tbmvgpu: tbmvgpu

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

Description

This function computes triangular banded matrix-vector multipication x = op(A) x by using CUDA cublas function cublasDtbmv

Usage

tbmvgpu(fillmode = 1, trans = 1, diagmode = 1, k, A, x)

Arguments

fillmode	indicates if matrix A lower or upper part is stored, the other symmetric part is not referenced and is inferred from the stored elements. if fillmode == 1 then the triagular banded matrix A is stored column by column, with the main diagonal of the matrix stored in row 1, the first subdiagonal in row 2 (starting at first position), the second subdiagonal in row 3 (starting at first position), etc. if fillmode == 2 then the triangular banded matrix A is stored column by column, with the main diagonal of the matrix stored in row k+1, the first superdiagonal in row k (starting at second position), the second superdiagonal in row k-1 (starting at third position), etc.
trans	matrix A transpose operator, 1 (non-transpose), 2 (transpose), 3 (conjugate transpose); default at 1 (non-transpose)
diagmode	indicates whether the main diagonal of the matrix A is unity and consequently should not be touched or modified by the function. if diagmode = 1, the matrix diagonal has non-unit elements, if diagmode = 2, the matrix diagonal has unit elements
k	number of sub- or super- diagonals
A	input matrix; list of R external GPU pointer and dimension
x	input/output vector; list of R external GPU pointer and dimension

Value

updated vector x, a list consisting of

• ptr: GPU pointer

• m: length of vector x

• n: 1

See Also

gemvgpu

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