A container to store requested and minimum neccessary resource requests across all resource types currently supported by ExecutorResourceMgr/ExecutorResourcePool. It also includes a ChunkRequestInfo struct to denote which DataMgr chunks (with their sizes in bytes) are neccesary for the query. More...

#include <ResourceRequest.h>

Collaboration diagram for ExecutorResourceMgr_Namespace::RequestInfo:

Public Member Functions
	RequestInfo (const ExecutorDeviceType request_device_type, const size_t priority_level, const size_t cpu_slots, const size_t min_cpu_slots, const size_t gpu_slots, const size_t min_gpu_slots, const size_t cpu_result_mem, const size_t min_cpu_result_mem, const ChunkRequestInfo &chunk_request_info, const bool output_buffers_reusable_intra_thread)

	RequestInfo (const ExecutorDeviceType request_device_type, const size_t num_kernels, const size_t cpu_result_mem)
	Simple constructor assuming no difference between min and requested resources, and no intra-thread cpu mem sharing. More...

Public Attributes
ExecutorDeviceType	request_device_type

size_t	priority_level {0}

size_t	cpu_slots {0}

size_t	min_cpu_slots {0}

size_t	gpu_slots {0}

size_t	min_gpu_slots {0}

size_t	cpu_result_mem {0}

size_t	min_cpu_result_mem {0}

ChunkRequestInfo	chunk_request_info

bool	output_buffers_reusable_intra_thread {false}

bool	chunk_memory_scales_by_num_threads {false}

bool	request_must_run_alone {false}

bool	request_must_run_alone_for_device_type {false}

Detailed Description

A container to store requested and minimum neccessary resource requests across all resource types currently supported by ExecutorResourceMgr/ExecutorResourcePool. It also includes a ChunkRequestInfo struct to denote which DataMgr chunks (with their sizes in bytes) are neccesary for the query.

RequestInfo is the principal data interface between Executor::launchKernelsViaResourceMgr and ExecutorResourceMgr.

Definition at line 70 of file ResourceRequest.h.

Constructor & Destructor Documentation

ExecutorResourceMgr_Namespace::RequestInfo::RequestInfo	(	const ExecutorDeviceType	request_device_type,
		const size_t	priority_level,
		const size_t	cpu_slots,
		const size_t	min_cpu_slots,
		const size_t	gpu_slots,
		const size_t	min_gpu_slots,
		const size_t	cpu_result_mem,
		const size_t	min_cpu_result_mem,
		const ChunkRequestInfo &	chunk_request_info,
		const bool	output_buffers_reusable_intra_thread
	)

inline

Definition at line 85 of file ResourceRequest.h.

       : request_device_type(request_device_type)
       , priority_level(priority_level)
       , cpu_slots(cpu_slots)
       , min_cpu_slots(min_cpu_slots)
       , gpu_slots(gpu_slots)
       , min_gpu_slots(min_gpu_slots)
       , cpu_result_mem(cpu_result_mem)
       , min_cpu_result_mem(min_cpu_result_mem)
       , chunk_request_info(chunk_request_info)
       , output_buffers_reusable_intra_thread(output_buffers_reusable_intra_thread) {}

ExecutorResourceMgr_Namespace::RequestInfo::RequestInfo	(	const ExecutorDeviceType	request_device_type,
		const size_t	num_kernels,
		const size_t	cpu_result_mem
	)

inline

Simple constructor assuming no difference between min and requested resources, and no intra-thread cpu mem sharing.

Definition at line 110 of file ResourceRequest.h.

References GPU.

       : request_device_type(request_device_type)
       , priority_level(static_cast<size_t>(0))
       , cpu_slots(num_kernels)
       , min_cpu_slots(num_kernels)
       , gpu_slots(request_device_type == ExecutorDeviceType::GPU ? num_kernels
                                                                  : static_cast<size_t>(0))
       , min_gpu_slots(request_device_type == ExecutorDeviceType::GPU
                           ? num_kernels
                           : static_cast<size_t>(0))
       , cpu_result_mem(cpu_result_mem)
       , min_cpu_result_mem(cpu_result_mem) {}