CUDA Fortran

Programming Guide and Reference

Release 2011

The Portland Group

While every precaution has been taken in the preparation of this document, The Portland Group® (PGI®), a wholly-owned subsidiary of STMicroelectronics, Inc., makes no warranty for the use of its products and assumes no responsibility for any errors that may appear, or for damages resulting from the use of the information contained herein. The Portland Group retains the right to make changes to this information at any time, without notice. The software described in this document is distributed under license from STMicroelectronics and/or The Portland Group and may be used or copied only in accordance with the terms of the license agreement ("EULA"). PGI Workstation, PGI Server, PGI Accelerator, PGF95, PGF90, PGFORTRAN, and PGI Unified Binary are trademarks; and PGI, PGHPF, PGF77, PGCC, PGC++, PGI Visual Fortran, PVF, PGI CDK, Cluster Development Kit, PGPROF, PGDBG, and The Portland Group are registered trademarks of The Portland Group Incorporated. No part of this document may be reproduced or transmitted in any form or by any means, for any purpose other than the purchaser's or the end user's personal use without the express written permission of STMicroelectronics and/or The Portland Group.

Copyright © 2010 The Portland Group® and STMicroelectronics, Inc.All rights reserved. Printed in the United States of America Release 2010, version 10.8, August 2010 Release 2010, version 10.9, September 2010 Release 2011, version 11.0, November 2010

Technical support: trs@pgroup.com Sales: sales@pgroup.com Web: www.pgroup.com

ID: 103151117

Contents
Preface ...................................................................................................................................... ix
Intended Audience ................................................................................................................. ix Organization .......................................................................................................................... ix Conventions ........................................................................................................................... ix Terminology ........................................................................................................................... x Related Publications ................................................................................................................ x

1. Introduction ........................................................................................................................ 1 2. Programming Guide .......................................................................................................... 3
CUDA Fortran Kernels .............................................................................................................. 3 Thread Blocks ........................................................................................................................ 4 Memory Hierarchy .................................................................................................................. 4 Subroutine / Function Qualifiers ................................................................................................ 4 Attributes(host) .............................................................................................................. 5 Attributes(global) ............................................................................................................ 5 Attributes(device) ........................................................................................................... 5 Restrictions .................................................................................................................... 5 Variable Qualifiers .................................................................................................................. 5 Attributes(device) ........................................................................................................... 5 Attributes(constant) ........................................................................................................ 6 Attributes(shared) ........................................................................................................... 6 Attributes(pinned) .......................................................................................................... 6 Datatypes in Device Subprograms ............................................................................................. 6 Predefined Variables in Device Subprograms .............................................................................. 7 Execution Configuration ........................................................................................................... 7 Asynchronous concurrent execution .......................................................................................... 8 Concurrent Host and Device Execution .............................................................................. 8 Concurrent Stream Execution ........................................................................................... 8 Kernel Loop Directive .............................................................................................................. 8 Restrictions on the CUF kernel directive ........................................................................... 10 Building a CUDA Fortran Program ........................................................................................... 11 Emulation Mode ................................................................................................................... 11 iii

3. Reference ........................................................................................................................... 13
New Subroutine and Function Attributes ................................................................................... Host Subroutines and Functions ...................................................................................... Global Subroutines ........................................................................................................ Device Subroutines and Functions ................................................................................... Restrictions on Device Subprograms ................................................................................ Variable Attributes ................................................................................................................. Device data .................................................................................................................. Constant data ................................................................................................................ Shared data .................................................................................................................. Value dummy arguments ................................................................................................ Pinned arrays ............................................................................................................... Allocating Device and Pinned Arrays ........................................................................................ Allocating Device Memory .............................................................................................. Allocating Device Memory Using Runtime Routines ............................................................ Allocating Pinned Memory .............................................................................................. Data transfer between host and device memory ......................................................................... Data Transfer Using Assignment Statements ....................................................................... Implicit Data Transfer in Expressions ............................................................................... Data Transfer Using Runtime Routines ............................................................................. Invoking a kernel subroutine .................................................................................................. Device code ......................................................................................................................... Datatypes allowed ......................................................................................................... Builtin variables ............................................................................................................ Fortran intrinsics .......................................................................................................... New Intrinsic Functions ................................................................................................. Warp-Vote Operations .................................................................................................... Atomic Functions .......................................................................................................... Restrictions .................................................................................................................. Host code ............................................................................................................................ SIZEOF Intrinsic ............................................................................................................ 13 13 13 14 14 14 14 15 16 17 17 17 17 18 18 19 19 19 20 20 21 21 22 22 24 26 27 29 29 29

4. Runtime APIs .................................................................................................................... 31
Initialization ......................................................................................................................... Device Management ............................................................................................................... cudaGetDeviceCount ...................................................................................................... cudaSetDevice ............................................................................................................... cudaSetDeviceFlags ........................................................................................................ cudaGetDevice .............................................................................................................. cudaGetDeviceProperties ................................................................................................ cudaChooseDevice ......................................................................................................... Thread Management .............................................................................................................. cudaThreadSynchronize ................................................................................................. cudaThreadExit ............................................................................................................. Memory Management ............................................................................................................ cudaMalloc .................................................................................................................. iv 31 31 31 31 31 32 32 32 32 32 32 32 33

cudaMallocPitch ............................................................................................................ cudaFree ...................................................................................................................... cudaMallocArray ........................................................................................................... cudaFreeArray .............................................................................................................. cudaMemset ................................................................................................................. cudaMemset2D ............................................................................................................. cudaMemcpy ................................................................................................................ cudaMemcpyAsync ........................................................................................................ cudaMemcpy2D ............................................................................................................ cudaMemcpy2DAsync .................................................................................................... cudaMemcpyToArray ...................................................................................................... cudaMemcpy2DToArray .................................................................................................. cudaMemcpyFromArray .................................................................................................. cudaMemcpy2DFromArray .............................................................................................. cudaMemcpyArrayToArray .............................................................................................. cudaMemcpy2DArrayToArray .......................................................................................... cudaMalloc3D .............................................................................................................. cudaMalloc3DArray ....................................................................................................... cudaMemset3D ............................................................................................................. cudaMemcpy3D ............................................................................................................ cudaMemcpy3DAsync .................................................................................................... cudaMemcpyToSymbol ................................................................................................... cudaMemcpyFromSymbol ............................................................................................... cudaMemcpyToSymbolAsync ........................................................................................... cudaMemcpyFromSymbolAsync ....................................................................................... cudaGetSymbolAddress ................................................................................................... cudaGetSymbolSize ........................................................................................................ cudaHostAlloc ............................................................................................................... cudaHostGetDevicePointer .............................................................................................. cudaMallocHost ............................................................................................................ cudaFreeHost ................................................................................................................ Stream Management .............................................................................................................. cudaStreamCreate .......................................................................................................... cudaStreamQuery .......................................................................................................... cudaStreamSynchronize .................................................................................................. cudaStreamDestroy ........................................................................................................ Event Management ................................................................................................................ cudaEventCreate ............................................................................................................ cudaEventRecord .......................................................................................................... cudaEventQuery ............................................................................................................ cudaEventSynchronize .................................................................................................... cudaEventDestroy .......................................................................................................... cudaEventElapsedTime ................................................................................................... Error Handling ..................................................................................................................... cudaGetLastError .......................................................................................................... cudaGetErrorString ........................................................................................................

33 33 33 33 34 34 34 34 34 35 35 35 35 35 36 36 36 36 36 36 37 37 37 37 37 38 38 38 38 39 39 39 39 39 39 39 40 40 40 40 40 40 40 41 41 41 v

.............................................................................................................................................................................. 47 vi ................................... 41 cudaRuntimeGetVersion ............ 43 Matrix Multiplication Example .............................................................Version Management ........................................................................................ 43 Source Code Listing ........................................................................................................................................................................ 41 cudaDriverGetVersion ..................................................................... 43 Source Code Description ........................................................................................ Contact Information ................................................................................................ Examples .......... 45 6................ 41 5...............

.................................................... 6 3........................................ 23 3....... Fortran Numeric and Logical Intrinsics ......3....................................... 22 3...........................................................................8................ Fortran Random Number Intrinsics ...................... 21 3................. 23 3...................................5..............................................10..................................................................................................7............... 28 3....................... Counting Atomic Functions ................... 27 3...2..................9......................................................... Device Code Intrinsic Datatypes ................................................................................................ Fortran Reduction Intrinsics ...6.................... Fortran Numeric Inquiry Intrinsics ...............................1............................................. 23 3......................Tables 2.............................................................................. 24 3........ 29 vii ............. 24 3.....1.................................................................... Fortran Mathematical Intrinsics ........ Fortran Bit Manipulation Intrinsics ........................ Arithmetic and Bitwise Atomic Functions ................... Compare and Swap Atomic Function .4..................................... Intrinsic Datatypes ..

viii .

Intended Audience This guide is intended for application programmers. C. scientists and engineers proficient in programming with the Fortran.Preface This document describes CUDA Fortran. a small set of extensions to Fortran that supports and is built upon the CUDA computing architecture. “Reference” describes the CUDA Fortran language reference Chapter 4. and Intel 64 hardware platforms. “Runtime APIs” describes the interface between CUDA Fortran and the CUDA Runtime API Chapter 5. Conventions This guide uses the following conventions: italic is used for emphasis. The PGI tools are available on a variety of operating systems for the X86. “Introduction” contains a general introduction Chapter 2. “Examples” provides sample code and an explanation of the simple example. “Programming Guide” serves as a programming guide for CUDA Fortran Chapter 3. AMD64. ix . Organization The organization of this document is as follows: Chapter 1. This guide assumes familiarity with basic operating system usage. and/or C++ languages.

