Skip to content

GitLab

"vscode:/vscode.git/clone" did not exist on "18f78c1eb0e32c3d6c9c8e00f1bde059af68bb9a"
Commit 06db7ac4 authored by peastman's avatar peastman
Browse files

Very minor optimization

parent 291484f2
Show whitespace changes
Inline Side-by-side
Showing with 62 additions and 62 deletions
platforms/cuda/src/kernels/customGBEnergyN2.cu
View file @ 06db7ac4
......@@ -65,7 +65,7 @@ extern "C" __global__ void computeN2Energy(unsigned long long* __restrict__ forc
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
real dEdR = 0;
......@@ -117,7 +117,7 @@ extern "C" __global__ void computeN2Energy(unsigned long long* __restrict__ forc
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+tj;
real dEdR = 0;
......@@ -268,7 +268,7 @@ extern "C" __global__ void computeN2Energy(unsigned long long* __restrict__ forc
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
real dEdR = 0;
......@@ -313,7 +313,7 @@ extern "C" __global__ void computeN2Energy(unsigned long long* __restrict__ forc
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
real dEdR = 0;
......
platforms/cuda/src/kernels/customGBValueN2.cu
View file @ 06db7ac4
......@@ -60,7 +60,7 @@ extern "C" __global__ void computeN2Value(const real4* __restrict__ posq, const
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
real tempValue1 = 0;
......@@ -108,7 +108,7 @@ extern "C" __global__ void computeN2Value(const real4* __restrict__ posq, const
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+tj;
real tempValue1 = 0;
......@@ -241,7 +241,7 @@ extern "C" __global__ void computeN2Value(const real4* __restrict__ posq, const
real r2 = delta.x*delta.x + delta.y*delta.y + delta.z*delta.z;
if (r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
real tempValue1 = 0;
......@@ -275,7 +275,7 @@ extern "C" __global__ void computeN2Value(const real4* __restrict__ posq, const
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
real tempValue1 = 0;
......
platforms/cuda/src/kernels/customNonbondedGroups.cu
View file @ 06db7ac4
......@@ -58,7 +58,7 @@ extern "C" __global__ void computeInteractionGroups(
if (!isExcluded && r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
real dEdR = 0.0f;
real tempEnergy = 0.0f;
......
platforms/cuda/src/kernels/gbsaObc1.cu
View file @ 06db7ac4
......@@ -116,7 +116,7 @@ extern "C" __global__ void computeBornSum(unsigned long long* __restrict__ globa
if (atom1 < NUM_ATOMS && y*TILE_SIZE+j < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = make_float2(localData[tbx+j].radius, localData[tbx+j].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if ((j != tgx) && (params1.x < rScaledRadiusJ)) {
......@@ -163,7 +163,7 @@ extern "C" __global__ void computeBornSum(unsigned long long* __restrict__ globa
if (atom1 < NUM_ATOMS && y*TILE_SIZE+tj < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = make_float2(localData[tbx+tj].radius, localData[tbx+tj].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if (params1.x < rScaledRadiusJ) {
......@@ -305,7 +305,7 @@ extern "C" __global__ void computeBornSum(unsigned long long* __restrict__ globa
int atom2 = atomIndices[tbx+tj];
if (atom1 < NUM_ATOMS && atom2 < NUM_ATOMS && r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = make_float2(localData[tbx+tj].radius, localData[tbx+tj].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if (params1.x < rScaledRadiusJ) {
......@@ -355,7 +355,7 @@ extern "C" __global__ void computeBornSum(unsigned long long* __restrict__ globa
if (atom1 < NUM_ATOMS && atom2 < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = make_float2(localData[tbx+tj].radius, localData[tbx+tj].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if (params1.x < rScaledRadiusJ) {
......@@ -461,7 +461,7 @@ extern "C" __global__ void computeGBSAForce1(unsigned long long* __restrict__ fo
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[tbx+j].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......@@ -518,7 +518,7 @@ extern "C" __global__ void computeGBSAForce1(unsigned long long* __restrict__ fo
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[tbx+tj].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......@@ -672,7 +672,7 @@ extern "C" __global__ void computeGBSAForce1(unsigned long long* __restrict__ fo
real r2 = delta.x*delta.x + delta.y*delta.y + delta.z*delta.z;
if (r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[tbx+tj].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......@@ -723,7 +723,7 @@ extern "C" __global__ void computeGBSAForce1(unsigned long long* __restrict__ fo
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[tbx+tj].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......
platforms/cuda/src/kernels/nonbonded.cu
View file @ 06db7ac4
......@@ -161,7 +161,7 @@ extern "C" __global__ void computeNonbonded(
#endif
real r2 = delta.x*delta.x + delta.y*delta.y + delta.z*delta.z;
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
#ifdef USE_SYMMETRIC
......@@ -232,7 +232,7 @@ extern "C" __global__ void computeNonbonded(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+tj;
#ifdef USE_SYMMETRIC
......@@ -433,7 +433,7 @@ extern "C" __global__ void computeNonbonded(
real r2 = delta.x*delta.x + delta.y*delta.y + delta.z*delta.z;
if (r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
#ifdef USE_SYMMETRIC
......@@ -507,7 +507,7 @@ extern "C" __global__ void computeNonbonded(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
#ifdef USE_SYMMETRIC
......
platforms/opencl/src/kernels/customGBEnergyN2.cl
View file @ 06db7ac4
......@@ -67,7 +67,7 @@ __kernel void computeN2Energy(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
real dEdR = 0;
......@@ -117,7 +117,7 @@ __kernel void computeN2Energy(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+tj;
real dEdR = 0;
......@@ -279,7 +279,7 @@ __kernel void computeN2Energy(
real r2 = delta.x*delta.x + delta.y*delta.y + delta.z*delta.z;
if (r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
real dEdR = 0;
......@@ -317,7 +317,7 @@ __kernel void computeN2Energy(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
real dEdR = 0;
......
platforms/opencl/src/kernels/customGBEnergyN2_cpu.cl
View file @ 06db7ac4
......@@ -67,7 +67,7 @@ __kernel void computeN2Energy(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
......@@ -133,7 +133,7 @@ __kernel void computeN2Energy(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
......@@ -285,7 +285,7 @@ __kernel void computeN2Energy(
real r2 = dot(delta.xyz, delta.xyz);
if (atom1 < NUM_ATOMS && atomIndices[j] < NUM_ATOMS && r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[j];
......@@ -341,7 +341,7 @@ __kernel void computeN2Energy(
if (atom1 < NUM_ATOMS && atomIndices[j] < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[j];
......
platforms/opencl/src/kernels/customGBValueN2.cl
View file @ 06db7ac4
......@@ -59,7 +59,7 @@ __kernel void computeN2Value(__global const real4* restrict posq, __local real4*
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
real tempValue1 = 0;
......@@ -107,7 +107,7 @@ __kernel void computeN2Value(__global const real4* restrict posq, __local real4*
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+tj;
real tempValue1 = 0;
......@@ -252,7 +252,7 @@ __kernel void computeN2Value(__global const real4* restrict posq, __local real4*
real r2 = delta.x*delta.x + delta.y*delta.y + delta.z*delta.z;
if (r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
real tempValue1 = 0;
......@@ -285,7 +285,7 @@ __kernel void computeN2Value(__global const real4* restrict posq, __local real4*
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
real tempValue1 = 0;
......
platforms/opencl/src/kernels/customGBValueN2_cpu.cl
View file @ 06db7ac4
......@@ -59,7 +59,7 @@ __kernel void computeN2Value(__global const real4* restrict posq, __local real4*
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
......@@ -116,7 +116,7 @@ __kernel void computeN2Value(__global const real4* restrict posq, __local real4*
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
......@@ -251,7 +251,7 @@ __kernel void computeN2Value(__global const real4* restrict posq, __local real4*
real r2 = dot(delta.xyz, delta.xyz);
if (atom1 < NUM_ATOMS && atomIndices[j] < NUM_ATOMS && r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[j];
......@@ -296,7 +296,7 @@ __kernel void computeN2Value(__global const real4* restrict posq, __local real4*
if (atom1 < NUM_ATOMS && atomIndices[j] < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[j];
......
platforms/opencl/src/kernels/customNonbondedGroups.cl
View file @ 06db7ac4
......@@ -91,7 +91,7 @@ __kernel void computeInteractionGroups(
if (!isExcluded && r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
real dEdR = 0.0f;
real tempEnergy = 0.0f;
......
platforms/opencl/src/kernels/gbsaObc.cl
View file @ 06db7ac4
......@@ -67,7 +67,7 @@ __kernel void computeBornSum(
if (atom1 < NUM_ATOMS && y*TILE_SIZE+j < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = (float2) (localData[tbx+j].radius, localData[tbx+j].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if ((j != tgx) && (params1.x < rScaledRadiusJ)) {
......@@ -114,7 +114,7 @@ __kernel void computeBornSum(
if (atom1 < NUM_ATOMS && y*TILE_SIZE+tj < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = (float2) (localData[tbx+tj].radius, localData[tbx+tj].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if (params1.x < rScaledRadiusJ) {
......@@ -268,7 +268,7 @@ __kernel void computeBornSum(
int atom2 = atomIndices[tbx+tj];
if (atom1 < NUM_ATOMS && atom2 < NUM_ATOMS && r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = (float2) (localData[tbx+tj].radius, localData[tbx+tj].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if (params1.x < rScaledRadiusJ) {
......@@ -317,7 +317,7 @@ __kernel void computeBornSum(
if (atom1 < NUM_ATOMS && atom2 < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = (float2) (localData[tbx+tj].radius, localData[tbx+tj].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if (params1.x < rScaledRadiusJ) {
......@@ -437,7 +437,7 @@ __kernel void computeGBSAForce1(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[tbx+j].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......@@ -492,7 +492,7 @@ __kernel void computeGBSAForce1(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[tbx+tj].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......@@ -659,7 +659,7 @@ __kernel void computeGBSAForce1(
real r2 = delta.x*delta.x + delta.y*delta.y + delta.z*delta.z;
if (r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[tbx+tj].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......@@ -707,7 +707,7 @@ __kernel void computeGBSAForce1(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[tbx+tj].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......
platforms/opencl/src/kernels/gbsaObc_cpu.cl
View file @ 06db7ac4
......@@ -71,7 +71,7 @@ __kernel void computeBornSum(
if (atom1 < NUM_ATOMS && y*TILE_SIZE+j < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = (float2) (localData[j].radius, localData[j].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if ((j != tgx) && (params1.x < rScaledRadiusJ)) {
......@@ -120,7 +120,7 @@ __kernel void computeBornSum(
if (atom1 < NUM_ATOMS && y*TILE_SIZE+j < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = (float2) (localData[j].radius, localData[j].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if (params1.x < rScaledRadiusJ) {
......@@ -269,7 +269,7 @@ __kernel void computeBornSum(
int atom2 = atomIndices[j];
if (atom1 < NUM_ATOMS && atom2 < NUM_ATOMS && r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = (float2) (localData[j].radius, localData[j].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if (params1.x < rScaledRadiusJ) {
......@@ -331,7 +331,7 @@ __kernel void computeBornSum(
if (atom1 < NUM_ATOMS && atom2 < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
float2 params2 = (float2) (localData[j].radius, localData[j].scaledRadius);
real rScaledRadiusJ = r+params2.y;
if (params1.x < rScaledRadiusJ) {
......@@ -461,7 +461,7 @@ __kernel void computeGBSAForce1(
if (atom1 < NUM_ATOMS && y*TILE_SIZE+j < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[j].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......@@ -520,7 +520,7 @@ __kernel void computeGBSAForce1(
if (atom1 < NUM_ATOMS && y*TILE_SIZE+j < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[j].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......@@ -677,7 +677,7 @@ __kernel void computeGBSAForce1(
int atom2 = atomIndices[j];
if (atom1 < NUM_ATOMS && atom2 < NUM_ATOMS && r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[j].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......@@ -737,7 +737,7 @@ __kernel void computeGBSAForce1(
if (atom1 < NUM_ATOMS && atom2 < NUM_ATOMS) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
real bornRadius2 = localData[j].bornRadius;
real alpha2_ij = bornRadius1*bornRadius2;
real D_ij = r2*RECIP(4.0f*alpha2_ij);
......
platforms/opencl/src/kernels/nonbonded.cl
View file @ 06db7ac4
......@@ -71,7 +71,7 @@ __kernel void computeNonbonded(
#endif
real r2 = delta.x*delta.x + delta.y*delta.y + delta.z*delta.z;
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
#ifdef USE_SYMMETRIC
......@@ -128,7 +128,7 @@ __kernel void computeNonbonded(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+tj;
#ifdef USE_SYMMETRIC
......@@ -297,7 +297,7 @@ __kernel void computeNonbonded(
real r2 = delta.x*delta.x + delta.y*delta.y + delta.z*delta.z;
if (r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
#ifdef USE_SYMMETRIC
......@@ -347,7 +347,7 @@ __kernel void computeNonbonded(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[tbx+tj];
#ifdef USE_SYMMETRIC
......
platforms/opencl/src/kernels/nonbonded_cpu.cl
View file @ 06db7ac4
......@@ -72,7 +72,7 @@ __kernel void computeNonbonded(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
......@@ -140,7 +140,7 @@ __kernel void computeNonbonded(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = y*TILE_SIZE+j;
......@@ -307,7 +307,7 @@ __kernel void computeNonbonded(
real r2 = dot(delta.xyz, delta.xyz);
if (r2 < CUTOFF_SQUARED) {
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[j];
......@@ -371,7 +371,7 @@ __kernel void computeNonbonded(
if (r2 < CUTOFF_SQUARED) {
#endif
real invR = RSQRT(r2);
real r = RECIP(invR);
real r = r2*invR;
unsigned int atom2 = j;
LOAD_ATOM2_PARAMETERS
atom2 = atomIndices[j];
......
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