Skip to content

GitLab

Commit 773dd0f0 authored by Peter Eastman's avatar Peter Eastman
Browse files

Optimizations to PME

parent 95f1884c
Show whitespace changes
Inline Side-by-side
Showing with 22 additions and 36 deletions
plugins/amoeba/platforms/cuda/src/kernels/kCalculateAmoebaCudaPME.cu
View file @ 773dd0f0
...@@ -206,13 +206,7 @@ void kFindAmoebaAtomRangeForGrid_kernel() ...@@ -206,13 +206,7 @@ void kFindAmoebaAtomRangeForGrid_kernel()
} }
} }
__global__ __global__
#if (__CUDA_ARCH__ >= 200) __launch_bounds__(64, 10)
__launch_bounds__(768, 1)
#elif (__CUDA_ARCH__ >= 120)
__launch_bounds__(384, 1)
#else
__launch_bounds__(192, 1)
#endif
void kGridSpreadFixedMultipoles_kernel() void kGridSpreadFixedMultipoles_kernel()
{ {
const float xscale = cSim.pmeGridSize.x*cSim.invPeriodicBoxSizeX; const float xscale = cSim.pmeGridSize.x*cSim.invPeriodicBoxSizeX;
...@@ -303,13 +297,7 @@ void kGridSpreadFixedMultipoles_kernel() ...@@ -303,13 +297,7 @@ void kGridSpreadFixedMultipoles_kernel()
} }
__global__ __global__
#if (__CUDA_ARCH__ >= 200) __launch_bounds__(64, 10)
__launch_bounds__(768, 1)
#elif (__CUDA_ARCH__ >= 120)
__launch_bounds__(384, 1)
#else
__launch_bounds__(192, 1)
#endif
void kGridSpreadInducedDipoles_kernel() void kGridSpreadInducedDipoles_kernel()
{ {
const float xscale = cSim.pmeGridSize.x*cSim.invPeriodicBoxSizeX; const float xscale = cSim.pmeGridSize.x*cSim.invPeriodicBoxSizeX;
...@@ -435,11 +423,11 @@ void kAmoebaReciprocalConvolution_kernel() ...@@ -435,11 +423,11 @@ void kAmoebaReciprocalConvolution_kernel()
__global__ __global__
#if (__CUDA_ARCH__ >= 200) #if (__CUDA_ARCH__ >= 200)
__launch_bounds__(768, 1)
#elif (__CUDA_ARCH__ >= 120)
__launch_bounds__(384, 1) __launch_bounds__(384, 1)
#else #elif (__CUDA_ARCH__ >= 120)
__launch_bounds__(192, 1) __launch_bounds__(192, 1)
#else
__launch_bounds__(96, 1)
#endif #endif
void kComputeFixedPotentialFromGrid_kernel() void kComputeFixedPotentialFromGrid_kernel()
{ {
...@@ -551,11 +539,11 @@ void kComputeFixedPotentialFromGrid_kernel() ...@@ -551,11 +539,11 @@ void kComputeFixedPotentialFromGrid_kernel()
__global__ __global__
#if (__CUDA_ARCH__ >= 200) #if (__CUDA_ARCH__ >= 200)
__launch_bounds__(768, 1) __launch_bounds__(256, 1)
#elif (__CUDA_ARCH__ >= 120) #elif (__CUDA_ARCH__ >= 120)
__launch_bounds__(384, 1) __launch_bounds__(128, 1)
#else #else
__launch_bounds__(192, 1) __launch_bounds__(64, 1)
#endif #endif
void kComputeInducedPotentialFromGrid_kernel() void kComputeInducedPotentialFromGrid_kernel()
{ {
...@@ -978,15 +966,9 @@ void kCalculateAmoebaPMEFixedMultipoleField(amoebaGpuContext amoebaGpu) ...@@ -978,15 +966,9 @@ void kCalculateAmoebaPMEFixedMultipoleField(amoebaGpuContext amoebaGpu)
{ {
// Compute B-spline coefficients and sort the atoms. // Compute B-spline coefficients and sort the atoms.
