gemmgpu: gemmgpu

In yuanli22/RCUDA: GPU Enabled BLAS, LAPACK, Statistical Functions and Random Number Generators

Description

This function computes the matrix-matrix multiplication C = a op ( A ) op ( B ) + b C by using CUDA cublas function cublasDgemm

Usage

gemmgpu(transa = 1, transb = 1, alpha = 1, A, B, beta = 0, C)

Arguments

transa	matrix A transpose operator, 1 (non-transpose), 2 (transpose), 3 (conjugate transpose); default at 1 (non-transpose)
transb	matrix B transpose operator, 1 (non-transpose), 2 (transpose), 3 (conjugate transpose); default at 1 (non-transpose)
alpha	scale factor a of matrix A; default 1
A	input matrix; list of R external GPU pointer and dimension
B	input matrix; list of R external GPU pointer and dimension
beta	scale factor b of matrix C; default 0
C	input/output matrix; list of R external GPU pointer and dimension

Value

updated matrix C, a list consisting of

• ptr: GPU pointer

• m: matrix C's number of rows

• n: matrix C's number of columns

See Also

gemvgpu

Examples

A_gpu <- creategpu(1:6, 3, 2)
B_gpu <- creategpu(1:6, 3, 2)
C_gpu <- creategpu(1:4, 2, 2)
gemmgpu(2, 1, 1, A_gpu, B_gpu, beta=1, C_gpu)
gathergpu(C_gpu)

yuanli22/RCUDA documentation built on May 4, 2019, 6:35 p.m.