
auto* axis_tensor = concat_param_->axis_tensor;
if (axis_tensor != nullptr) {
// auto* axis_tensor_data = axis_tensor->data<int>(TARGET(kARM));
// axis = axis_tensor_data[0];
}

if (inputs.size() == 2) {
kernel_func_name_ = "concat2";
Expand Down Expand Up @@ -100,8 +98,7 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
width_ = output_tensor_dims[0]; // n
flag_ = 2;
break;
case 3:
case -1: // width
case 3: // width
width_ = output_tensor_dims[1]; // c
flag_ = 3;
break;
Expand All @@ -113,17 +110,12 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
auto input0_tensor_dims = inputs[0]->dims();
for (int i = 1; i < inputs.size(); i++) {
auto dims = inputs[i]->dims();
// auto flag = CHECK_EQ_OR_FALSE(input0_tensor_dims.size(), dims.size());
if (input0_tensor_dims.size() != dims.size()) {
printf("input shape must be same \n");
return;
}
CHECK(input0_tensor_dims.size() == dims.size())
<< "All inputs must have the same axes!";
for (int i = 0; i < dims.size(); i++) {
if (i != axis_) {
if (input0_tensor_dims[i] != dims[i]) {
printf("input shape must be same \n");
return;
}
CHECK(input0_tensor_dims[i] == dims[i])
<< "All inputs must have the same shape, except at concat axis!";
}
}
}
Expand Down Expand Up @@ -151,29 +143,18 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
VLOG(4) << "concat input shape: ";
for (size_t i = 0; i < inputs.size(); i++) {
VLOG(4) << "inputs [" << i << "]"
<< "[" << inputs[i]->dims().size() << "D]:"
<< " dims:" << inputs[i]->dims()[0] << " "
<< inputs[i]->dims()[1] << " " << inputs[i]->dims()[2] << " "
<< inputs[i]->dims()[3];
<< " dims:" << inputs[i]->dims();
}

VLOG(4) << "concat output shape: ";
VLOG(4) << " out dims: "
<< "[" << output_tensor_dims.size()
<< "D]:" << output_tensor_dims[0] << " " << output_tensor_dims[1]
<< " " << output_tensor_dims[2] << " " << output_tensor_dims[3];
VLOG(4) << " out dims: " << output_tensor_dims;
VLOG(4) << "axis_: " << axis_;
VLOG(4) << "flag_: " << flag_;

VLOG(4) << TargetToStr(concat_param_->output->target());
VLOG(4) << "output_image_shape(w,h):" << output_image_shape["width"] << " "
VLOG(4) << "output_image_shape(w,h): " << output_image_shape["width"] << " "
<< output_image_shape["height"];
VLOG(4) << "output_tensor_dims[" << output_tensor_dims.size()
<< "D]:" << output_tensor_dims[0] << " " << output_tensor_dims[1]
<< " " << output_tensor_dims[2] << " " << output_tensor_dims[3]
<< "output_tensor_dims[output_tensor_dims.size() - 1]"
<< output_tensor_dims[output_tensor_dims.size() - 1];
VLOG(4) << "output_tensor_w: " << output_tensor_w << ", flag_: " << flag_;
VLOG(4) << "output_tensor_w: " << output_tensor_w;
VLOG(4) << "width_:" << width_;
VLOG(4) << "global_work_size: "
<< output_tensor_dims[output_tensor_dims.size() - 1] << " "
Expand Down Expand Up @@ -433,6 +414,7 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
}
#endif

private:
int axis_ = 1;
int flag_ = 1;
int width_ = 1;
Expand Down