• ISO/IEC 1539-1:1997. x . Version 3.. • NVIDIA CUDA Compute Unified Device Architecture Reference Manual. multiple filenames are allowed. June 2010.1. and for language statements in the text.pdf. In the context of p/tsets. There are a wide variety of releases and distributions of each of these types of operating systems. { item2 | item 3} braces indicate that a selection is required. NVIDIA. 1997 (Fortran 95).1. square brackets indicate optional items. The Portland Group.nvidia. The PGI compilers and tools are supported on both 32-bit and 64-bit variants of the Linux and Windows operating systems on a variety of x86-compatible processors.pgroup.htm Related Publications The following documents contain additional information related to CUDA Fortran programming.com/cuda. [ item1 ] in general. FORTRAN Fortran language statements are shown in the text of this guide using a reduced fixed point size. 2009.com/support/definitions. Geneva. examples. PGI provides a glossary of terms which you can access at www. Zero or more of the preceding item may occur.com/cuda. In this example. In this case item1 is optional. September. • NVIDIA CUDA Programming Guide.1. ellipsis indicate a repetition. arguments. NVIDIA.pgroup. including assembly language statements. C/C++ C/C++ language statements are shown in the test of this guide using a reduced fixed point size. Information Technology – Programming Languages – FORTRAN.Terminology Constant Width is used for filenames. Version 3. 7/21/2010.nvidia. Available online at http://www. Release 10. filename . square brackets are required to specify a p/t-set. Bold is used for commands. directories. • PGI Users Guide. Terminology If there are terms in this guide with which you are unfamiliar. Available online at http:// www. In this case. Available online at http:// www.9.com/doc/pgiug. you must select either item2 or item3.. options.

Compared to the PGI Accelerator directives-based model and compilers. here called CUDA C. CUDA Fortran includes a Fortran 2003 compiler and tool chain for programming NVIDIA GPUs using Fortran. Graphic processing units or GPUs have evolved into programmable. The CUDA Fortran extensions described in this document allow the following operations in a Fortran program: • Declaring variables that are allocated in the GPU device memory • Allocating dynamic memory in the GPU device memory • Copying data from the host memory to the GPU memory. highly parallel computational units with very high memory bandwidth. A properly designed CUDA program will run on any CUDA-enabled GPU. Mac OS X and Windows. and coordinated independent thread groups organized into a grid. A CUDA programmer must partition the program into coarse grain blocks that can be executed in parallel. but are general enough to be useful in many data-parallel. Introduction Welcome to Release 2011 of PGI CUDA Fortran. The programming model supports four key abstractions: cooperating threads organized into thread groups. shared memory and barrier synchronization within thread groups. and back • Writing subroutines and functions to execute on the GPU • Invoking GPU subroutines from the host • Allocating pinned memory on the host • Using asynchronous transfers between the host and GPU 1 . Each block is partitioned into fine grain threads. and tremendous potential for many applications. with compilers and libraries to support the programming of NVIDIA GPUs. GPU designs are optimized for the computations found in graphics rendering. a general purpose parallel programming architecture. PGI 2011 includes support for CUDA Fortran on Linux. CUDA Fortran is an analog to NVIDIA's CUDA C compiler. NVIDIA introduced CUDA™. compute-intensive programs. CUDA Fortran is a lower-level explicit programming model with substantial runtime library components that give expert programmers direct control of all aspects of GPGPU programming. CUDA comes with an extended C compiler. allowing direct programming of the GPU from a high level language.Chapter 1. which can cooperate using shared memory and barrier synchronization. regardless of the number of available processor cores. a small set of extensions to Fortran that supports and is built upon the CUDA computing architecture.

2 .

Programming Guide This chapter introduces the CUDA programming model through examples written in CUDA Fortran. y real :: a integer :: n ! call the kernel call ksaxpy<<<n/64. each thread performs one iteration of the common SAXPY loop operation. “Reference. 64>>>( n. x. y ) real. x. value :: a integer. 3 . each instance will be executed by a different CUDA thread. A kernel is defined using the attributes(global) specifier on the subroutine statement. each with 64 threads. a. a kernel is called using special chevron syntax to specify the number of thread blocks and threads within each thread block: ! Kernel definition attributes(global) subroutine ksaxpy( n. x. and each thread has a global thread block index. Each thread is assigned a thread block index accessed through the built-in blockidx variable.y real.” on page 13. a. and a thread index accessed through threadidx.Chapter 2. Such a subroutine is called a device kernel or kernel. The CUDA threads are organized into thread blocks. y ) end subroutine In this case. and a local thread index within its thread block. refer to Chapter 3. i i = (blockidx%x-1) * blockdim%x + threadidx%x if( i <= n ) y(i) = a * x(i) + y(i) end subroutine ! Host subroutine subroutine solve( n. A call to a kernel specifies how many parallel instances of the kernel must be executed. dimension(*) :: x. y ) real. device. dimension(*) :: x. CUDA Fortran Kernels CUDA Fortran allows the definition of Fortran subroutines that execute in parallel on the GPU when called from the Fortran program which has been invoked and is running on the host. a. In this example. value :: n. the call to the kernel ksaxpy specifies n/64 thread blocks. For a reference for CUDA Fortran.

the thread ID is (x+Dx(y-1)+Dy(z-1)). Each thread can also have private local memory. It can also directly copy data to and from the device global memory. data in global device memory can be read or written by all threads. designating whether it is executed on the host or on the device. or with the host attribute by default. the first argument in the chevron <<<>>> syntax is the grid size. is called a host subprogram. The shared memory acts like a low-latency. the maximum number of threads in a thread block is 512.Dy. For a one-dimensional thread block. On the host side. and the second argument is the thread block size. so are slow relative to the host memory. and a thread index accessed through threadidx. or three-dimensional index. data in shared memory has a lifetime of the thread block.Dy). The host can also set the values in the device constant memory.Thread Blocks Thread Blocks Each thread is assigned a thread block index accessed through the built-in blockidx variable. and a block index within the grid. but it is read-only to the threads and limited in size. or called from another device subprogram. Currently. In CUDA Fortran. Data in constant memory space is initialized by the host program. Accesses to constant memory are typically faster than accesses to global memory. whether it is a kernel. the host program can directly access data in the host main memory. On the device side. Thread blocks must be able to execute independently. Each thread in the block waits at the call to SYNCTHREADS() until all threads have reached that call. the thread index for each dimension starts at one. For a two-dimensional thread block of size (Dx. • A subprogram declared with attributes(host). and by synchronizing at a barrier using the SYNCTHREADS() intrinsic. Threads in the same thread block can access and share data in shared memory. two-. and if the latter.or two-dimensional grid of blocks. A kernel may be invoked with many thread blocks. computed from the thread index. The thread index may be a one-. by the same core or by different cores. all threads can read data in constant memory. For a three-dimensional thread block of size (Dx. The thread blocks are organized into a one. • A subprogram declared with attributes(global) or attributes(device) is called a device subprogram. 4 . the thread index is equal to the thread ID. Memory Hierarchy CUDA Fortran programs have access to several memory spaces. again implemented using DMA access. the thread ID is equal to (x+Dx(y-1)). so each thread has a thread index within the block. data in thread local memory may be implemented as processor registers or may be allocated in the global device memory. called from the host. Threads in the same thread block may cooperate by using shared memory. two thread blocks may be executed in parallel or one after the other. each with the same thread block size. A unique thread ID is assigned to each thread. best performance will often be obtained when thread local data Subroutine / Function Qualifiers A subroutine or function in CUDA Fortran has an additional attribute. such data copies require DMA access to the device. high bandwidth software managed cache memory. When invoking a kernel.Dz).

and may only be called from the host using a kernel call containing the chevron syntax and runtime mapping parameters. • Calls to a kernel subroutine must specify the execution configuration. Such a call is asynchronous. • A device subprogram must not contain variables with the SAVE attribute. • A kernel subroutine may not also have the device or host attribute. to be executed on the device. the variable may be accessed by any subprogram in that module and by any subprogram that uses the module. such a routine must be called from a subprogram with the global or device attribute. or with data initialization. constant.” on page 7. and may not contain any subroutines or functions. if none of the host. • Device subprograms may not be contained in a host subroutine or function. By default. as described in “Predefined Variables in Device Subprograms. it declares that the subroutine is a kernel subroutine. or device attributes is specified. Attributes(device) The device attribute. 5 . specified on the subroutine or function statement. or pinned attributes may be specified for a variable. specified on the subroutine or function statement. the host routine making the call continues to execute before the device has completed its execution of the kernel subroutine. The default is attributes(host). Programming Guide • A subroutine declared with attributes(global) is also called a kernel subroutine. variables declared in modules or host subprograms are allocated in the host main memory. Dummy arguments in a device subprogram must not have the pointer attribute. and is allocated in the device global memory. Attributes(global) The global attribute may only be specified on a subroutine statement. shared. Attributes(device) A variable with the device attribute is called a device variable. Variable Qualifiers Variables in CUDA Fortran have a new attribute. At most one of the device. that is. • A device subprogram must not have optional arguments. Such a subprogram can only be called from another host subprogram. which declares in which memory the data is allocated. Restrictions The following restrictions apply to subprograms: • A device subprogram must not be recursive.Chapter 2. Attributes(host) The host attribute. global. declares that the subprogram is to be executed on the device. • If declared in a module. declares that the subroutine or function is to be executed on the host.

