Add abstract "MatGeneric<FPP, GPU2>::multiply(MatDense<FPP, GPU2>&, const...

Add abstract "MatGeneric<FPP, GPU2>::multiply(MatDense<FPP, GPU2>&, const char) const" and a fix of its overload in MatDense<FPP, GPU2>.

src/faust_linear_operator/GPU2/faust_MatButterfly_gpu.h

@@ -94,7 +94,7 @@ namespace Faust
 
 			void Display() const;
 			Real<FPP> norm() const;
-			void multiply(MatDense<FPP, GPU2> &other, const char op_this);
+			void multiply(MatDense<FPP, GPU2> &other, const char op_this) const;
 			void multiply(MatSparse<FPP, GPU2> &other, const char op_this);
 			MatSparse<FPP, GPU2> toMatSparse() const;
 			~MatButterfly();

src/faust_linear_operator/GPU2/faust_MatButterfly_gpu.hpp

@@ -24,7 +24,7 @@ namespace Faust
 
 
 	template<typename FPP>
-		void MatButterfly<FPP, GPU2>::multiply(MatDense<FPP, GPU2> &other, const char op_this)
+		void MatButterfly<FPP, GPU2>::multiply(MatDense<FPP, GPU2> &other, const char op_this) const
 		{
 			if(op_this != 'N' && op_this != 'T')
 				throw std::runtime_error("MatButtermfly::multiply only handle 'N' and 'T' for op_this");

src/faust_linear_operator/GPU2/faust_MatDense_gpu.cpp.in

@@ -67,17 +67,6 @@ namespace Faust
 				const void* stream/*=nullptr*/) : MatDense<@FAUST_SCALAR_FOR_GM@,GPU2>(mat.getNbRow(), mat.getNbCol(), mat.getData(), /*no_alloc*/ mat.getData() == nullptr, dev_id, stream){}
 
 
-
-	template<>
-		void Faust::MatDense<@FAUST_SCALAR_FOR_GM@,GPU2>::multiply(const MatDense<@FAUST_SCALAR_FOR_GM@, GPU2> &other, const char op_this)
-		{
-			// other = this * other
-			gm_Op gop_this;
-			char2gm_Op(op_this, gop_this);
-			auto dsm_funcs = GPUModHandler::get_singleton()->dsm_funcs(@FAUST_SCALAR_FOR_GM@(0));
-			dsm_funcs->mul_gpu_dsm_ext(this->gpu_mat, other.gpu_mat, other.gpu_mat, gop_this, OP_NOTRANSP);
-		}
-
 	template<>
 		void Faust::MatDense<@FAUST_SCALAR_FOR_GM@,GPU2>::multiply(MatDense<@FAUST_SCALAR_FOR_GM@, Cpu> &other, const char op_this)
 		{
@@ -632,6 +621,28 @@ namespace Faust
 			return *this;
 		}
 
+	template<>
+		void Faust::MatDense<@FAUST_SCALAR_FOR_GM@,GPU2>::multiply(MatDense<@FAUST_SCALAR_FOR_GM@, GPU2> &other, const char op_this) const
+		{
+			// other = this * other
+			gm_Op gop_this;
+			char2gm_Op(op_this, gop_this);
+			auto dsm_funcs = GPUModHandler::get_singleton()->dsm_funcs(@FAUST_SCALAR_FOR_GM@(0));
+			faust_unsigned_int m, n; // prod dims
+			m = op_this == 'N'?this->getNbRow():this->getNbCol();
+			n = other.getNbCol();
+			// is other large enough to receive this * other ?
+			if(other.getNbRow() < m)
+			{
+				MatDense<@FAUST_SCALAR_FOR_GM@, GPU2> out(m, n);
+				dsm_funcs->mul_gpu_dsm_ext(this->gpu_mat, other.gpu_mat, out.gpu_mat, gop_this, OP_NOTRANSP);
+				other = std::move(out);
+			}
+			else
+				dsm_funcs->mul_gpu_dsm_ext(this->gpu_mat, other.gpu_mat, other.gpu_mat, gop_this, OP_NOTRANSP);
+		}
+
+
 	template<>
 		MatDense<@FAUST_SCALAR_FOR_GM@,GPU2>::MatDense(MatDense<@FAUST_SCALAR_FOR_GM@,GPU2>&& mat)
 		{

src/faust_linear_operator/GPU2/faust_MatDense_gpu.h

@@ -103,7 +103,7 @@ namespace Faust
 				void eltwise_mul(const Vect<FPP, GPU2> &vec, const int *ids=nullptr);
 				//  TODO: other shouldn't be const if it is the output
 				//  other = (*this) * other
-				void multiply(const MatDense<FPP, GPU2> &other, const char op_this='N');
+				void multiply(MatDense<FPP, GPU2> &other, const char op_this='N') const;
 				//  other = (*this) * other
 				void multiply(MatDense<FPP, Cpu> &other, const char op_this='N');
 //				void multiply(MatSparse<FPP, Cpu> &other, MatDense<FPP, GPU2>& output, const char op_this='N');

src/faust_linear_operator/GPU2/faust_MatGeneric_gpu.h

@@ -46,6 +46,8 @@ namespace Faust
 
 				virtual void Display() const=0;
 				virtual Real<FPP> norm() const=0;
+
+				virtual void multiply(MatDense<FPP,GPU2> &A, const char opThis) const =0;
 				MatGeneric();
 
 				virtual ~MatGeneric();

src/faust_linear_operator/GPU2/faust_MatPerm_gpu.h

@@ -90,7 +90,7 @@ namespace Faust
 
 			void Display() const;
 			Real<FPP> norm() const;
-			void multiply(MatDense<FPP, GPU2> &other, const char op_this);
+			void multiply(MatDense<FPP, GPU2> &other, const char op_this) const;
 			void multiply(MatSparse<FPP, GPU2> &other, const char op_this);
 			MatSparse<FPP, GPU2> toMatSparse() const;
 

src/faust_linear_operator/GPU2/faust_MatPerm_gpu.hpp

@@ -21,7 +21,7 @@ namespace Faust
 
 
 	template<typename FPP>
-		void MatPerm<FPP, GPU2>::multiply(MatDense<FPP, GPU2> &other, const char op_this)
+		void MatPerm<FPP, GPU2>::multiply(MatDense<FPP, GPU2> &other, const char op_this) const
 		{
 			if(op_this != 'N' && op_this != 'T')
 				throw std::runtime_error("MatButtermfly::multiply only handle 'N' and 'T' for op_this");

src/faust_linear_operator/GPU2/faust_MatSparse_gpu.cpp.in

@@ -474,7 +474,7 @@ namespace Faust
 		}
 
 	template<>
-		void MatSparse<FSFG, GPU2>::multiply(MatDense<FSFG,GPU2>& mat, char opThis/*='N'*/) const
+		void MatSparse<FSFG, GPU2>::multiply(MatDense<FSFG,GPU2>& mat, const char opThis/*='N'*/) const
 		{
 			gm_Op gop_this;
 			char2gm_Op(opThis, gop_this);

src/faust_linear_operator/GPU2/faust_MatSparse_gpu.h

@@ -100,7 +100,7 @@ namespace Faust
 				std::string to_string(const bool transpose=false, const bool displaying_small_mat_elts=false) const;
 				MatType getType() const;
 				void multiply(Vect<FPP,GPU2>& vec, char opThis='N') const;
-				void multiply(MatDense<FPP,GPU2>& vec, char opThis='N') const;
+				void multiply(MatDense<FPP,GPU2>& mat, const char opThis='N') const;
 				static void spgemm(const MatSparse<FPP,GPU2> & A, const MatDense<FPP,GPU2> & B, MatDense<FPP,GPU2> & C, const FPP & alpha, const FPP & beta, const char opA, const char opB);
 				MatBSR<FPP, GPU2> to_bsr(int bsize) const;
 				~MatSparse();