New implementation for ABGDXYZ in CUDA

nond-rep
Salvatore Filippone 11 months ago
parent 105aa3c570
commit 4681767ef8

@ -241,6 +241,30 @@ int axpbyMultiVecDeviceDouble(int n,double alpha, void* devMultiVecX,
return(i);
}
int abgdxyzMultiVecDeviceDouble(int n,double alpha,double beta, double gamma, double delta,
void* devMultiVecX, void* devMultiVecY, void* devMultiVecZ)
{ int j=0, i=0;
int pitch = 0;
struct MultiVectDevice *devVecX = (struct MultiVectDevice *) devMultiVecX;
struct MultiVectDevice *devVecY = (struct MultiVectDevice *) devMultiVecY;
struct MultiVectDevice *devVecZ = (struct MultiVectDevice *) devMultiVecZ;
spgpuHandle_t handle=psb_cudaGetHandle();
pitch = devVecY->pitch_;
if ((n > devVecY->size_) || (n>devVecX->size_ ))
return SPGPU_UNSUPPORTED;
#if 1
spgpuDabgdxyz(handle,n, alpha,beta,gamma,delta,
(double*)devVecX->v_,(double*) devVecY->v_,(double*) devVecZ->v_);
#else
for(j=0;j<devVecY->count_;j++)
spgpuDaxpby(handle,(double*)devVecY->v_+pitch*j, n, beta,
(double*)devVecY->v_+pitch*j, alpha,(double*) devVecX->v_+pitch*j);
#endif
return(i);
}
int axyMultiVecDeviceDouble(int n, double alpha, void *deviceVecA, void *deviceVecB)
{ int i = 0;
struct MultiVectDevice *devVecA = (struct MultiVectDevice *) deviceVecA;

@ -947,7 +947,8 @@ contains
if ((nx<m).or.(ny<m).or.(nz<m)) then
info = psb_err_internal_error_
else
!info = axpbyMultiVecDevice(m,alpha,xx%deviceVect,beta,y%deviceVect)
info = abgdxyzMultiVecDevice(m,alpha,beta,gamma,delta,&
& xx%deviceVect,yy%deviceVect,zz%deviceVect)
end if
call yy%set_dev()
call zz%set_dev()

@ -316,6 +316,19 @@ module psb_d_vectordev_mod
end function axpbyMultiVecDeviceDouble
end interface
interface abgdxyzMultiVecDevice
function abgdxyzMultiVecDeviceDouble(n,alpha,beta,gamma,delta,deviceVecX,&
& deviceVecY,deviceVecZ) &
& result(res) bind(c,name='abgdxyzMultiVecDeviceDouble')
use iso_c_binding
integer(c_int) :: res
integer(c_int), value :: n
real(c_double), value :: alpha, beta,gamma,delta
type(c_ptr), value :: deviceVecX, deviceVecY, deviceVecZ
end function abgdxyzMultiVecDeviceDouble
end interface abgdxyzMultiVecDevice
interface axyMultiVecDevice
function axyMultiVecDeviceDouble(n,alpha,deviceVecA,deviceVecB) &
& result(res) bind(c,name='axyMultiVecDeviceDouble')

@ -12,7 +12,7 @@ CINCLUDES=-I$(INCDIR)
OBJS=cabs.o camax.o casum.o caxpby.o caxy.o cdot.o cgath.o \
cnrm2.o cscal.o cscat.o csetscal.o \
dabs.o damax.o dasum.o daxpby.o daxy.o ddot.o dgath.o \
dabs.o damax.o dasum.o daxpby.o daxy.o ddot.o dgath.o dabgdxyz.o\
dia_cspmv.o dia_dspmv.o dia_sspmv.o dia_zspmv.o dnrm2.o \
dscal.o dscat.o dsetscal.o ell_ccsput.o ell_cspmv.o \
ell_dcsput.o ell_dspmv.o ell_scsput.o ell_sspmv.o ell_zcsput.o ell_zspmv.o \

@ -0,0 +1,79 @@
/*
* spGPU - Sparse matrices on GPU library.
*
* Copyright (C) 2010 - 2012
* Davide Barbieri - University of Rome Tor Vergata
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* version 3 as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*/
#include "cudadebug.h"
#include "cudalang.h"
#include <cuda_runtime.h>
extern "C"
{
#include "core.h"
#include "vector.h"
}
#include "debug.h"
#define BLOCK_SIZE 512
__global__ void spgpuDabgdxyz_krn(int n, double alpha, double beta, double gamma, double delta,
double* x, double *y, double *z)
{
int id = threadIdx.x + BLOCK_SIZE*blockIdx.x;
unsigned int gridSize = blockDim.x * gridDim.x;
double t;
for ( ; id < n; id +=gridSize)
//if (id,n)
{
if (beta == 0.0)
t = PREC_DMUL(alpha,x[id]);
else
t = PREC_DADD(PREC_DMUL(alpha, x[id]), PREC_DMUL(beta,y[id]));
if (delta == 0.0)
z[id] = gamma * t;
else
z[id] = PREC_DADD(PREC_DMUL(gamma, t), PREC_DMUL(delta,z[id]));
y[id] = t;
}
}
void spgpuDabgdxyz(spgpuHandle_t handle,
int n,
double alpha,
double beta,
double gamma,
double delta,
__device double* x,
__device double* y,
__device double *z)
{
int msize = (n+BLOCK_SIZE-1)/BLOCK_SIZE;
int num_mp, max_threads_mp, num_blocks_mp, num_blocks;
dim3 block(BLOCK_SIZE);
cudaDeviceProp deviceProp;
cudaGetDeviceProperties(&deviceProp, 0);
num_mp = deviceProp.multiProcessorCount;
max_threads_mp = deviceProp.maxThreadsPerMultiProcessor;
num_blocks_mp = max_threads_mp/BLOCK_SIZE;
num_blocks = num_blocks_mp*num_mp;
dim3 grid(num_blocks);
spgpuDabgdxyz_krn<<<grid, block, 0, handle->currentStream>>>(n, alpha, beta, gamma, delta,
x, y, z);
}
Loading…
Cancel
Save