Kernel Tuning Toolkit 2.3
 
Loading...
Searching...
No Matches
ktt::KernelCompilationData Struct Reference

#include <KernelCompilationData.h>

Public Member Functions

 KernelCompilationData ()
 

Public Attributes

uint64_t m_MaxWorkGroupSize
 
uint64_t m_LocalMemorySize
 
uint64_t m_PrivateMemorySize
 
uint64_t m_ConstantMemorySize
 
uint64_t m_RegistersCount
 

Detailed Description

Structure which holds compilation information about specific kernel configuration.

Constructor & Destructor Documentation

◆ KernelCompilationData()

ktt::KernelCompilationData::KernelCompilationData ( )

Constructor which initializes all data values to zero.

Member Data Documentation

◆ m_ConstantMemorySize

uint64_t ktt::KernelCompilationData::m_ConstantMemorySize

The size in bytes of user-allocated constant memory required by the kernel. This value is valid only for CUDA backend.

◆ m_LocalMemorySize

uint64_t ktt::KernelCompilationData::m_LocalMemorySize

The size in bytes of statically-allocated local memory (shared memory in CUDA) required by the kernel.

◆ m_MaxWorkGroupSize

uint64_t ktt::KernelCompilationData::m_MaxWorkGroupSize

The maximum size of a work-group (thread block in CUDA), beyond which a launch of the kernel would fail. Depends on both kernel and device.

◆ m_PrivateMemorySize

uint64_t ktt::KernelCompilationData::m_PrivateMemorySize

The size in bytes of private memory (local memory in CUDA) used by each work-item of the kernel.

◆ m_RegistersCount

uint64_t ktt::KernelCompilationData::m_RegistersCount

The number of registers used by each work-item of the kernel. This value is valid only for CUDA backend.


The documentation for this struct was generated from the following file: