Skip to content

GitLab

Commit ca63e6d8 authored by Mark Friedrichs's avatar Mark Friedrichs
Browse files

Fix allocation of shared memory for shake kernels

parent 8868141e
Show whitespace changes
Inline Side-by-side
Showing with 7 additions and 17 deletions
platforms/cuda/src/kernels/kShakeH.cu
View file @ ca63e6d8
......@@ -74,13 +74,7 @@ __launch_bounds__(G8X_SHAKE_THREADS_PER_BLOCK, 1)
#endif
void kApplyFirstShake_kernel()
{
#if (__CUDA_ARCH__ >= 200)
__shared__ Atom sA[GF1XX_SHAKE_THREADS_PER_BLOCK];
#elif (__CUDA_ARCH__ >= 130)
__shared__ Atom sA[GT2XX_SHAKE_THREADS_PER_BLOCK];
#else
__shared__ Atom sA[G8X_SHAKE_THREADS_PER_BLOCK];
#endif
extern __shared__ Atom sA[];
Atom* psA = &sA[threadIdx.x];
unsigned int pos = threadIdx.x + blockIdx.x * blockDim.x;
while (pos < cSim.ShakeConstraints)
......@@ -235,9 +229,10 @@ void kApplyFirstShake(gpuContext gpu)
// printf("kApplyFirstShake\n");
if (gpu->sim.ShakeConstraints > 0)
{
kApplyFirstShake_kernel<<<gpu->sim.blocks, gpu->sim.shake_threads_per_block>>>();
kApplyFirstShake_kernel<<<gpu->sim.blocks, gpu->sim.shake_threads_per_block, sizeof(Atom)*gpu->sim.shake_threads_per_block>>>();
LAUNCHERROR("kApplyFirstShake");
}
}
__global__
......@@ -250,13 +245,7 @@ __launch_bounds__(G8X_SHAKE_THREADS_PER_BLOCK, 1)
#endif
void kApplySecondShake_kernel()
{
#if (__CUDA_ARCH__ >= 200)
__shared__ Atom sA[GF1XX_SHAKE_THREADS_PER_BLOCK];
#elif (__CUDA_ARCH__ >= 130)
__shared__ Atom sA[GT2XX_SHAKE_THREADS_PER_BLOCK];
#else
__shared__ Atom sA[G8X_SHAKE_THREADS_PER_BLOCK];
#endif
extern __shared__ Atom sA[];
Atom* psA = &sA[threadIdx.x];
unsigned int pos = threadIdx.x + blockIdx.x * blockDim.x;
while (pos < cSim.ShakeConstraints)
......@@ -468,7 +457,7 @@ void kApplySecondShake(gpuContext gpu)
// printf("kApplySecondShake\n");
if (gpu->sim.ShakeConstraints > 0)
{
kApplySecondShake_kernel<<<gpu->sim.blocks, gpu->sim.shake_threads_per_block>>>();
kApplySecondShake_kernel<<<gpu->sim.blocks, gpu->sim.shake_threads_per_block, sizeof(Atom)*gpu->sim.shake_threads_per_block>>>();
LAUNCHERROR("kApplySecondShake");
}
else if (gpu->sim.NonShakeConstraints > 0)
......@@ -477,5 +466,6 @@ void kApplySecondShake(gpuContext gpu)
kApplyNoShake_kernel<<<gpu->sim.blocks, gpu->sim.nonshake_threads_per_block>>>();
LAUNCHERROR("kApplyNoShake");
}
}
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