the variable will be allocated in the normal host paged memory. Device constant variables may not be allocatable.8 equivalent to real(kind=8) 6 .8 4.4. A device array may be an explicit-shape array. it will be allocated in host pagelocked memory. Device constant data may not be assigned or modified in any device subprogram.Datatypes in Device Subprograms • If declared in a host subprogram. Device constant variables are allocated in the device constant memory space. or an assumed-shape dummy array. the variable may be accessed by that subprogram or subprograms contained in that subprogram. Attributes(constant) A variable with the constant attributes is called a device constant variable. It can be read or written by all threads in the block. the variable may be accessed by any subprogram in that module and by any subprogram that uses the module.1. or size of page-locked memory. from when it is allocated until it is deallocated. Some operating systems or installations may restrict the use. Datatypes in Device Subprograms The following intrinsic datatypes are allowed in device subprograms and device data: Table 2. and has a lifetime of the thread block. Other device variables have a lifetime of the entire application. When a pinned variable is allocated. or by other device subprograms to which it is passed as an argument. If declared in a module. Intrinsic Datatypes Type integer logical real double precision Type Kind 1. but may be modified in host subprograms. and may only be accessed within that subprogram.8 1. A shared variable is allocated in the device shared memory for a thread block. Attributes(shared) A variable with the shared attributed is called a device shared variable or a shared variable.2. an allocatable array. if the allocation in page-locked memory fails. A shared variable may not be data initialized. An allocatable device variable has a dynamic lifetime. A shared variable may only be declared in a device subprogram. and have a lifetime of the entire application.4.2. A pinned variable must be an allocatable array. Attributes(pinned) A variable with the pinned attributes is called a pinned variable. availability. The advantage of using pinned variables is that copies from page-locked memory to device memory are faster than copies from normal paged host memory. though a write in one thread is only guaranteed to be visible to other threads after the next call to the SYNCTHREADS() intrinsic.

as with threadidx. • If grid is type(dim3). in addition to the statically allocated shared memory. the value of each component must be equal to or greater than one. or are type(dim3). A kernel subroutine call looks like this: call kernel<<<grid. These variables are of type dim3. and the product of the component values must be less than or equal to 512. except for warpsize. This memory is used for the 7 .block[.…) where • grid and block are either integer expressions (for one-dimensional grids and thread blocks).Chapter 2. • The variable blockidx contains the block index within the grid. • The variable blockdim contains the dimensions of the thread block. for one-dimensional grids. • The value of bytes must be an integer. • The variable griddim contains the dimensions of the grid.or two-dimensional grids and one-. for one. or three-dimensional thread blocks. for one. are not accessible in host subprograms. two-.or two-dimensional thread blocks. the threadidx%y and/or threadidx%z components have the value one. Threads are executed in groups of 32.y. • If block is type(dim3). and block%x and block%y must be equal to or greater than one. It may also specify a dynamic shared memory extent. to support concurrent stream execution on the device. blockdim has the same value for all thread blocks in the same grid.arg2. blockidx%y has the value one. called warps. in bytes. the value of grid%z must be one. The execution configuration defines the dimensionality and extent of the grid and thread blocks that execute the subroutine. The value of blockidx%z is always one. it specifies the number of bytes of shared memory to be allocated for each thread block. Predefined Variables in Device Subprograms Device subprograms have access to block and grid indices and dimensions through several builtin read-only variables. warpsize contains the number of threads in a warp.bytes[. Programming Guide complex character(len=1) 4. and a stream identifier.streamid]]>>>(arg1. the value of griddim%z is always one. Execution Configuration A call to a kernel subroutine must specify an execution configuration.z end type These predefined variables. • The variable is warpsize is declared to be type integer. • The variable threadidx contains the thread index within its thread block.8 1 Derived types may contain members with these intrinsic datatypes or other allowed derived types. the module cudafor defines the derived type dim3 as follows: type(dim3) integer(kind=4) :: x.

Launch configuration and mapping of the loop iterations onto the hardware is controlled and specified as part of the directive body using the familiar CUDA chevron syntax. where the launch actually corresponds to placing the kernel on a queue for execution by the device. including calling or queueing more kernels for execution on the device. The default stream. it is the programmer's responsibility to ensure that parallel execution is legal and produces the correct answer.Asynchronous concurrent execution assumed-size shared variables in the thread block. operations on different stream queues may execute out-of-order with respect to when they were placed on the queues. For these operations. and no subsequent operations on any queue begin until the stream zero operation is complete. the host program should synchronize with the device to avoid erroneous results. Concurrent Stream Execution Operations involving the device. In this way. it specifies the stream to which this call is associated. the host can continue executing.2. Kernel Loop Directive CUDA Fortran allows automatic kernel generation and invocation from a region of host code containing one or more tightly nested loops. An application can manage more concurrency by using multiple streams. its value is treated as zero. The work in the loops specified by the directive is executed in parallel. As with any kernel. Each user-created stream manages its own queue. are implemented using stream queues. and will only be initiated when all previous operations on that queue have been completed. across the thread blocks and grid. the host program can synchronize and wait for all previously launched or queued kernels. see Section 3.3. An operation is placed at the end of the stream queue. the launch is asynchronous. the call actually returns to the host program before the kernel subroutine begins execution. The program can use cudaThreadSynchronize() or CUDA Events to wait for the completion of the kernel. The call can be treated as a kernel launch operation. including kernel execution and data copies to and from device memory. and may execute concurrently with each other. inserting synchronization into the kernel as appropriate. used when no stream identifier is specified. is stream zero. such as summing the values in a vector or matrix. By calling the runtime routine cudaThreadSynchronize. Concurrent Host and Device Execution When a host subprogram calls a kernel subroutine. Asynchronous concurrent execution There are two components to asynchronous concurrent execution with CUDA Fortran. 8 . the compiler handles the generation of the final reduction kernel. Programmers must be careful when using concurrent host and device execution. • The value of streamid must be an integer greater than or equal to zero. stream zero is special in that operations on the stream zero queue will begin only after all preceding operations on all queues are complete. The one exception to this rule is a scalar reduction operation. in cases where the host program reads or modifies device or constant data. If not specified.

Alternatively. n a(i. m do i = 1. n a(i. is computed by the compiler and runtime by dividing the loop trip counts n and m by the thread block size.j) +c(i. so all iterations are computed. using asterisks tells the compiler to choose a thread block shape and/or compute the grid shape from the thread block shape and the loop limits. m do i = 1. !$cuf kernel do(2) <<< *. The body of the doubly-nested loop is turned in the kernel body: ThreadIdx%x runs from 1 to 32 and is mapped onto the inner i loop. m do i = 1.*). the schedule applies just to the outermost loop. Alternatively.j) = b(i.j) = b(i.j) = b(i. ThreadIdx%y runs from 1 to 4 and is mapped onto the outer j loop. Example 2 !$cuf kernel do <<< *.j) end do end do 9 .j) +c(i.Chapter 2. 256 >>> do j = 1. The grid and block values can be an integer scalar or a parenthesized list. only the outer j loop is run in parallel with a thread block size of 256. Example 1 !$cuf kernel do(2) <<< (*. that is.1) >>> do j = 1. (256. specified as (*. the directive defines a two-dimensional thread block of size 32x4. You might consider if the code in Example 2 would perform better if the two loops were interchanged. (32.4) >>> do j = 1. block >>> The compiler maps the launch configuration specified by the grid and block values onto the outermost n loops. In this case.j) end do end do In this example. The inner i dimension is run sequentially on each thread.j) end do end do Without an explicit n on the do. Loops which are not mapped onto the grid and block values are run sequentially on each thread. starting at loop n and working out. you could specify a configuration like the following in which the threads read and write the matrices in coalesced fashion. n a(i.*). Programming Guide Syntax The general form of the kernel directive is: !$cuf kernel do[(n)] <<< grid. the default value is 1. The grid shape.j) +c(i.

if they are allowed in device code. specify non-asterisk values for the grid. Specifically. upper limit. n a(i. the 256 threads in each block each do one element of the matrix addition. m do i = 1.Kernel Loop Directive Example 3 In Example 2.j) = b(i. • Subroutine and function calls to attributes(device) subprograms are allowed if they are in the same module as the code containing the directive. as illustrated here: !$cuf kernel do(2) <<< (1. incrementing by 256. are not. and GOTO statements. Summation Example The simplest directive form for performing a dot product on two device arrays takes advantage of the properties for scalar use outlined previously. but the device-specific intrinsics such as syncthreads. Further expansion of the work along the i direction and all work across the j dimension is handled by the mapping onto the grid dimensions.j) end do end do Now the threads in a thread block handle all values in the i direction. the compiler generates a correct sequence of synchronized operations to produce one copy either in device global memory or on the host. • The body of the loops may contain assignment statements.*). • Scalars used or assigned in the loop must either have the device attribute.j) +c(i. • CUDA Fortran intrinsic functions are allowed. • Only CUDA Fortran dataypes are allowed within the loops. it must be followed by at least that many tightly-nested DO loops. loops. one for each thread. Restrictions on the CUF kernel directive The following restrictions apply to CUF kernel directives: • If the directive specifies n dimensions. • There can be no GOTO or EXIT statements within or between any loops that have been mapped onto the grid and block configuration values. To "unroll" more work into each thread. (256. in concert. and increment must be invariant with respect to any other loop in the kernel do. • The tightly-nested DO loops must have invariant loop limits: the lower limit. • Implicit loops and F90 array syntax are not allowed within the directive loops.1) >>> do j = 1. • Arrays used or assigned in the loop must have the device attribute. In CUDA built-in language. etc. the j loop is mapped onto the grid x-dimension. atomic functions. One thread block is created for each j. gridDim%x is equal to m. because the compiler skips over the constant 1 in the i loop grid size. when a reduction has a scalar target. Except in the case of reductions. 10 . IF statements. or the compiler will make a device copy of that variable live for the duration of the loops.

This CUF kernel can be followed by another CUF kernel in the same subprogram: !cuf kernel do <<< *. Building a CUDA Fortran Program CUDA Fortran is supported by the PGI Fortran compilers when the filename uses a CUDA Fortran extension. 11 . allowing the programmer to use a host debugger. To build a program using emulation mode.0 !cuf kernel do <<< *. such as memory races.CUF extension may also be used.cuf extension specifies that the file is a free-format CUDA Fortran program. the device code is compiled for and runs on the host. the compiler recognizes rsum as a scalar temporary which should be allocated locally on every thread. in which case the program is processed by the preprocessor before being compiled. the host floating point units and intrinsics are used. Emulation Mode PGI Fortran compilers support an emulation mode for program development on workstations or systems without a CUDA-enabled GPU and for debugging.Chapter 2. The . In emulation mode. use of rsum on the host following this loop is undefined. compile and link with the –Mcuda=emu command line option. CUDA Fortran extensions can be enabled in any Fortran source file by adding the –Mcuda command line option. which may produce slightly different answers than the device units and intrinsics. emulation mode may execute a single thread block at a time. the compiler recognizes the use of the scalar and generates just one final result. Programming Guide rsum = 0. To compile a fixed-format program. It’s important to note that the emulation is far from exact. This will not expose certain errors. In emulation mode. the . In particular. However. * >>> do i = 1. add the command line option –Mfixed. * >>> do i = 1. n rsum = rsum + x(i) * y(i) end do For reductions. n rsum = x(i) * y(i) z(i) = rsum end do In this CUF kernel.

