Skip to content

GitLab

"wrappers/vscode:/vscode.git/clone" did not exist on "572abf661899686fc0bd98041ef66c898601a5b5"
Unverified Commit 8e8923a7 authored by Peter Eastman's avatar Peter Eastman Committed by GitHub
Browse files

Converted AMOEBA to common platform (#3120) 

* Began converting AMOEBA to common platform

* Beginning of OpenCL platform for AMOEBA

* Converted AmoebaVdwForce to common platform

* Cleaned up reference AMOEBA tests

* Began converting AmoebaMultipoleForce to common platform

* Continue converting AmoebaMultipoleForce to common platform

* Bug fixes

* Bug fix

* Continue converting AmoebaMultipoleForce to common platform

* Converting AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce to common platform

* Converting AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce to common platform

* Creating OpenCL version of AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce

* Creating OpenCL version of AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce

* Creating OpenCL version of AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce

* Converted arrays from real3 to real

* Bug fix to OpenCL AmoebaGeneralizedKirkwoodForce

* Fixes for AMD GPUs

* Began converting HippoNonbondedForce to common platform

* Continuing to convert HippoNonbondedForce to common platform

* Continuing to convert HippoNonbondedForce to common platform

* Working on unifying PME kernels

* Fixed error on devices without 64 bit atomics

* Unified PME kernels

* Converted HippoNonbondedForce to common platform

* Creating OpenCL implementation of HippoNonbondedForce

* Continuing OpenCL implementation of HippoNonbondedForce

* Mostly finished OpenCL implementation of HippoNonbondedForce

* Eliminated three component vector types in host code

* Fix errors on CPU OpenCL

* Skip double precision tests for AMOEBA on OpenCL

* Bug fixes

* Bug fixes

* Fixed compilation error
parent 393a4dbd
Expand all Show whitespace changes
Inline Side-by-side
Showing with 3489 additions and 4942 deletions
plugins/amoeba/platforms/cuda/src/kernels/hippoFixedField.cu → plugins/amoeba/platforms/common/src/kernels/hippoFixedField.cc
View file @ 8e8923a7
......@@ -32,7 +32,7 @@ real bn3 = (5*bn2+alsq2n*exp2a)*invR2;
// Calculate the field at particle 1 due to multipoles at particle 2
real fdamp3, fdamp5, fdamp7;
computeDirectFieldDampingFactors(alpha2, r, fdamp3, fdamp5, fdamp7);
computeDirectFieldDampingFactors(alpha2, r, &fdamp3, &fdamp5, &fdamp7);
#ifndef COMPUTING_EXCEPTIONS
real scale = 1;
#endif
......@@ -58,7 +58,7 @@ tempField1 = -delta*factor2 - dipole2*rr3j + qDotDelta2*2*rr5j;
// Calculate the field at particle 2 due to multipoles at particle 1
computeDirectFieldDampingFactors(alpha1, r, fdamp3, fdamp5, fdamp7);
computeDirectFieldDampingFactors(alpha1, r, &fdamp3, &fdamp5, &fdamp7);
#ifdef USE_EWALD
real rr3i = bn1 - (1-scale*fdamp3)*invR3;
real rr5i = bn2 - (1-scale*fdamp5)*3*invR5;
......
plugins/amoeba/platforms/cuda/src/kernels/hippoMutualField.cu → plugins/amoeba/platforms/common/src/kernels/hippoMutualField.cc
View file @ 8e8923a7
real fdamp3, fdamp5;
computeMutualFieldDampingFactors(alpha1, alpha2, r, fdamp3, fdamp5);
computeMutualFieldDampingFactors(alpha1, alpha2, r, &fdamp3, &fdamp5);
#ifdef COMPUTING_EXCEPTIONS
fdamp3 *= scale;
fdamp5 *= scale;
......
plugins/amoeba/platforms/cuda/src/kernels/hippoNonbondedExceptions.cu → plugins/amoeba/platforms/common/src/kernels/hippoNonbondedExceptions.cc
