Defines the PME GPU data structures (the GPU function parameters used both on host and device sides).
- Author
- Aleksei Iupinov a.yup.nosp@m.inov.nosp@m.@gmai.nosp@m.l.co.nosp@m.m
|
struct | PmeGpuConstParams |
| A GPU data structure for storing the constant PME data. This only has to be initialized once. More...
|
|
struct | PmeGpuGridParams |
| A GPU data structure for storing the PME data related to the grid sizes and cut-off. This only has to be updated at every DD step. More...
|
|
struct | PmeGpuAtomParams |
| A GPU data structure for storing the PME data of the atoms, local to this process' domain partition. This only has to be updated every DD step. More...
|
|
struct | PmeGpuDynamicParams |
| A GPU data structure for storing the PME data which might change for each new PME computation. More...
|
|
struct | PmeGpuKernelParamsBase |
| A single structure encompassing all the PME data used in GPU kernels on device. To extend the list with platform-specific parameters, this can be inherited by the GPU framework-specific structure. More...
|
|
#define HIDE_FROM_OPENCL_COMPILER |
( |
|
x | ) |
x |
A workaround to hide DeviceBuffer template from OpenCL kernel compilation.
- to turn it into a dummy of the same size as host implementation of device buffer. As we only care about 64-bit, 8 bytes is fine. TODO: what we should be doing is providing separate device-side views of the same structures - then there would be no need for macro.