#include <cstring>
#include "./opr_impl.h"
#include "src/common/utils.h"
#include "src/naive/handle.h"
namespace megdnn {
namespace naive {
template <typename T>
void exec_internal(_megdnn_tensor_in src, _megdnn_tensor_out dst, bool clockwise) {
auto N = src.layout.shape[0], IH = src.layout.shape[1], IW = src.layout.shape[2],
IC = src.layout.shape[3];
rep(n, N) rep(ih, IH) rep(iw, IW) {
int ow = clockwise ? IH - ih - 1 : ih;
int oh = clockwise ? iw : IW - iw - 1;
rep(c, IC) {
dst.ptr<T>()
[n * dst.layout.stride[0] + oh * dst.layout.stride[1] +
ow * dst.layout.stride[2] + c * dst.layout.stride[3]] =
src.ptr<T>()
[n * src.layout.stride[0] + ih * src.layout.stride[1] +
iw * src.layout.stride[2] + c * dst.layout.stride[3]];
}
}
}
void RotateImpl::exec(
_megdnn_tensor_in src, _megdnn_tensor_in dst, _megdnn_workspace workspace) {
check_exec(src.layout, dst.layout, workspace.size);
#define cb(DType) \
if (src.layout.dtype.enumv() == DTypeTrait<DType>::enumv) { \
using ctype = typename DTypeTrait<DType>::ctype; \
MEGDNN_DISPATCH_CPU_KERN_OPR( \
exec_internal<ctype>(src, dst, param().clockwise)); \
return; \
}
MEGDNN_FOREACH_COMPUTING_DTYPE(cb)
MEGDNN_FOREACH_QUANTIZED_DTYPE(cb)
#undef cb
megdnn_assert_internal(0);
}
} }