View file @ 8e8923a7
......@@ -3,15 +3,15 @@
/**
* Compute exceptions for HIPPO.
*/
extern "C" __global__ void computeNonbondedExceptions(
unsigned long long* __restrict__ forceBuffers, mixed* __restrict__ energyBuffer, unsigned long long* __restrict__ torqueBuffers,
const real4* __restrict__ posq, const real3* __restrict__ extDipole, const int2* __restrict__ exceptionAtoms, const real* __restrict__ mmScale,
const real* __restrict__ dmScale, const real* __restrict__ ddScale, const real* __restrict__ dispScale, const real* __restrict__ repScale, const real* __restrict__ ctScale,
const real* __restrict__ coreCharge, const real* __restrict__ valenceCharge, const real* __restrict__ alpha, const real* __restrict__ epsilon,
const real* __restrict__ damping, const real* __restrict__ c6, const real* __restrict__ pauliK, const real* __restrict__ pauliQ,
const real* __restrict__ pauliAlpha, const real3* __restrict__ dipole, const real3* __restrict__ inducedDipole, const real* __restrict__ qXX,
const real* __restrict__ qXY, const real* __restrict__ qXZ, const real* __restrict__ qYY, const real* __restrict__ qYZ,
const real3* __restrict__ extrapolatedDipole
KERNEL void computeNonbondedExceptions(
GLOBAL mm_ulong* RESTRICT forceBuffers, GLOBAL mixed* RESTRICT energyBuffer, GLOBAL mm_ulong* RESTRICT torqueBuffers,
GLOBAL const real4* RESTRICT posq, GLOBAL const int2* RESTRICT exceptionAtoms, GLOBAL const real* RESTRICT mmScale,
GLOBAL const real* RESTRICT dmScale, GLOBAL const real* RESTRICT ddScale, GLOBAL const real* RESTRICT dispScale, GLOBAL const real* RESTRICT repScale, GLOBAL const real* RESTRICT ctScale,
GLOBAL const real* RESTRICT coreCharge, GLOBAL const real* RESTRICT valenceCharge, GLOBAL const real* RESTRICT alpha, GLOBAL const real* RESTRICT epsilon,
GLOBAL const real* RESTRICT damping, GLOBAL const real* RESTRICT c6, GLOBAL const real* RESTRICT pauliK, GLOBAL const real* RESTRICT pauliQ,
GLOBAL const real* RESTRICT pauliAlpha, GLOBAL const real* RESTRICT dipole, GLOBAL const real* RESTRICT inducedDipole, GLOBAL const real* RESTRICT qXX,
GLOBAL const real* RESTRICT qXY, GLOBAL const real* RESTRICT qXZ, GLOBAL const real* RESTRICT qYY, GLOBAL const real* RESTRICT qYZ,
GLOBAL const real* RESTRICT extrapolatedDipole
#ifdef USE_CUTOFF
, real4 periodicBoxSize, real4 invPeriodicBoxSize, real4 periodicBoxVecX, real4 periodicBoxVecY, real4 periodicBoxVec
#endif
......@@ -19,7 +19,7 @@ extern "C" __global__ void computeNonbondedExceptions(
mixed energy = 0;
const bool isExcluded = false;
const real interactionScale = 1.0f;
for (int index = blockIdx.x*blockDim.x+threadIdx.x; index < NUM_EXCEPTIONS; index += blockDim.x*gridDim.x) {
for (int index = GLOBAL_ID; index < NUM_EXCEPTIONS; index += GLOBAL_SIZE) {
int2 atoms = exceptionAtoms[index];
int atom1 = atoms.x;
int atom2 = atoms.y;
......@@ -42,8 +42,8 @@ extern "C" __global__ void computeNonbondedExceptions(
real pauliK1 = pauliK[atom1];
real pauliQ1 = pauliQ[atom1];
real pauliAlpha1 = pauliAlpha[atom1];
real3 dipole1 = dipole[atom1];
real3 inducedDipole1 = inducedDipole[atom1];
real3 dipole1 = make_real3(dipole[3*atom1], dipole[3*atom1+1], dipole[3*atom1+2]);
real3 inducedDipole1 = make_real3(inducedDipole[3*atom1], inducedDipole[3*atom1+1], inducedDipole[3*atom1+2]);
real qXX1 = qXX[atom1];
real qXY1 = qXY[atom1];
real qXZ1 = qXZ[atom1];
......@@ -58,8 +58,8 @@ extern "C" __global__ void computeNonbondedExceptions(
real pauliK2 = pauliK[atom2];
real pauliQ2 = pauliQ[atom2];
real pauliAlpha2 = pauliAlpha[atom2];
real3 dipole2 = dipole[atom2];
real3 inducedDipole2 = inducedDipole[atom2];
real3 dipole2 = make_real3(dipole[3*atom2], dipole[3*atom2+1], dipole[3*atom2+2]);
real3 inducedDipole2 = make_real3(inducedDipole[3*atom2], inducedDipole[3*atom2+1], inducedDipole[3*atom2+2]);
real qXX2 = qXX[atom2];
real qXY2 = qXY[atom2];
real qXZ2 = qXZ[atom2];
......@@ -79,21 +79,21 @@ extern "C" __global__ void computeNonbondedExceptions(
real tempEnergy = 0.0f;
COMPUTE_INTERACTION
energy += tempEnergy;
atomicAdd(&forceBuffers[atom1], static_cast<unsigned long long>((long long) (tempForce.x*0x100000000)));
atomicAdd(&forceBuffers[atom1+PADDED_NUM_ATOMS], static_cast<unsigned long long>((long long) (tempForce.y*0x100000000)));
atomicAdd(&forceBuffers[atom1+2*PADDED_NUM_ATOMS], static_cast<unsigned long long>((long long) (tempForce.z*0x100000000)));
atomicAdd(&forceBuffers[atom2], static_cast<unsigned long long>((long long) (-tempForce.x*0x100000000)));
atomicAdd(&forceBuffers[atom2+PADDED_NUM_ATOMS], static_cast<unsigned long long>((long long) (-tempForce.y*0x100000000)));
atomicAdd(&forceBuffers[atom2+2*PADDED_NUM_ATOMS], static_cast<unsigned long long>((long long) (-tempForce.z*0x100000000)));
atomicAdd(&torqueBuffers[atom1], static_cast<unsigned long long>((long long) (tempTorque1.x*0x100000000)));
atomicAdd(&torqueBuffers[atom1+PADDED_NUM_ATOMS], static_cast<unsigned long long>((long long) (tempTorque1.y*0x100000000)));
atomicAdd(&torqueBuffers[atom1+2*PADDED_NUM_ATOMS], static_cast<unsigned long long>((long long) (tempTorque1.z*0x100000000)));
atomicAdd(&torqueBuffers[atom2], static_cast<unsigned long long>((long long) (tempTorque2.x*0x100000000)));
atomicAdd(&torqueBuffers[atom2+PADDED_NUM_ATOMS], static_cast<unsigned long long>((long long) (tempTorque2.y*0x100000000)));
atomicAdd(&torqueBuffers[atom2+2*PADDED_NUM_ATOMS], static_cast<unsigned long long>((long long) (tempTorque2.z*0x100000000)));
ATOMIC_ADD(&forceBuffers[atom1], (mm_ulong) ((mm_long) (tempForce.x*0x100000000)));
ATOMIC_ADD(&forceBuffers[atom1+PADDED_NUM_ATOMS], (mm_ulong) ((mm_long) (tempForce.y*0x100000000)));
ATOMIC_ADD(&forceBuffers[atom1+2*PADDED_NUM_ATOMS], (mm_ulong) ((mm_long) (tempForce.z*0x100000000)));
ATOMIC_ADD(&forceBuffers[atom2], (mm_ulong) ((mm_long) (-tempForce.x*0x100000000)));
ATOMIC_ADD(&forceBuffers[atom2+PADDED_NUM_ATOMS], (mm_ulong) ((mm_long) (-tempForce.y*0x100000000)));
ATOMIC_ADD(&forceBuffers[atom2+2*PADDED_NUM_ATOMS], (mm_ulong) ((mm_long) (-tempForce.z*0x100000000)));
ATOMIC_ADD(&torqueBuffers[atom1], (mm_ulong) ((mm_long) (tempTorque1.x*0x100000000)));
ATOMIC_ADD(&torqueBuffers[atom1+PADDED_NUM_ATOMS], (mm_ulong) ((mm_long) (tempTorque1.y*0x100000000)));
ATOMIC_ADD(&torqueBuffers[atom1+2*PADDED_NUM_ATOMS], (mm_ulong) ((mm_long) (tempTorque1.z*0x100000000)));
ATOMIC_ADD(&torqueBuffers[atom2], (mm_ulong) ((mm_long) (tempTorque2.x*0x100000000)));
ATOMIC_ADD(&torqueBuffers[atom2+PADDED_NUM_ATOMS], (mm_ulong) ((mm_long) (tempTorque2.y*0x100000000)));
ATOMIC_ADD(&torqueBuffers[atom2+2*PADDED_NUM_ATOMS], (mm_ulong) ((mm_long) (tempTorque2.z*0x100000000)));
#ifdef USE_CUTOFF
}
#endif
}
energyBuffer[blockIdx.x*blockDim.x+threadIdx.x] += energy;
energyBuffer[GLOBAL_ID] += energy;
}
plugins/amoeba/platforms/common/src/kernels/multipolePme.cc 0 → 100644
View file @ 8e8923a7
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