Commit 2e451b9d authored by Peter Eastman's avatar Peter Eastman
Browse files

Deleted the old CUDA platform

parent 352e2fc7
/* -------------------------------------------------------------------------- *
* OpenMM *
* -------------------------------------------------------------------------- *
* This is part of the OpenMM molecular simulation toolkit originating from *
* Simbios, the NIH National Center for Physics-Based Simulation of *
* Biological Structures at Stanford, funded under the NIH Roadmap for *
* Medical Research, grant U54 GM072970. See https://simtk.org. *
* *
* Portions copyright (c) 2010 Stanford University and the Authors. *
* Authors: Peter Eastman *
* Contributors: *
* *
* This program is free software: you can redistribute it and/or modify *
* it under the terms of the GNU Lesser General Public License as published *
* by the Free Software Foundation, either version 3 of the License, or *
* (at your option) any later version. *
* *
* 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 Lesser General Public License for more details. *
* *
* You should have received a copy of the GNU Lesser General Public License *
* along with this program. If not, see <http://www.gnu.org/licenses/>. *
* -------------------------------------------------------------------------- */
#include <stdio.h>
#include <cuda.h>
#include <vector_functions.h>
#include <cstdlib>
#include <string>
#include <iostream>
//#include <fstream>
using namespace std;
#include "gputypes.h"
#include "cudaKernels.h"
__global__ void kPrepareConstraints_kernel(int numAtoms, float4* oldPos, float4* posq, float4* posqP) {
for (int index = threadIdx.x+blockIdx.x*blockDim.x; index < numAtoms; index += blockDim.x*gridDim.x) {
float4 pos = posq[index];
oldPos[index] = pos;
posqP[index] = make_float4(0.0f, 0.0f, 0.0f, pos.w);
}
}
__global__ void kFinishConstraints_kernel(int numAtoms, float4* posq, float4* posqP) {
for (int index = threadIdx.x+blockIdx.x*blockDim.x; index < numAtoms; index += blockDim.x*gridDim.x) {
float4 pos = posq[index];
float4 delta = posqP[index];
posq[index] = make_float4(pos.x+delta.x, pos.y+delta.y, pos.z+delta.z, pos.w);
}
}
void kApplyConstraints(gpuContext gpu)
{
kPrepareConstraints_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(gpu->natoms, gpu->sim.pOldPosq, gpu->sim.pPosq, gpu->sim.pPosqP);
LAUNCHERROR("kPrepareConstraints");
kApplyShake(gpu);
kApplySettle(gpu);
kApplyCCMA(gpu);
kFinishConstraints_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(gpu->natoms, gpu->sim.pPosq, gpu->sim.pPosqP);
LAUNCHERROR("kFinishConstraints");
}
/* -------------------------------------------------------------------------- *
* OpenMM *
* -------------------------------------------------------------------------- *
* This is part of the OpenMM molecular simulation toolkit originating from *
* Simbios, the NIH National Center for Physics-Based Simulation of *
* Biological Structures at Stanford, funded under the NIH Roadmap for *
* Medical Research, grant U54 GM072970. See https://simtk.org. *
* *
* Portions copyright (c) 2009 Stanford University and the Authors. *
* Authors: Scott Le Grand, Peter Eastman *
* Contributors: *
* *
* This program is free software: you can redistribute it and/or modify *
* it under the terms of the GNU Lesser General Public License as published *
* by the Free Software Foundation, either version 3 of the License, or *
* (at your option) any later version. *
* *
* 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 Lesser General Public License for more details. *
