YASK
Yet Another Stencil Kit: a software framework for creating HPC stencil code. Copyright 2014-2023 Intel Corporation.
|
Classes | |
class | yask::yk_factory |
Bootstrap factory to create a stencil solution. More... | |
class | yask::yk_env |
Kernel environment. More... | |
class | yask::yk_solution |
Stencil solution as defined by the generated code from the YASK stencil compiler. More... | |
class | yask::yk_stats |
Statistics from calls to run_solution(). More... | |
class | yask::yk_var |
A run-time YASK data container. More... | |
Typedefs | |
typedef std::shared_ptr< yk_env > | yask::yk_env_ptr |
Shared pointer to yk_env. | |
typedef std::shared_ptr< yk_solution > | yask::yk_solution_ptr |
Shared pointer to yk_solution. | |
typedef std::shared_ptr< yk_var > | yask::yk_var_ptr |
Shared pointer to yk_var. | |
typedef std::shared_ptr< yk_stats > | yask::yk_stats_ptr |
Shared pointer to yk_stats. | |
Variables | |
YASK_DEPRECATED typedef yk_var | yask::yk_grid |
[Deprecated] Use yk_var. | |
YASK_DEPRECATED typedef yk_var_ptr | yask::yk_grid_ptr |
[Deprecated] Use yk_var_ptr. | |
const int | yask::yask_numa_local = -1 |
Allocate vars on local NUMA node. | |
const int | yask::yask_numa_interleave = -2 |
Allocate vars across all available NUMA nodes. | |
const int | yask::yask_numa_none = -9 |
Do not specify any NUMA binding. | |
const int | yask::yask_numa_offload = -11 |
Do not specify any NUMA binding and use allocations optimized for offloading. | |
Types, clases, and functions used in both the YASK Stencil Kernel API.
const int yask::yask_numa_local = -1 |
Allocate vars on local NUMA node.
This is used in yk_solution::set_default_numa_preferred and yk_var::set_numa_preferred. In Python, specify as yask_kernel.cvar.yask_numa_local
.
const int yask::yask_numa_interleave = -2 |
Allocate vars across all available NUMA nodes.
This is used in yk_solution::set_default_numa_preferred and yk_var::set_numa_preferred. In Python, specify as yask_kernel.cvar.yask_numa_interleave
.
const int yask::yask_numa_none = -9 |
Do not specify any NUMA binding.
This is used in yk_solution::set_default_numa_preferred and yk_var::set_numa_preferred. In Python, specify as yask_kernel.cvar.yask_numa_none
.
const int yask::yask_numa_offload = -11 |
Do not specify any NUMA binding and use allocations optimized for offloading.
This is used in yk_solution::set_default_numa_preferred and yk_var::set_numa_preferred. In Python, specify as yask_kernel.cvar.yask_numa_host
.