int threads;
gpuContext gpu = amoebaGpu->gpuContext; gpuContext gpu = amoebaGpu->gpuContext;
if (gpu->sm_version >= SM_20) int bsplineThreads = (gpu->sm_version >= SM_20 ? 448 : (gpu->sm_version >= SM_12 ? 160 : 160));
threads = 448; kComputeAmoebaBsplines_kernel<<<gpu->sim.blocks, bsplineThreads, bsplineThreads*AMOEBA_PME_ORDER*AMOEBA_PME_ORDER*sizeof(float)>>>();
else if (gpu->sm_version >= SM_12)
threads = 160;
else
threads = 160;
kComputeAmoebaBsplines_kernel<<<gpu->sim.blocks, threads, threads*AMOEBA_PME_ORDER*AMOEBA_PME_ORDER*sizeof(float)>>>();
LAUNCHERROR("kComputeAmoebaBsplines"); LAUNCHERROR("kComputeAmoebaBsplines");
bbSort(gpu->psPmeAtomGridIndex->_pDevData, gpu->natoms); bbSort(gpu->psPmeAtomGridIndex->_pDevData, gpu->natoms);
kFindAmoebaAtomRangeForGrid_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(); kFindAmoebaAtomRangeForGrid_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>();
...@@ -994,13 +976,14 @@ void kCalculateAmoebaPMEFixedMultipoleField(amoebaGpuContext amoebaGpu) ...@@ -994,13 +976,14 @@ void kCalculateAmoebaPMEFixedMultipoleField(amoebaGpuContext amoebaGpu)
// Perform PME for the fixed multipoles. // Perform PME for the fixed multipoles.
kGridSpreadFixedMultipoles_kernel<<<8*gpu->sim.blocks, 64>>>(); kGridSpreadFixedMultipoles_kernel<<<10*gpu->sim.blocks, 64>>>();
LAUNCHERROR("kGridSpreadFixedMultipoles"); LAUNCHERROR("kGridSpreadFixedMultipoles");
cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_FORWARD); cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_FORWARD);
kAmoebaReciprocalConvolution_kernel<<<gpu->sim.blocks, gpu->sim.nonbond_threads_per_block>>>(); kAmoebaReciprocalConvolution_kernel<<<gpu->sim.blocks, gpu->sim.nonbond_threads_per_block>>>();
LAUNCHERROR("kAmoebaReciprocalConvolution"); LAUNCHERROR("kAmoebaReciprocalConvolution");
cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_INVERSE); cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_INVERSE);
kComputeFixedPotentialFromGrid_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(); int potentialThreads = (gpu->sm_version >= SM_20 ? 384 : (gpu->sm_version >= SM_12 ? 192 : 96));
kComputeFixedPotentialFromGrid_kernel<<<gpu->sim.blocks, potentialThreads>>>();
LAUNCHERROR("kComputeFixedPotentialFromGrid"); LAUNCHERROR("kComputeFixedPotentialFromGrid");
kRecordFixedMultipoleField_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(amoebaGpu->psE_Field->_pDevData); kRecordFixedMultipoleField_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(amoebaGpu->psE_Field->_pDevData);
LAUNCHERROR("kRecordFixedMultipoleField"); LAUNCHERROR("kRecordFixedMultipoleField");
...@@ -1014,13 +997,14 @@ void kCalculateAmoebaPMEInducedDipoleField(amoebaGpuContext amoebaGpu) ...@@ -1014,13 +997,14 @@ void kCalculateAmoebaPMEInducedDipoleField(amoebaGpuContext amoebaGpu)
// Perform PME for the induced dipoles. // Perform PME for the induced dipoles.
gpuContext gpu = amoebaGpu->gpuContext; gpuContext gpu = amoebaGpu->gpuContext;
kGridSpreadInducedDipoles_kernel<<<8*gpu->sim.blocks, 64>>>(); kGridSpreadInducedDipoles_kernel<<<10*gpu->sim.blocks, 64>>>();
LAUNCHERROR("kGridSpreadInducedDipoles"); LAUNCHERROR("kGridSpreadInducedDipoles");
cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_FORWARD); cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_FORWARD);
kAmoebaReciprocalConvolution_kernel<<<gpu->sim.blocks, gpu->sim.nonbond_threads_per_block>>>(); kAmoebaReciprocalConvolution_kernel<<<gpu->sim.blocks, gpu->sim.nonbond_threads_per_block>>>();
LAUNCHERROR("kAmoebaReciprocalConvolution"); LAUNCHERROR("kAmoebaReciprocalConvolution");
cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_INVERSE); cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_INVERSE);
kComputeInducedPotentialFromGrid_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(); int potentialThreads = (gpu->sm_version >= SM_20 ? 256 : (gpu->sm_version >= SM_12 ? 128 : 64));
kComputeInducedPotentialFromGrid_kernel<<<gpu->sim.blocks, potentialThreads>>>();
LAUNCHERROR("kComputeInducedPotentialFromGrid"); LAUNCHERROR("kComputeInducedPotentialFromGrid");
kRecordInducedDipoleField_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(amoebaGpu->psWorkVector[0]->_pDevData, amoebaGpu->psWorkVector[1]->_pDevData); kRecordInducedDipoleField_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(amoebaGpu->psWorkVector[0]->_pDevData, amoebaGpu->psWorkVector[1]->_pDevData);
LAUNCHERROR("kRecordInducedDipoleField"); LAUNCHERROR("kRecordInducedDipoleField");
...@@ -1034,13 +1018,14 @@ void kCalculateAmoebaPME(amoebaGpuContext amoebaGpu) ...@@ -1034,13 +1018,14 @@ void kCalculateAmoebaPME(amoebaGpuContext amoebaGpu)
// Perform PME for the fixed multipoles. // Perform PME for the fixed multipoles.