* *
* You should have received a copy of the GNU Lesser General Public License *
* along with this program. If not, see <http://www.gnu.org/licenses/>. *
* -------------------------------------------------------------------------- */
#include <stdio.h>
#include <cuda.h>
#include <vector_functions.h>
#include <cstdlib>
#include <string>
#include <iostream>
//#include <fstream>
using namespace std;
#include "gputypes.h"
static __constant__ cudaGmxSimulation cSim;
void SetBrownianUpdateSim(gpuContext gpu)
{
cudaError_t status;
status = cudaMemcpyToSymbol(cSim, &gpu->sim, sizeof(cudaGmxSimulation));
RTERROR(status, "cudaMemcpyToSymbol: SetSim copy to cSim failed");
}
void GetBrownianUpdateSim(gpuContext gpu)
{
cudaError_t status;
status = cudaMemcpyFromSymbol(&gpu->sim, cSim, sizeof(cudaGmxSimulation));
RTERROR(status, "cudaMemcpyFromSymbol: SetSim copy from cSim failed");
}
__global__
#if (__CUDA_ARCH__ >= 200)
__launch_bounds__(GF1XX_UPDATE_THREADS_PER_BLOCK, 1)
#elif (__CUDA_ARCH__ >= 120)
__launch_bounds__(GT2XX_UPDATE_THREADS_PER_BLOCK, 1)
#else
__launch_bounds__(G8X_UPDATE_THREADS_PER_BLOCK, 1)
#endif
void kBrownianUpdatePart1_kernel()
{
unsigned int pos = threadIdx.x + blockIdx.x * blockDim.x;
unsigned int rpos = cSim.pRandomPosition[blockIdx.x];
__syncthreads();
while (pos < cSim.atoms)
{
float4 random4a = cSim.pRandom4[rpos + pos];
float4 apos = cSim.pPosq[pos];
float4 force = cSim.pForce4[pos];
float invMass = cSim.pVelm4[pos].w;
float forceScale = cSim.tauDeltaT*invMass;
float noiseScale = cSim.noiseAmplitude*sqrtf(invMass);
cSim.pOldPosq[pos] = apos;
apos.x = force.x*forceScale + noiseScale*random4a.x;
apos.y = force.y*forceScale + noiseScale*random4a.y;
apos.z = force.z*forceScale + noiseScale*random4a.z;
cSim.pPosqP[pos] = apos;
pos += blockDim.x * gridDim.x;
}
}
void kBrownianUpdatePart1(gpuContext gpu)
{
// printf("kBrownianUpdatePart1\n");
kBrownianUpdatePart1_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>();
LAUNCHERROR("kBrownianUpdatePart1");
}
__global__
#if (__CUDA_ARCH__ >= 200)
__launch_bounds__(GF1XX_UPDATE_THREADS_PER_BLOCK, 1)
#elif (__CUDA_ARCH__ >= 120)
__launch_bounds__(GT2XX_UPDATE_THREADS_PER_BLOCK, 1)
#else
__launch_bounds__(G8X_UPDATE_THREADS_PER_BLOCK, 1)
#endif
void kBrownianUpdatePart2_kernel()
{
unsigned int pos = threadIdx.x + blockIdx.x * blockDim.x;
unsigned int rpos = cSim.pRandomPosition[blockIdx.x];
__syncthreads();
while (pos < cSim.atoms)
{
float4 velocity = cSim.pVelm4[pos];
float4 apos = cSim.pPosq[pos];
float4 xPrime = cSim.pPosqP[pos];
velocity.x = cSim.oneOverDeltaT*(xPrime.x);
velocity.y = cSim.oneOverDeltaT*(xPrime.y);
velocity.z = cSim.oneOverDeltaT*(xPrime.z);
xPrime.x += apos.x;
xPrime.y += apos.y;
xPrime.z += apos.z;
cSim.pPosq[pos] = xPrime;
cSim.pVelm4[pos] = velocity;
pos += blockDim.x * gridDim.x;
}
// Update random position pointer
if (threadIdx.x == 0)
{
rpos += cSim.paddedNumberOfAtoms;
if (rpos > cSim.randoms)
rpos -= cSim.randoms;
cSim.pRandomPosition[blockIdx.x] = rpos;
}
}
extern void kGenerateRandoms(gpuContext gpu);
void kBrownianUpdatePart2(gpuContext gpu)
{
// printf("kBrownianUpdatePart2\n");
kBrownianUpdatePart2_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>();
LAUNCHERROR("kBrownianUpdatePart2");
// Update randoms if necessary
gpu->iterations++;
if (gpu->iterations == gpu->sim.randomIterations)
{
kGenerateRandoms(gpu);
gpu->iterations = 0;
}
}
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment