◆ topology
◆ cpu_depth
int _starpu_machine_config::cpu_depth |
◆ pu_depth
int _starpu_machine_config::pu_depth |
◆ current_bindid
int _starpu_machine_config::current_bindid |
Where to bind next worker ?
◆ currently_bound
char _starpu_machine_config::currently_bound[STARPU_NMAXWORKERS] |
◆ currently_shared
char _starpu_machine_config::currently_shared[STARPU_NMAXWORKERS] |
◆ current_cuda_gpuid
int _starpu_machine_config::current_cuda_gpuid |
Which GPU(s) do we use for CUDA ?
◆ current_opencl_gpuid
int _starpu_machine_config::current_opencl_gpuid |
Which GPU(s) do we use for OpenCL ?
◆ current_mic_deviceid
int _starpu_machine_config::current_mic_deviceid |
◆ current_mpi_deviceid
int _starpu_machine_config::current_mpi_deviceid |
◆ cpus_nodeid
int _starpu_machine_config::cpus_nodeid |
Memory node for cpus, if only one
◆ cuda_nodeid
int _starpu_machine_config::cuda_nodeid |
Memory node for CUDA, if only one
◆ opencl_nodeid
int _starpu_machine_config::opencl_nodeid |
Memory node for OpenCL, if only one
◆ mic_nodeid
int _starpu_machine_config::mic_nodeid |
Memory node for MIC, if only one
◆ mpi_nodeid
int _starpu_machine_config::mpi_nodeid |
Memory node for MPI, if only one
◆ padding1
char _starpu_machine_config::padding1[STARPU_CACHELINE_SIZE] |
Separate out previous variables from per-worker data.
◆ workers
Basic workers : each of this worker is running its own driver and can be combined with other basic workers.
◆ combined_workers
Combined workers: these worker are a combination of basic workers that can run parallel tasks together.
◆ submitted_mutex
starpu_pthread_mutex_t _starpu_machine_config::submitted_mutex |
◆ padding2
char _starpu_machine_config::padding2[STARPU_CACHELINE_SIZE] |
Separate out previous mutex from the rest of the data.
struct { ... } * _starpu_machine_config::bindid_workers |
Translation table from bindid to worker IDs
◆ nbindid
unsigned _starpu_machine_config::nbindid |
◆ worker_mask
uint32_t _starpu_machine_config::worker_mask |
This bitmask indicates which kinds of worker are available. For instance it is possible to test if there is a CUDA worker with the result of (worker_mask & STARPU_CUDA).
◆ conf
struct starpu_conf _starpu_machine_config::conf |
either the user given configuration passed to starpu_init or a default configuration
◆ running
unsigned _starpu_machine_config::running |
this flag is set until the runtime is stopped
◆ disable_kernels
int _starpu_machine_config::disable_kernels |
◆ pause_depth
int _starpu_machine_config::pause_depth |
Number of calls to starpu_pause() - calls to starpu_resume(). When >0, StarPU should pause.
◆ sched_ctxs
all the sched ctx of the current instance of starpu
◆ submitting
unsigned _starpu_machine_config::submitting |
this flag is set until the application is finished submitting tasks
◆ watchdog_ok
int _starpu_machine_config::watchdog_ok |
The documentation for this struct was generated from the following file: