tThreads per block limited to 256 - sphere - GPU-based 3D discrete element method algorithm with optional fluid coupling
(HTM) git clone git://src.adamsgaard.dk/sphere
(DIR) Log
(DIR) Files
(DIR) Refs
(DIR) LICENSE
---
(DIR) commit bcca6eaf973e97dbcf0d970ce22ff944864f0276
(DIR) parent 6bbe20b846311d165bb831422b509c93f00bfa30
(HTM) Author: Anders Damsgaard <adc@geo.au.dk>
Date: Wed, 13 Mar 2013 23:58:38 +0100
Threads per block limited to 256
Diffstat:
M src/device.cu | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
---
(DIR) diff --git a/src/device.cu b/src/device.cu
t@@ -529,6 +529,8 @@ __host__ void DEM::startTime()
unsigned int blocksPerGrid = iDivUp(np, threadsPerBlock);
dim3 dimGrid(blocksPerGrid, 1, 1); // Blocks arranged in 1D grid
dim3 dimBlock(threadsPerBlock, 1, 1); // Threads arranged in 1D block
+ unsigned int blocksPerGridBonds = iDivUp(params.nb0, threadsPerBlock);
+ dim3 dimGridBonds(blocksPerGridBonds, 1, 1); // Blocks arranged in 1D grid
// Shared memory per block
unsigned int smemSize = sizeof(unsigned int)*(threadsPerBlock+1);
t@@ -731,7 +733,8 @@ __host__ void DEM::startTime()
// Process particle pairs
if (params.nb0 > 0) {
- bondsLinear<<< 1, params.nb0 >>>(
+ //bondsLinear<<< 1, params.nb0 >>>(
+ bondsLinear<<<dimGridBonds, dimBlock>>>(
dev_bonds,
dev_bonds_delta,
dev_bonds_omega,