gpuContext gpu = amoebaGpu->gpuContext; gpuContext gpu = amoebaGpu->gpuContext;
kGridSpreadFixedMultipoles_kernel<<<8*gpu->sim.blocks, 64>>>(); kGridSpreadFixedMultipoles_kernel<<<10*gpu->sim.blocks, 64>>>();
LAUNCHERROR("kGridSpreadFixedMultipoles"); LAUNCHERROR("kGridSpreadFixedMultipoles");
cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_FORWARD); cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_FORWARD);
kAmoebaReciprocalConvolution_kernel<<<gpu->sim.blocks, gpu->sim.nonbond_threads_per_block>>>(); kAmoebaReciprocalConvolution_kernel<<<gpu->sim.blocks, gpu->sim.nonbond_threads_per_block>>>();
LAUNCHERROR("kAmoebaReciprocalConvolution"); LAUNCHERROR("kAmoebaReciprocalConvolution");
cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_INVERSE); cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_INVERSE);
kComputeFixedPotentialFromGrid_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(); int potentialThreads = (gpu->sm_version >= SM_20 ? 384 : (gpu->sm_version >= SM_12 ? 192 : 96));
kComputeFixedPotentialFromGrid_kernel<<<gpu->sim.blocks, potentialThreads>>>();
LAUNCHERROR("kComputeFixedPotentialFromGrid"); LAUNCHERROR("kComputeFixedPotentialFromGrid");
kComputeFixedMultipoleForceAndEnergy_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(); kComputeFixedMultipoleForceAndEnergy_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>();
LAUNCHERROR("kComputeFixedMultipoleForceAndEnergy"); LAUNCHERROR("kComputeFixedMultipoleForceAndEnergy");
...@@ -1048,13 +1033,14 @@ void kCalculateAmoebaPME(amoebaGpuContext amoebaGpu) ...@@ -1048,13 +1033,14 @@ void kCalculateAmoebaPME(amoebaGpuContext amoebaGpu)
// Perform PME for the induced dipoles. // Perform PME for the induced dipoles.
kGridSpreadInducedDipoles_kernel<<<8*gpu->sim.blocks, 64>>>(); kGridSpreadInducedDipoles_kernel<<<10*gpu->sim.blocks, 64>>>();
LAUNCHERROR("kGridSpreadInducedDipoles"); LAUNCHERROR("kGridSpreadInducedDipoles");
cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_FORWARD); cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_FORWARD);
kAmoebaReciprocalConvolution_kernel<<<gpu->sim.blocks, gpu->sim.nonbond_threads_per_block>>>(); kAmoebaReciprocalConvolution_kernel<<<gpu->sim.blocks, gpu->sim.nonbond_threads_per_block>>>();
LAUNCHERROR("kAmoebaReciprocalConvolution"); LAUNCHERROR("kAmoebaReciprocalConvolution");
cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_INVERSE); cufftExecC2C(gpu->fftplan, gpu->psPmeGrid->_pDevData, gpu->psPmeGrid->_pDevData, CUFFT_INVERSE);
kComputeInducedPotentialFromGrid_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(); potentialThreads = (gpu->sm_version >= SM_20 ? 256 : (gpu->sm_version >= SM_12 ? 128 : 64));
kComputeInducedPotentialFromGrid_kernel<<<gpu->sim.blocks, potentialThreads>>>();
LAUNCHERROR("kComputeInducedPotentialFromGrid"); LAUNCHERROR("kComputeInducedPotentialFromGrid");
kComputeInducedDipoleForceAndEnergy_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>(); kComputeInducedDipoleForceAndEnergy_kernel<<<gpu->sim.blocks, gpu->sim.update_threads_per_block>>>();
LAUNCHERROR("kComputeInducedDipoleForceAndEnergy"); LAUNCHERROR("kComputeInducedDipoleForceAndEnergy");
......
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