cuda-0.7.0.0: FFI binding to the CUDA interface for programming NVIDIA GPUs

Copyright[2009..2014] Trevor L. McDonell
LicenseBSD
Safe HaskellNone
LanguageHaskell98

Foreign.CUDA.Driver.Module.Link

Contents

Description

Module linking for low-level driver interface

Since CUDA-5.5

Synopsis

JIT module linking

data LinkState Source

A pending JIT linker state

Instances

data JITOption Source

Just-in-time compilation and linking options

Constructors

MaxRegisters !Int

maximum number of registers per thread

ThreadsPerBlock !Int

number of threads per block to target for

OptimisationLevel !Int

level of optimisation to apply (1-4, default 4)

Target !Compute

compilation target, otherwise determined from context

FallbackStrategy !JITFallback

fallback strategy if matching cubin not found

GenerateDebugInfo

generate debug info (-g) (requires cuda >= 5.5)

GenerateLineInfo

generate line number information (-lineinfo) (requires cuda >= 5.5)

Verbose

verbose log messages (requires cuda >= 5.5)

Instances

create :: [JITOption] -> IO LinkState Source

Create a pending JIT linker invocation. The returned LinkState should be destroyed once no longer needed. The device code machine size will match the calling application.

Requires CUDA-5.5.

http://docs.nvidia.com/cuda/cuda-driver-api/group__CUDA__MODULE.html#group__CUDA__MODULE_1g86ca4052a2fab369cb943523908aa80d

complete :: LinkState -> IO Module Source

Complete a pending linker invocation and load the current module. The link state will be destroyed.

Requires CUDA-5.5.

http://docs.nvidia.com/cuda/cuda-driver-api/group__CUDA__MODULE.html#group__CUDA__MODULE_1g818fcd84a4150a997c0bba76fef4e716

addDataFromPtr :: LinkState -> Int -> Ptr Word8 -> JITInputType -> [JITOption] -> IO () Source

As addData, but read the specified number of bytes of image data from the given pointer.