B | |
| block [Kernel] |
A block is a 3 dimension group of threads
|
C | |
| clDeviceExecutionCapabilities [Devices] | |
| clDeviceFPConfig [Devices] | |
| clDeviceGlobalMemCacheType [Devices] | |
| clDeviceID [Devices] | |
| clDeviceLocalMemType [Devices] | |
| clDeviceQueueProperties [Devices] | |
| context [Devices] | |
| couple [Vector] | |
| cudaInfo [Devices] |
Specific informations for Cuda-managed devices
|
| cuda_extra [Kernel.Cuda] | |
| custom [Vector] |
Spoc offers many predefined vectors types.
|
| customarray [Vector] | |
D | |
| deviceType [Devices] |
Device type GPU|CPU|Accelerator
|
| device_vec [Vector] | |
| dim3 [Devices] | |
G | |
| generalInfo [Devices] |
General informations shared
by all devices (managed by Cuda or OpenCL)
|
| grid [Kernel] |
A grid is a 3 dimension group of blocks
|
K | |
| kernel [Kernel] |
Manages Kernels
|
| kernelArgs [Kernel] |
type of parameters usable with kernels
|
| kind [Vector] | |
O | |
| openCLInfo [Devices] |
Specific informations for OpenCL-managed devices
|
P | |
| platformInfo [Devices] |
Informations about the OpenCL platform
|
S | |
| specificLibrary [Devices] |
Different frameworks Spoc can use
(currently Nvidia Cuda and OpenCL)
|
| spoc_vec [Vector] |
a spoc_vector is a Bigarray or a custom vector
|
| sub [Vector] |
sub vectors are vector parts sharing memory space on cpu memory BUT not on gpu memory,
allowing easy computation distribution over multiple GPUs.
|
V | |
| vec_device [Vector] | |
| vector [Vector] |
a vector represents every information needed by Spoc to manage it
It uses Bigarrays to manage data on the cpu side (see the OCaml Bigarray Module for more informations)
|