[blas/OpenCL] Added multiply OpenCL kernel and unit test

Added sscal equivalent kernel and multiply function. Added unit test setup to test standalone kernels. Signed-off-by: Debadri Samaddar <[email protected]>
nnstreamer · Jul 2, 2024 · 602bc5e · 602bc5e
1 parent b9c27e9
commit 602bc5e
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 1 deletion.
diff --git a/nntrainer/tensor/cl_operations/blas_kernel_interface.h b/nntrainer/tensor/cl_operations/blas_kernel_interface.h
@@ -30,6 +30,18 @@ namespace nntrainer {
 Tensor dotCl(Tensor const &input, Tensor const &m, RunLayerContext &context,
              bool trans = false, bool trans_m = false);
 
+/**
+ * @brief Process data and dimensions for OpenCL dot operation
+ * @param[in] input Tensor
+ * @param[in] m Tensor
+ * @param[in] result Tensor
+ * @param[in] RunLayerContext reference
+ * @param[in] trans bool
+ * @param[in] trans_m bool
+ */
+Tensor dotCl(Tensor const &input, Tensor const &m, RunLayerContext &context,
+             bool trans = false, bool trans_m = false);
+
 /**
  * @brief Process data and dimensions for OpenCL dot operation
  * @param[in] input Tensor

diff --git a/nntrainer/tensor/cl_operations/blas_kernels.cpp b/nntrainer/tensor/cl_operations/blas_kernels.cpp
@@ -421,4 +421,5 @@ void sscal_cl(float *X, const unsigned int N, const float alpha,
 
   } while (false);
 }
+
 } // namespace nntrainer
diff --git a/nntrainer/tensor/cl_operations/blas_kernels.h b/nntrainer/tensor/cl_operations/blas_kernels.h
@@ -158,7 +158,7 @@ void addition_cl(const __fp16 *input, __fp16 *res, unsigned int size,
 
 /**
  * @brief     fp16 sscal value element by element immediately
- * @param[in] X __fp16 * input
+ * @param[in] X float * input
  * @param[in] N unsigned int number of elements
  * @param[in] alpha float multiplier
  * @param[in] context RunLayerContext reference

diff --git a/nntrainer/tensor/cl_operations/blas_kernels_fp16.cpp b/nntrainer/tensor/cl_operations/blas_kernels_fp16.cpp
@@ -441,4 +441,5 @@ void sscal_cl(__fp16 *X, const unsigned int N, const float alpha,
 
   } while (false);
 }
+
 } // namespace nntrainer
Original file line number	Diff line number	Diff line change
Expand Up		@@ -421,4 +421,5 @@ void sscal_cl(float *X, const unsigned int N, const float alpha,

		} while (false);
		}

		} // namespace nntrainer
Original file line number	Diff line number	Diff line change
Expand Up		@@ -441,4 +441,5 @@ void sscal_cl(__fp16 *X, const unsigned int N, const float alpha,

		} while (false);
		}

		} // namespace nntrainer