|
Uni20
Asynchronous tensor and runtime library.
|
cuSOLVER handle slots and dynamically paired stream execution leases. More...
#include <uni20/backend/cuda/resource_pool.hpp>#include <uni20/backend/cuda/runtime.hpp>#include <cusolverDn.h>#include <cstddef>#include <optional>Go to the source code of this file.
Classes | |
| class | uni20::cusolver::ExecutionLease |
| Move-only pairing of one cuSOLVER handle slot and one idle stream. More... | |
| class | uni20::cusolver::ExecutionPool |
| Device-local cuSOLVER handle pool paired with the shared stream pool. More... | |
| class | uni20::cusolver::HandleSlot |
| One reusable cuSOLVER handle and its handle-local state. More... | |
Namespaces | |
| namespace | uni20 |
| namespace | uni20::cusolver |
Functions | |
| ExecutionPool & | uni20::cusolver::execution_pool (cuda::DeviceResources &resources) |
| Return the lazily constructed cuSOLVER execution pool for one device. | |
| ExecutionPool & | uni20::cusolver::execution_pool (cuda::DeviceResources &resources, std::size_t handle_count) |
| Configure or return the canonical cuSOLVER execution pool for one device. | |
cuSOLVER handle slots and dynamically paired stream execution leases.