12 .

A subprogram with no attributes prefix has the host attribute by default. A kernel subroutine may not be contained in another subroutine or function. or elemental. or both. to be called from a host routine using an execution configuration. or may have both host and device attribute. A subroutine with the global attribute is called a kernel subroutine. their meaning and restrictions. A Function may have the host or device attribute. pure. and may not contain any other subprogram. A host subprogram is compiled for execution on the host processor. 13 . This chapter describes how to specify the new attributes. if there is no attributes prefix on the subprogram statement. function return datatype). global. or a host subprogram. Reference This chapter is the CUDA Fortran Language Reference. These attributes are specified using the attributes(attr) prefix on the Subroutine or Function statement. New Subroutine and Function Attributes CUDA Fortran adds new attributes to subroutines and functions.Chapter 3. Host Subroutines and Functions The host attribute may be explicitly specified on the Subroutine or Function statement as follows: attributes(host) subroutine sub(…) attributes(host) integer function func(…) integer attributes(host) function func(…) The host attributes prefix may be preceded or followed by any other allowable subroutine or function prefix specifiers (recursive. A kernel subroutine may not be recursive. then default rules are used. or device attribute. as described in the following sections. pure. Global Subroutines The global attribute may be explicitly specified on the Subroutine statement as follows: attributes(global) subroutine sub(…) Functions may not have the global attribute. so no other subroutine prefixes are allowed. A subroutine or function with the host attribute is called a host subroutine or function. A Subroutine may have the host. A kernel subroutine is compiled as a kernel for execution on the device. elemental.

” on page 21. or elemental. • It may not be pure or elemental. nor have the recursive prefix on the subprogram statement. A subroutine or function with the device attribute must appear within a Fortran module. or in thread local memory.Variable Attributes Device Subroutines and Functions The device attribute may be explicitly specified on the Subroutine or Function statement as follows: attributes(device) subroutine sub(…) attributes(device) datatype function func(…) datatype attributes(device) function func(…) A subroutine or function with the device attribute may not be recursive. pure. • It may not be contained in another subroutine or function. Device data A variable or array with the device attribute is defined to reside in the device global memory. nor have the pure or elemental prefix on the subprogram statement. except for the function return datatype. Variable Attributes CUDA Fortran adds new attributes for variables and arrays. This section describes how to specify the new attributes and their meaning and restriction. in the thread block shared memory. so no other subroutine or function prefixes are allowed. 14 . to be device arrays of size 100. Variables in modules may be declared to be in device global memory or constant memory space. CUDA Fortran adds a new attribute for allocatable arrays in host memory. refer to “Device code. and may only be called from device subprograms in the same module. Variables declared in a device subprogram may have one of four attributes: they may be declared to be in device global memory. in page-locked host memory space. in constant memory space. The device attribute can be specified with the attributes statement. A device subprogram is compiled for execution on the device. The advantage of using pinned memory is that transfers between the device and pinned memory are faster and can be asynchronous. • It may not contain another subprogram. For more information. that is. a and b. the array may be declared to be in pinned memory. The following example declares two arrays. A subroutine or function with the device or kernel attribute is called a device subprogram. or as an attribute on the type declaration statement. Restrictions on Device Subprograms A subroutine or function with the device or global attribute must satisfy the following restrictions: • It may not be recursive.

• Device variables and arrays declared in a host subprogram cannot have the Save attribute. 15 . • Constant variables and arrays may appear in modules. or as an attribute on the type declaration statement. • Device variables and arrays may not have the Pointer or Target attributes. In host subprograms. Reference real :: a(100) attributes(device) :: a real. Constant data A variable or array with the constant attribute is defined to reside in the device constant memory space. • Device variables and arrays may appear in modules. The following example declares two arrays. device data may only be used in the following manner: • In declaration statements • In Allocate and Deallocate statements • As an argument to the Allocated intrinsic function • As the source or destination in a data transfer assignment statement • As an actual argument to a kernel subroutine • As an actual argument to another host subprogram or runtime API call • As a dummy argument in a host subprogram A device array may have the allocatable attribute. but may not be in a Common block or an Equivalence statement. The constant attribute can be specified with the attributes statement. Target.Chapter 3. constant :: d(100) These rules apply to constant data: • Constant data may not have the Pointer. or may have adjustable extent. the subprogram interface must be explicit (in the Fortran sense). • Device variables and arrays may be passed as actual arguments to host and device subprograms. c and d. • Members of a derived type may not have the device attribute unless they are allocatable. but may not be in a Common block or an Equivalence statement. in that case. and the matching dummy argument must also have the device attribute. real :: c(100) attributes(constant) :: c real. device :: b(100) These rules apply to device data: • An allocatable device array dynamically allocates device global memory. to be constant arrays of size 100. or Allocatable attributes.

Shared arrays that are not dummy arguments may be declared as assumed-size arrays. or Allocatable attributes. In host subprograms.Variable Attributes • Members of a derived type may not have the constant attribute. variables and arrays with the constant attribute may be read and written. or as an attribute on the type declaration statement. s and t. to be shared arrays of size 100. Its size is determined at run time by the call to the kernel. This memory is used for the 16 . • Within host subprograms. shared :: x(*) Such an array has special significance. shared :: d(100) These rules apply to shared data: • Shared data may not have the Pointer. • Shared variables and arrays may be passed as actual arguments to from a device subprogram to another device subprogram. The following example declares two arrays. as long as the subprogram interface is explicit. • Members of a derived type may not have the shared attribute. • Within device subprograms. and the matching dummy argument also has the constant attribute. Target. real :: c(100) attributes(shared) :: c real. data with the constant attribute may only be used in the following manner: • In declaration statements • As the source or destination in a data transfer assignment statement • As an actual argument to another host subprogram • As a dummy argument in a host subprogram Shared data A variable or array with the shared attribute is defined to reside in the shared memory space of a thread block. • Arrays with the constant attribute must have fixed size. When the kernel is called. the last dimension of a shared array may have an asterisk as its upper bound: real. as long as the interface is explicit and the matching dummy argument has the shared attribute. The shared attribute can be specified with the attributes statement. A shared variable or array may only be declared and used inside a device subprogram. that is. • Constant variables and arrays may be passed as actual arguments to host and device subprograms. the value of the bytes argument in the execution configuration is used to specify the number of bytes of shared memory that is dynamically allocated for each thread block. variables and arrays with the constant attribute may not be assigned or modified. • Shared variables may not be in a Common block or Equivalence statement.

If a shared array is not a dummy argument and not assumed-size. to be pinned allocatable arrays. allocatable. Where the array is deallocated. or whether the dummy argument has the pinned and allocatable attributes. by adding the value attribute to the variable declaration. b integer.n) :: a. n ) real. If a device array declared in a host subprogram does not have the Save attribute. they are all implicitly equivalenced. Pinned arrays An allocatable array with the pinned attribute will be allocated in special page-locked host memory. specifically for dynamically allocating device arrays and host pinned arrays. dummy arguments are passed by default by reference. and the address of the argument is passed to the subprogram. Reference assumed-size shared memory arrays in that thread block. An array with the pinned attribute may be declared in a module or in a host subprogram. The variable arrays corresponding to the dummy arguments a and b must be set up before the call to reside on the device. Scalar arguments can be passed by value. Value dummy arguments In device subprograms. These arrays are dynamically allocated in host subprograms using the Allocate statement. Programmers must take this into account when coding. p and q. b. value :: n In this case. pinned :: q(:) Pinned arrays may be passed as arguments to host subprograms regardless of whether the interface is explicit. the value of n can be passed from the host without needing to reside in device memory. it will be automatically deallocated when the subprogram returns. if there is more than one assumed-size shared memory array. Allocating Device Memory Device arrays can have the allocatable attribute. Allocating Device and Pinned Arrays This section describes extensions to the Allocate statement. as is done in C. and other supported methods for allocating device memory. The following example declares two arrays. it must be fixed size. when such memory is available. the declaration for the array must still have the pinned attribute. This means the actual argument must be stored in device global memory. real :: p(:) allocatable :: p attributes(pinned) :: p real. starting at the same shared memory address. attributes(global) subroutine madd( a. The pinned attribute can be specified with the attributes statement. dimension(n. 17 . or the deallocation may fail. or as an attribute on the type declaration statement. following the rules of Fortran.Chapter 3. and dynamically deallocated using the Deallocate statement.

pinned allocatable arrays work and act like any other allocatable array on the host.c. or if it is exhausted. allocatable. allocatable.stat=istat) … if(allocated(b)) deallocate(b) Scalar variables can be allocated on the device using the Fortran 2003 allocatable scalar feature. and thus the arrays are treated more like C malloc’ed areas. declare and initialize the scalar on the host as: integer. The compiler will generate code to allocate the array in host page-locked memory. device :: adev(n) real. These arrays will also have a lifetime of the subprogram as is usual with the Fortran language: subroutine vfunc(a.n) real. if available.Allocating Device and Pinned Arrays real. and cudaFree can be used to free it: real. Fortran interfaces to the CUDA memory management runtime routines are provided. device :: ndev allocate(ndev) ndev = 100 The language also supports the ability to create the equivalent of automatic and local device arrays without using the allocate statement. allocatable. These functions return memory which will bypass certain Fortran allocatable properties such as automatic deallocation. 100) … istat = cudaFree(v) For a complete list of the memory management runtime routines. Mixing standard Fortran allocate/deallocate with the runtime Malloc/Free for a given array is not supported. Allocating Pinned Memory Allocatable arrays with the pinned attribute are dynamically allocated using the Allocate statement. If no such memory space is available.stat=istat) … if(allocated(p)) deallocate(p) 18 . the compiler allocates the array in normal paged host memory. Otherwise. device :: atmp(4) … end subroutine vfunc ! adev and atmp are deallocated Allocating Device Memory Using Runtime Routines For programmers comfortable with the CUDA C programming environment. device :: b(:) allocate(b(5024). refer to“Memory Management. pinned :: p(:) allocate(p(5000). device :: v(:) istat = cudaMalloc(v. The cudaMalloc function can be used to allocate single-dimensional arrays of the supported intrinsic datatypes.” on page 32. To use these. allocatable. real.

