Implement c++ Vect<FPP, GPU2>::multiplyLeft(MatSparse<FPP, GPU2>).

src/faust_linear_operator/GPU2/faust_Vect_gpu.cpp.in

@@ -13,10 +13,10 @@ namespace Faust
 
 	template<>
 		Vect<FSFG,GPU2>::Vect(const faust_unsigned_int size,
-				const FSFG* cpu_data,
-				const bool no_alloc,
-				const int32_t dev_id,
-				const void* stream): MatDense<FSFG,GPU2>(size, 1, cpu_data, no_alloc, dev_id, stream)
+				const FSFG* cpu_data/*=nullptr*/,
+				const bool no_alloc/*=false*/,
+				const int32_t dev_id/*=-1*/,
+				const void* stream/*=nullptr*/): MatDense<FSFG,GPU2>(size, 1, cpu_data, no_alloc, dev_id, stream)
 	{
 	}
 
@@ -104,7 +104,7 @@ namespace Faust
 		template<>
 		void Vect<FSFG,GPU2>::multiplyLeft(MatSparse<FSFG,GPU2> const& S, const char transS)
 		{
-
+			S.multiply(*this);
 		}
 
 		template<>

src/faust_linear_operator/GPU2/faust_Vect_gpu.h

@@ -46,7 +46,7 @@ namespace Faust
 			FPP mean_relative_error(const Vect<FPP,GPU2>& ref_vec) const;
 			// delete parent methods that don't apply to a vector
 			void setEyes() = delete;
-			void multiplyLeft(MatSparse<FPP,GPU2> const& S, const char transS);
+			void multiplyLeft(MatSparse<FPP,GPU2> const& S, const char transS='N');
 			FPP operator[](faust_unsigned_int i);
 //			FPP& operator[](faust_unsigned_int i); // impossible to return a reference to a GPU buffer value (therefore it's impossible to modify a vector coeff as it's done with Vect<FPP,Cpu> (e.g. v[i] = 3)
 			void set_coeff(faust_unsigned_int i, const FPP& val);