Reference To determine whether or not the allocation from page-locked memory was successful. It returns a logical success value. Implicit Data Transfer in Expressions Some limited data transfer can be enclosed within expressions. b. pinned=plog) if (. if a. Data Transfer Using Assignment Statements You can copy variables and arrays from the host memory to the device memory by using simple assignment statements in host subprograms. Note Using assignment statements to read or write device or constant data implicitly uses CUDA stream zero. Similarly. • An assignment statement where the left hand side is a host variable or host array or array section. and c are conforming host arrays. and adev. For instance. copies data from the host memory to the device global memory. the rule of thumb is all arithmetic or operations must occur on the host. you can use simple assignment statements to copy or assign variables or arrays with the constant attribute. plog) then .not. and the right hand side is a device variable or device array or array section. Data transfer between host and device memory This section provides methods to transfer data between the host and device memory.Chapter 3. and cdev are conforming device arrays. Compiler-generated temporary arrays are generated to accommodate the host copies of device data as needed. bdev. . and the right hand is a host variable or host array or array section. stat=istat. which normally only allows one device array to appear on the righthand-side of an expression. the following expressions are legal: a = adev adev = a b = a + adev c = x * adev + b 19 . . • An assignment statement where the left hand side is a device variable or device array or array section. In general. logical plog allocate(p(5000). This means such data copies are synchronous. an additional PINNED keyword is added to the allocate statement. meaning the data copy waits until all previous kernels and data copies complete. copies data from the device global memory to the host memory. • An assignment statement with a device variable or device array or array section on both sides of the assignment statement copies data between two device variables or arrays.

it is converted to dim3(block. or of type(dim3). device to host. refer to “Memory Management. bl • block is an integer. the kind parameter to the Memcpy routines is optional in Fortran since the attributes of the arrays are explicitly declared.1. The product grid%x*grid%y gives the number of thread blocks to launch. If block is an integer.” on page 32. cur. Array slices are also supported. device :: wrk(1024) real cur(512) istat = cudaMemcpy(wrk.1. If it is type(dim3). If grid is an integer. or of type(dim3). The execution configuration gives the size and shape of the grid and thread blocks that execute the function. it is converted to dim3(grid. 512) For those familiar with the CUDA C routines. or both: c = adev + bdev adev = adev + a b = sqrt(adev) Elemental transfers are supported by the language but will perform poorly. and their performance is dependent on the size of the slice. which must be less than the maximum supported by the device. bytes. Fortran interfaces to the CUDA memory management runtime routines are provided. For a complete list of memory management runtime routines. as well as the amount of shared memory to use for assumed-size shared memory arrays. the number of threads per thread block is block%x*block%y*block%z. block.1). These functions can transfer data either from the host to device. or from one device array to another. stream >>> • grid is an integer. it has the form: <<< grid. and the implementation. The execution configuration is specified after the subroutine name in the call statement. and the associated stream. Invoking a kernel subroutine A call to a kernel subroutine must give the execution configuration for the call. Counts expressed in arguments to the Fortran runtime routines are expressed in terms of data type elements. 20 . or would be more efficient written in another way. the value of grid%z must be one. not bytes. Data Transfer Using Runtime Routines For programmers comfortable with the CUDA C programming environment. If it is type(dim3). amount of contiguous data in the slices.Invoking a kernel subroutine The following expressions are not legal as they either promote a false impression of where the actual computation occurs. The cudaMemcpy function can be used to copy data between the host and the GPU: real.1).

a kernel subroutine attributes(global) subroutine sub( a ) can be called like: call sub <<< DG. they may be of derived type. are limited to the types described in this section. or declared in device subprograms. it must be an integer. such as the function arguments and execution configuration arguments. Device code Datatypes allowed Variables and arrays with the device. or a value returned by a call to cudaStreamCreate.8 1.4(default). refer to “Shared data. where the members of the derived type have one of the allowed intrinsic datatypes.1. if present. or shared attributes. They may have any of the intrinsic datatypes in the following table. If not specified. it must be a scalar integer. or if bytes is greater than the shared memory available. defined as type(dim3) integer(kind=4) :: x. See Section 4. Table 3. • stream is optional.2. DB >>> ( A ) The function call will fail if the grid or block arguments are greater than the maximum sizes allowed.z end type 21 .y.4(default). The system module cudafor includes definitions of the derived type dim3. if present. Device Code Intrinsic Datatypes Type integer logical real double precision complex character(len=1) Type Kind 1.8 4(default). and have a value of zero.8 1 (default) Additionally. Reference • bytes is optional.” on page 16.Chapter 3. allowing for static shared memory declared in the kernel and for other dedicated uses. For instance. or another allowed derived type.8 equivalent to real(kind=8) 4(default).5 on page 41.2. the value zero is used. constant. For more information. It specifies the stream to which this call is enqueued. and specifies the number of bytes of shared memory to be allocated for each thread block to use for assumed-size shared memory arrays.

griddim integer(4) :: warpsize • The variable threadidx contains the thread index within its thread block. Fortran Numeric and Logical Intrinsics Table 3. real. and griddim are available only in device subprograms. for one.Device code Builtin variables The system module cudafor declares several predefined variables. • The variable warpsize contains the number of threads in a warp. as with threadidx.real) complex integer. real real Name int logical max min mod modulo nint real sign Argument Datatypes integer. blockidx%y has the value one. They are declared as follows: type(dim3) :: threadidx. complex integer. • The variable blockidx contains the block index within the grid. The value of blockidx%z is always one. the value of griddim%z is always one. real integer. real integer. real real integer. blockdim.or two-dimensional thread blocks. The value of griddim is the same for all threads in the same grid. blockidx. Fortran intrinsics This section lists the Fortran intrinsic functions allowed in device subprograms. the value of griddim%y is one for one-dimensional grids. • The variables threadidx. • The variable blockdim contains the dimensions of the thread block. the value of griddim%z is always one. the blockdim%y and/or blockdim %z components have the value one. blockidx. The value of blockidx is the same for all threads in the same thread block. real 22 . currently defined to be 32. complex complex real real real real or (real. It has constant value.or two-dimensional thread blocks. blockdim has the same value for all threads in the same grid.2. real. for onedimensional grids. • The variable griddim contains the dimensions of the grid. Fortran Numeric and Logical Intrinsics Name abs aimag aint anint ceiling cmplx conjg dim floor Argument Datatypes integer. complex logical integer. real integer. real. These variables are read-only. for one. the threadidx%y and/or threadidx%z components have the value one. blockdim.

Chapter 3. complex real real Fortran Numeric Inquiry Intrinsics Table 3. real real real Name precision radix range selected_int_kind selected_real_kind tiny Argument Datatypes real. complex Name log log10 sin sinh sqrt tan tanh Argument Datatypes real. complex real real.3. complex integer (integer. real. complex integer. Fortran Bit Manipulation Intrinsics Name btest iand ibclr ibits ibset ieor ior Argument Datatypes integer integer integer integer integer integer integer Name ishft ishftc leadz mvbits not popcnt poppar Argument Datatypes integer integer integer integer integer integer integer 23 . real integer. Fortran Mathematical Intrinsics Name acos asin atan atan2 cos cosh exp Argument Datatypes real real real (real.4. Reference Fortran Mathematical Intrinsics Table 3.integer) real Fortran Bit Manipulation Intrinsics Table 3.5. Fortran Numeric Inquiry Intrinsics Name bit_size digits epsilon huge maxexponent minexponent Argument Datatypes integer integer.real) real. complex real real. complex real real. real real integer.

write-after-read.6. To avoid these potential issues. complex integer. refer to “Thread Management. real. Fortran Random Number Intrinsics Name random_number random_seed Argument Datatypes real integer New Intrinsic Functions This section describes the new intrinsic functions and subroutines supported in device subprograms.7. thus creating potential read-after-write. syncthreads syncthreads_count syncthreads_and syncthread_or threadfence threadfence_block threadfence_system For detailed information on these functions.” on page 32.Device code Fortran Reduction Intrinsics Table 3. real integer. complex Fortran Random Number Intrinsics Table 3. real integer. This 24 . or write-after-write hazards for some of these memory accesses. Fortran Reduction Intrinsics Name all any count maxloc maxval Argument Datatypes logical logical logical integer. Synchronization Functions The synchronization functions control the synchronization of various threads during execution of thread blocks. SYNCTHREADS The syncthreads intrinsic subroutine acts as a barrier synchronization for all threads in a single thread block. use syncthreads()to specify synchronization points in the kernel. real integer. real. real Name minloc minval product sum Argument Datatypes integer. . it has no arguments: void syncthreads() Sometimes threads within a block access the same addresses in shared or global memory.

acts a as a barrier at which all threads in the block must wait before any thread is allowed to proceed. Typically. like syncthreads. Each block first sums a subset of the array and stores the result in global memory. If any thread in a thread block issues a call to syncthreads. other threads may see the effects of these memory writes in a different order. and threadfence_system() to create a memory fence to enforce ordering. each block atomically increments a counter to signal that it is done with computing and storing its partial sum. Reference intrinsic acts as a barrier at which all threads in the block must wait before any thread is allowed to proceed. the last block done reads each of these partial sums from global memory and sums them to obtain the final result. When all blocks are done. acts as a barrier at which all threads in the block must wait before any thread is allowed to proceed.Chapter 3. threadfence_block(). syncthreads_or evaluates the integer argument int_value for all threads of the block and returns non-zero if and only if int_value evaluates to non-zero for any of them. You can use threadfence(). all threads must also reach and execute the same call statement. SYNCTHREADS_AND integer syncthreads_and(int_value) syncthreads_and. acts as a barrier at which all threads in the block must wait before any thread is allowed to proceed. If no fence is placed between storing the partial sum and incrementing the counter. creating a wait. SYNCTHREADS_OR integer syncthreads_or(int_value) syncthreads_or. or the kernel fails to complete correctly. In addition. the counter might increment before the partial sum is stored. suppose you use a kernel to compute the sum of an array of N numbers in one call. 25 . other threads may see the effects of these memory writes in a different order. syncthreads_and evaluates the integer argument int_value for all threads of the block and returns non-zero if and only if int_value evaluates to non-zero for all of them. Each thread in a thread block pauses at the syncthreads call until all threads have reached that call. For example. when a thread issues a series of writes to memory in a particular order. Threads within a block cooperate and share data by synchronizing their execution to coordinate memory accesses. To determine which block is finished last. like syncthreads. In addition. In addition. when a thread issues a series of writes to memory in a particular order. THREADFENCE void threadfence() threadfence acts as a memory fence. Memory Fences In general. syncthreads_count evaluates the integer argument int_value for all threads of the block and returns the number of threads for which int_value evaluates to non-zero. like syncthreads. SYNCTHREADS_COUNT integer syncthreads_count(int_value) syncthreads_count.

26 . THREADFENCE_SYSTEM void threadfence_system() threadfence_system acts as a memory fence.Device code threadfence() is one method to enforce a specific order. The function allthreads evaluates its argument for all threads in the current warp. only if the value of the argument is . The value of the function is .0 or higher.true.true. GPU_TIME The gpu_time intrinsic returns the value of the clock cycle counter on the GPU.2 and higher.” on page 32. The clock frequency can be determined by calling cudaGetDeviceProperties. creating a wait until all global and shared memory accesses made by the calling thread prior to threadfence_block() are visible to all threads in the thread block for all accesses. as described in “cudaGetDeviceProperties.0) ) allneg = . creating a wait until all global and shared memory accesses made by the calling thread prior to threadfence_system() are visible to: • All threads in the thread block for shared memory accesses • All threads in the device for global memory accesses • Host threads for page-locked host memory accesses threadfence_system() is only supported by devices of compute capability 2. It has a single argument: integer(8) clock call gpu_time(clock) The argument to gpu_time is set to the value of the clock cycle counter.true. Each of these functions has a single argument. Warp-Vote Operations Warp-vote operations are only supported by devices with compute capability 1. All global and shared memory accesses made by the calling thread prior to threadfence() are visible to: • All threads in the thread block for shared memory accesses • All threads in the device for global memory accesses THREADFENCE_BLOCK void threadfence_block() threadfence_block acts as a memory fence. ALLTHREADS The allthreads function is a warp-vote operation with a single scalar logical argument: if( allthreads(a(i)<0. for all threads in the warp.

Atomic Functions The atomic functions read and write the value of their first operand. only if the value of the argument is . Atomic functions are only supported by devices with compute capability 1.true. The value of the function is . Arithmetic and Bitwise Atomic Functions Function atomicadd( mem. Arithmetic and Bitwise Atomic Functions These atomic functions read and return the value of the first argument. The function anythread evaluates its argument for all threads in the current warp. which must be a variable or array element in shared memory (with the shared attribute) or in device global memory (with the device attribute). Both arguments must be of type integer(kind=4). Note The return value for each of these functions is the first argument.true.false. BALLOT The ballot function is a warp-vote operation with a single integer argument: unsigned integer ballot(int_value) The function ballot evaluates the argument int_value for all threads of the warp and returns an integer whose Nth bit is set if and only if int_value evaluates to non-zero for the Nth thread of the warp. The atomic functions return correct values even if multiple threads in the same or different thread blocks try to read and update the same location without any synchronization.8.1 and higher. value ) Additional Atomic Update mem = mem + value mem = mem – value 27 . value ) atomicsub( mem.0.false. Reference ANYTHREAD The anythread function is a warp-vote operation with a single scalar logical argument: if( anythread(a(i)<0.2 or higher is required if the first argument has the shared attribute. and store the combined value back to the first argument location. Compute capability 1. depending on the function. This function is only supported by devices of compute capability 2. Example: if( ballot(int_value) ) allneg = .Chapter 3.0) ) allneg = . mem. These functions are: Table 3. They also combine that value with the value of the second argument. for all threads in the warp.

value) mem = min(mem. These functions are intended to implement circular counters. value ) atomicmin( mem.value) mem = ieor(mem. These functions are: Table 3. and atomically stores a new value back to the first argument location if the first and second argument are equal.and. imax ) Compare and Swap Atomic Function This atomic function reads and returns the value of the first argument.value) mem = ior(mem. Counting Atomic Functions Function atomicinc( mem. value ) atomicor( mem. All three arguments must be of type integer(kind=4). Note The return value for each of these functions is the first argument. value ) mem = max(mem. mem.value) mem = value Counting Atomic Functions These atomic functions read and return the value of the first argument. value ) atomicxor( mem.value) mem = iand(mem. Note The return value for this function is the first argument. depending on the result of the comparison. It also compares the first argument with the second argument.9. The function is: 28 . mem>0) then mem = mem-1 else mem = imax endif atomicdec( mem.Device code atomicmax( mem. They also compare the first argument with the second argument. value ) atomicand( mem. Both arguments must be of type integer(kind=4). mem. and stores a new value back to the first argument location. imax ) Additional Atomic Update if (mem<imax) then mem = mem+1 else mem = 0 endif if (mem<imax . counting up to or down from a maximum value specified in the second argument. value ) atomicexch( mem.

• ENTRY statements are not allowed. • Objects with character type must have LEN=1. SIZEOF Intrinsic A call to sizeof(A). Host code Host subprograms may use intrinsic functions. • Automatic arrays must be fixed size. • Objects with the Pointer and Allocatable attribute are not allowed.3 are not supported. returns the number of bytes required to hold the value of A. j j = sizeof(i) ! this assigns the value 4 to j 29 . • Subroutine and function calls are supported only if they can be inlined. • Input/Output statements are not allowed: READ. • Fortran intrinsic functions not listed in Section 3.10. • Alternate return specifications are not allowed. OPEN. CLOSE. WRITE. such as the new sizeof intrinsic function. BACKSPACE. PRINT. ENDFILE. integer(kind=4) :: i. Compare and Swap Atomic Function Function atomiccas(mem.comp.Chapter 3. NAMELIST. Reference Table 3. • Cray pointers are not supported. where A is a variable or expression.6. INQUIRE.val) Additional Atomic Update if (mem == comp) then mem = val endif Restrictions This section lists restrictions on statements and features that can appear in device subprograms. FORMAT. • Recursive subroutines and functions are not allowed. • STOP and PAUSE statements are not allowed. • Optional arguments are not allowed. character substrings are not supported. • Floating point exception handling is not supported. REWIND.

30 .

” on page 41. Runtime APIs The system module cudafor defines the interfaces to the Runtime API routines. 31 . Tip When doing timing runs. refer to “Error Handling. cudaSetDevice integer function cudaSetDevice( devnum ) integer. intent(in) :: flags cudaSetDeviceFlags records how the CUDA runtime interacts with this host thread. and a nonzero value if there was an error. Device Management Use the functions in this section for device management.Chapter 4. the runtime initializes and connects to the device the first time a runtime routine is called or a device array is allocated. intent(in) :: devnum cudaSetDevice selects the device to associate with this host thread. To interpret the error codes. Most of the runtime API routines are integer functions that return an error code. they return a value of zero if the call was successful. cudaSetDeviceFlags integer function cudaSetDevice( flags ) integer. Initialization No explicit initialization is required. cudaGetDeviceCount integer function cudaGetDeviceCount( numdev ) integer. intent(out) :: numdev cudaGetDeviceCount assigns the number of available devices to its first argument. be aware that initialization can add some overhead.

cudaThreadExit integer function cudaThreadExit() cudaThreadExit explicitly cleans up all runtime-related CUDA resources associated with the host thread. intent(out) :: devnum cudaGetDevice assigns the device number associated with this host thread to its first argument. it is implicitly called when the host thread exits. as arguments to simplify interfacing to existing CUDA C code. cudaGetDeviceProperties integer function cudaGetDeviceProperties( prop.Thread Management cudaGetDevice integer function cudaGetDevice( devnum ) integer. Any subsequent CUDA calls or operations will reinitialize the runtime. Some can also take C types. intent(in) :: devnum cudaGetDeviceProperties returns the properties of a given device. Calling cudaThreadExit is optional. It may return an error condition if one of the preceding operations fails. devnum ) type(cudadeviceprop). intent(in) :: prop cudaChooseDevice assigns the device number that best matches the properties given in prop to its first argument. defined in the cudafor module. Consistent use of TYPE(C_PTR) and 32 . Memory Management Many of the memory management routines can take device arrays as arguments. intent(out) :: prop integer. CUDA Fortran has extended the F2003 derived type TYPE(C_PTR) by providing a C device pointer. cudaThreadSynchronize integer function cudaThreadSynchronize() cudaThreadSynchronize blocks execution of the host subprogram until all preceding kernels and operations are complete. Thread Management Sometimes threads within a block access the same addresses in shared or global memory. prop ) integer. To avoid these potential issues. provided through the Fortran 2003 iso_c_binding module. cudaChooseDevice integer function cudaChooseDevice ( devnum. or write-after-write hazards for some of these memory accesses. write-after-read. intent(out) :: devnum type(cudadeviceprop). use the functions in this section for thread management. thus creating potential read-after-write. as TYPE(C_DEVPTR).

1. height) cudaMallocPitch allocates data on the device. width. cudaMallocPitch integer function cudaMallocPitch(devptr. an allocatable device array as the second argument. devptr may also be of TYPE(C_DEVPTR).1. Use the functions in this section for memory management. devptr may be any allocatable. cudaMalloc integer function cudaMalloc(devptr. cudaFreeArray integer function cudaFreeArray(carray) type(cudaArrayPtr) :: carray 33 . Both of these features are subject to change when. The height is an integer. Or. as well as consistency checks between Fortran device arrays and host arrays.” on page 21. in which case the integer values are expressed in bytes.1. there is also a function C_DEVLOC() defined which will create a TYPE(C_DEVPTR) that holds the C address of the Fortran device array argument. Similarly. one-dimensional device array of a supported type specified in Table 3. devptr may be of TYPE(C_DEVPTR). devptr may be any allocatable device array of a supported type specified in Table 3. it is possible to construct a Fortran device array out of a TYPE(C_DEVPTR) by using an extension of the iso_c_binding subroutine c_f_pointer. count) cudaMalloc allocates data on the device.” on page 21. Currently. pitch. c_f_pointer will take a TYPE(C_DEVPTR) as the first argument. The width is in terms of number of elements. devptr may be of TYPE(C_DEVPTR). should be of benefit. two-dimensional device array of a supported type specified in Table 3.” on page 21. and the padded width is returned in the variable pitch. proper Fortran pointers for device data are supported. Or. cdesc. cudaFree integer function cudaFree(devptr) cudaFree deallocates data on the device. in the future. height cudaMallocArray allocates a data array on the device. Under CUDA Fortran. and in effect transfer the allocation to the Fortran array. width. cudaMallocArray integer function cudaMallocArray(carray. “Device Code Intrinsic Datatypes. a shape as the third argument. in which case the count is in bytes. height) type(cudaArrayPtr) :: carray type(cudaChannelFormatDesc) :: cdesc integer :: width. devptr may be any allocatable. “Device Code Intrinsic Datatypes. Runtime APIs TYPE(C_DEVPTR).Chapter 4. The count is in terms of elements. “Device Code Intrinsic Datatypes. cudaMallocPitch may pad the data.

kdir may be optional. count) cudaMemset sets a location or array to the specified value. in which case the count is in term of bytes. “Device Code Intrinsic Datatypes. for more information. of a supported type specified in Table 3. src. count. height) cudaMemset2D sets an array to the specified value. and the lowest byte of value is used. dst and src may be any device or host.Memory Management cudaFreeArray frees an array that was allocated on the device. cudaMemcpyDeviceToHost. cudaMemset2D integer function cudaMemset2D(devptr. “Device Code Intrinsic Datatypes. height. for more information. “Device Code Intrinsic Datatypes.” on page 21. Or. or cudaMemcpyDeviceToDevice.1. devptr may be any device array of a supported type specified in Table 3. in which case the pitch.1. stream) cudaMemcpy copies data from one location to another.1. width. devptr may be any device scalar or array of a supported type specified in Table 3. The count is in terms of elements. count. This function operates on page-locked host memory only. devptr may be of TYPE(C_DEVPTR). The value must match in type and kind. The pitch. pitch. of a supported type specified in Table 3. The count is in terms of elements. width. The value must match in type and kind.” on page 20. otherwise the stream argument is optional and defaults to zero. and the lowest byte of value is used. kdir may be optional. “Device Code Intrinsic Datatypes. width. dpitch.” on page 21. If kdir is specified. Or. Alternatively. in which case the count is in term of bytes.” on page 21. or cudaMemcpyDeviceToDevice. scalar or array. in which case the count is in term of bytes. cudaMemcpy2D integer function cudaMemcpy2D(dst. dst and src may be of TYPE(C_DEVPTR) or TYPE(C_PTR). The width 34 . kdir) cudaMemcpy copies data from one location to another. src. devptr may be of TYPE(C_DEVPTR). cudaMemcpyAsync integer function cudaMemcpyAsync(dst. cudaMemcpy integer function cudaMemcpy(dst. kdir. width.” on page 21. Alternatively.1.” on page 20. scalar or array. The count is in terms of elements.1. it must be one of the defined enums cudaMemcpyHostToDevice. “Device Code Intrinsic Datatypes. dst and src may be of TYPE(C_DEVPTR) or TYPE(C_PTR). cudaMemset integer function cudaMemset(devptr. refer to “Data Transfer Using Runtime Routines. dst and src may be any device or host. dst and src may be any device or host array.” on page 21. spitch. src. refer to “Data Transfer Using Runtime Routines. value. kdir) cudaMemcpy2D copies data from one location to another. cudaMemcpyDeviceToHost. and height are in terms of bytes. If kdir is specified. of a supported type specified in Table 3. value. The copy can be associated with a stream by passing a non-zero stream argument. it must be one of the defined enums cudaMemcpyHostToDevice. and height are in terms of elements.

for more information. width. dsty. for more information. dsty.” on page 20. srca. dst and src may be of TYPE(C_DEVPTR) or TYPE(C_PTR). spitch. The width and height are in terms of elements. spitch. in which case the width and height are in term of bytes. srcx. kdir) type(cudaArrayPtr) :: srca integer :: dstx. dsty. height. kdir. dst and src may be of TYPE(C_DEVPTR) or TYPE(C_PTR). height. count. Alternatively. dst and src may be any device or host array. or cudaMemcpyDeviceToDevice. otherwise the stream argument is optional and defaults to zero. srca. it must be one of the defined enums cudaMemcpyHostToDevice. If kdir is specified. kdir cudaMemcpy2DFromArray copies array data to and from the device. or cudaMemcpyDeviceToDevice. Alternatively.1. kdir) type(cudaArrayPtr) :: srca integer :: dpitch. kdir cudaMemcpyToArray copies array data to and from the device. cudaMemcpy2DToArray integer function cudaMemcpy2DToArray(dsta.” on page 20. src. dpitch. cudaMemcpyToArray integer function cudaMemcpyToArray(dsta. dsty. srcy. refer to “Data Transfer Using Runtime Routines. width. dpitch. kdir cudaMemcpyFromArray copies array data to and from the device. height. “Device Code Intrinsic Datatypes. cudaMemcpyDeviceToHost. count. height. kdir cudaMemcpy2DToArray copies array data to and from the device. width. width. The copy can be associated with a stream by passing a non-zero stream argument. dsty. refer to “Data Transfer Using Runtime Routines. it must be one of the defined enums cudaMemcpyHostToDevice. kdir) type(cudaArrayPtr) :: dsta integer :: dstx. cudaMemcpy2DFromArray integer function cudaMemcpy2DFromArray(dst. cudaMemcpyFromArray integer function cudaMemcpyFromArray(dst. kdir may be optional. count. cudaMemcpyDeviceToHost.Chapter 4. If kdir is specified. spitch. src. of a supported type specified in Table 3. count. dstx. width. kdir) type(cudaArrayPtr) :: dsta integer :: dstx. srcy. stream) cudaMemcpy2D copies data from one location to another. srcx. height. srcy. src. Runtime APIs and height are in terms of elements. dstx.” on page 21. 35 . srcx. cudaMemcpy2DAsync integer function cudaMemcpy2DAsync(dst. in which case the width and height are in term of bytes. This function operates on page-locked host memory only. kdir may be optional.

Alternatively. cudaMalloc3D integer function cudaMalloc3D(pitchptr. count. dstx. cdesc. srcx. intent(in) :: cext cudaMalloc3D allocates data on the device. height. cudaMemcpy3D integer function cudaMemcpy3D(p) type(cudaMemcpy3DParms) :: p cudaMemcpy3D copies elements from one 3D array to another as specified by the data held in the derived type p. kdir cudaMemcpyArrayToArray copies array data to and from the device.Memory Management cudaMemcpyArrayToArray integer function cudaMemcpyArrayToArray(dsta. which was allocated with cudaMalloc3D to a specified value. intent(out) :: pitchptr type(cudaExtent). srcy. srca. count. srca integer :: dstx. srcy. pitchptr may be any allocatable. three-dimensional device array of a supported type specified in “Datatypes allowed. height. dstx. value. srcx. dsty. srcx. pitchptr is a derived type defined in the cudafor module. srca integer :: dstx. the extents in each dimension specified by cext. cext is also a derived type which holds the extents of the allocated array. width. dsty. 36 . kdir) type(cudaArrayPtr) :: dsta. srca. kdir) type(cudaArrayPtr) :: dsta. dsty. srcx. kdir cudaMemcpy2DArrayToArray copies array data to and from the device. cudaMemset3D integer function cudaMemset3D(pitchptr. srcy. cudaMemcpy2DArrayToArray integer function cudaMemcpy2DArrayToArray(dsta. dsty. cext) type(cudaPitchedPtr) :: pitchptr integer :: value type(cudaExtent) :: cext cudaMemset3D sets elements of an array. cext) type(cudaArrayPtr) :: carray type(cudaChannelFormatDesc) :: cdesc type(cudaExtent) :: cext cudaMalloc3DArray allocates array data on the device. cudaMalloc3DArray integer function cudaMalloc3DArray(carray. width.” on page 21. cext) type(cudaPitchedPtr). srcy.

This function operates on page-locked host memory only. src. count. offset. The count is in terms of elements.” on page 21. cudaMemcpyFromSymbol integer function cudaMemcpyFromSymbol(dst. offset. count. offset.” on page 21. cudaMemcpyFromSymbolAsync integer function cudaMemcpyFromSymbolAsync(dst. optional :: stream 37 . count. stream) type(cudaSymbol) :: symbol integer :: count. optional :: stream cudaMemcpyToSymbol copies data from the source to a device area in global or constant memory space referenced by a symbol. kdir. kdir integer. The copy can be associated with a stream by passing a non-zero stream argument. offset. kdir integer. src may be any host scalar or array of a supported type specified in “Datatypes allowed. offset. stream) type(cudaMemcpy3DParms) :: p integer :: stream cudaMemcpy3DAsync copies elements from one 3D array to another as specified by the data held in the derived type p. cudaMemcpyToSymbol integer function cudaMemcpyToSymbol(symbol. kdir) type(cudaSymbol) :: symbol integer :: count. symbol. src. kdir cudaMemcpyToSymbol copies data from the source to a device area in global or constant memory space referenced by a symbol. symbol. count. This function operates on page-locked host memory only. offset. The count is in terms of elements. src may be any host scalar or array of a supported type as specified in “Datatypes allowed. The count is in terms of elements. cudaMemcpyToSymbolAsync integer function cudaMemcpyToSymbolAsync(symbol. dst may be any host scalar or array of a supported type specified in “Datatypes allowed.” on page 21. kdir cudaMemcpyFromSymbol copies data from a device area in global or constant memory space referenced by a symbol to a destination on the host. kdir) type(cudaSymbol) :: symbol integer :: count. The copy can be associated with a stream by passing a non-zero stream argument. kdir. Runtime APIs cudaMemcpy3DAsync integer function cudaMemcpy3D(p. offset. offset.Chapter 4. stream) type(cudaSymbol) :: symbol integer :: count.

vx.Memory Management cudaMemcpyFromSymbol copies data from a device area in global or constant memory space referenced by a symbol to a destination on the host. Size is in bytes. cudaHostGetDevicePointer integer function cudaHostGetDevicePointer(devptr. It returns in devptr an address that can be passed to. cudaGetSymbolAddress integer function cudaGetSymbolAddress(devptr. flags cudaHostAlloc allocates pinned memory on the host. dst may be any host scalar or array of a supported type specified in “Datatypes allowed. The following code sequence initializes a global device array “vx” from a CUDA C kernel: type(cudaSymbol) :: csvx type(c_devptr) :: cdvx real. This function operates on page-locked host memory only. a kernel which runs on 38 . allocatable. flags) type(C_PTR) :: hostptr integer :: size. cudaHostAlloc integer function cudaHostAlloc(hostptr. and read and written by. The normal iso_c_binding subroutine c_f_pointer can be used to move the type(c_ptr) to a Fortran pointer.” on page 21. csvx) Call c_f_pointer(cdvx. The count is in terms of elements. The copy can be associated with a stream by passing a non-zero stream argument. symbol) type(C_DEVPTR) :: devptr type(cudaSymbol) :: symbol cudaGetSymbolAddress returns in the devptr argument the address of symbol on the device. hostptr is a pinned memory buffer that was allocated via cudaHostAlloc(). symbol) integer :: size type(cudaSymbol) :: symbol cudaGetSymbolSize sets the variable size to the size of a device area in global or constant memory space referenced by the symbol. flags) type(C_DEVPTR) :: devptr type(C_PTR) :: hostptr integer :: flags cudaHostGetDevicePointer returns a pointer to a device memory address corresponding to the pinned memory on the host. A symbol can be set to an external device name via a character string. It returns in hostptr the address of the pagelocked allocation. device :: vx(:) csvx = “vx” Istat = cudaGetSymbolAddress(cdvx.0 cudaGetSymbolSize integer function cudaGetSymbolSize(size. or returns an error if the memory is unavailable. The flags argument enables different options to be specified that affect the allocation. 100) Vx = 0. size. hostptr.

It may return error codes from previous.Chapter 4. or returns an error if the memory is unavailable. and the value cudaErrorNotReady if not. cudaStreamDestroy integer function cudaStreamDestroy( stream ) integer. intent(out) :: stream cudaStreamCreate creates an asynchronous stream and assigns its identifier to its first argument. Stream Management Use the functions in this section for stream management. Runtime APIs the device. The flags argument is provided for future releases. cudaFreeHost integer function cudaFreeHost(hostptr) type(C_PTR) :: hostptr cudaFreeHost deallocates pinned memory on the host allocated with cudaMalloHost. intent(in) :: stream cudaStreamSynchronize blocks execution of the host subprogram until all preceding kernels and operations associated with the given stream are complete. it returns zero (success) if all operations are complete. intent(in) :: stream cudaStreamDestroy releases any resources associated with the given stream. size) type(C_PTR) :: hostptr integer :: size cudaMallocHost allocates pinned memory on the host. cudaMallocHost integer function cudaMallocHost(hostptr. intent(in) :: stream cudaStreamQuery tests whether all operations enqueued to the selected stream are complete. 39 . cudaStreamCreate integer function cudaStreamCreate( stream ) integer. It may also return another error condition if some asynchronous operations failed. The normal iso_c_binding subroutine c_f_pointer can be used to move the type(c_devptr)to a device array. cudaStreamSynchronize integer function cudaStreamSynchronize( stream ) integer. It returns in hostptr the address of the page-locked allocation. The normal iso_c_binding subroutine c_f_pointer can be used to move the type(c_ptr) to a Fortran pointer. asynchronous operations. size is in bytes. cudaStreamQuery integer function cudaStreamQuery( stream ) integer.

intent(in) :: stream cudaEventRecord issues an operation to the given stream to record an event. It returns cudaErrorInvalidValue if either event has not yet been recorded. intent(in) :: event integer. cudaEventCreate integer function cudaEventCreate( event ) type(cudaEvent). end) float :: time type(cudaEvent). intent(in) :: event cudaEventDestroy destroys the resources associated with an event object. This function is only valid with events recorded on stream zero. cudaEventElapsedTime integer function cudaEventElapsedTime( time. If stream is zero. The event is recorded after all preceding operations in the stream are complete. intent() :: start. 40 .Event Management Event Management Use the functions in this section to manage events. intent(out) :: event cudaEventCreate creates an event object and assigns the event identifier to its first argument cudaEventRecord integer function cudaEventRecord( event. and cudaErrorNotReady if it has not. end cudaEventElapsedTime computes the elapsed time between two events (in milliseconds). cudaEventSynchronize integer function cudaEventSynchronize( event ) type(cudaEvent). It returns a value of cudaErrorInvalidValue if cudaEventRecord has not been called for this event. cudaEventDestroy integer function cudaEventDestroy( event ) type(cudaEvent). the event is recorded after all preceding operations in all streams are complete. It returns success (zero) if the event has been recorded. It returns cudaErrorInvalidValue if cudaEventRecord has not been called for this event. cudaEventQuery integer function cudaEventQuery( event ) type(cudaEvent). intent(in) :: event cudaEventSynchronize blocks until the event has been recorded. start. intent(in) :: event cudaEventQuery tests whether an event has been recorded. stream ) type(cudaEvent).

This function automatically returns cudaErrorInvalidValue if the iversion argument is NULL. 41 . cudaGetErrorString function cudaGetErrorString( errcode ) integer. cudaGetLastError integer function cudaGetLastError() cudaGetLastError returns the error code that was most recently returned from any runtime call in this host thread. cudaDriverGetVersion integer function cudaDriverGetVersion(iversion) integer :: iversion cudaDriverGetVersion returns the version number of the installed CUDA driver as iversion. then it returns 0 as iversion. Runtime APIs Error Handling Use the functions in this section for error handling. cudaRuntimeGetVersion integer function cudaRuntimeGetVersion(iversion) integer :: iversion cudaRuntimeGetVersion returns the version number of the installed CUDA Runtime as iversion. intent(in) :: errcode character*(*) :: cudaGetErrorString cudaGetErrorString returns the message string associated with the given error code. If no driver is installed. This function automatically returns cudaErrorInvalidValue if the iversion argument is NULL. Version Management Use the functions in this section for version management.Chapter 4.

42 .

16). N. M. accumulates the submatrix product. B is MxL. k. and has not been highly optimized for execution time. Source Code Listing ! start the module containing the matmul kernel module mmul_mod use cudafor contains ! mmul_kernel computes A*B into C where ! A is NxM.L). C. ty ! submatrices stored in shared memory real. C is then NxL attributes(global) subroutine mmul_kernel( A. C(N. kb. The submatrix size is chosen so the number of threads in a block is a multiple of the warp size (32) and is less than the maximum number of threads per thread block (512). B(M. The program caches the submatrices of A and B in the fast shared memory.com web page at: www. The program computes the products by accumulating submatrix products. L integer :: i. j. For simplicity. Examples This chapter contains For up-to-date information about the state of the current release.M).com/support/index. M. tx.j) being computed 43 . B.L) integer.Chapter 5.htm Matrix Multiplication Example This example shows a program to compute the product C of two matrices A and B. value :: N. it reads a block submatrix of A and a block submatrix of B. Each element of the result is the product of one row of A by one column of B. L ) real :: A(N. shared :: Asub(16. visit the frequently asked questions (FAQ) section of the pgroup. as follows: • Each thread block computes one 16x16 submatrix of C. • Each thread within the block computes one element of the submatrix. Bsub(16. then moves to the next submatrix of A rowwise and of B columnwise. the program assumes the matrix sizes are a multiple of 16.pgroup.16) ! the value of C(i.

M.kb+ty-1) Bsub(tx. dimBlock ! Get the array sizes N = size( A.j) = sum(A(i. 16.k) * Bsub(k. ! Copy the results back and free up memory C(1:N. 1 ) dimBlock = dim3( 16. 1 ) M = size( A. 2 ) ! Allocate the device arrays allocate( Adev(N.ty) enddo ! Synchronize to make sure all threads are done ! reading the submatrices before overwriting them ! in the next iteration of the kb loop call syncthreads() enddo ! Each of the 16x16 threads stores its element ! to the global C array C(i. Bdev(M.j) do k = 1. device. B.Matrix Multiplication Example real :: Cij ! Get the thread indices tx = threadidx%x ty = threadidx%y ! This thread computes C(i. Bdev.j) ! Wait until all elements are filled call syncthreads() ! Multiply the two submatrices ! Each of the 16x16 threads accumulates the ! dot product for its element of C(i. allocatable.j)) i = (blockidx%x-1) * 16 + tx j = (blockidx%y-1) * 16 + ty Cij = 0. the block size do kb = 1. dimension(:.:) * B(:. C ! allocatable device arrays real.16 Cij = Cij + Asub(tx.1:L) = Cdev deallocate( Adev.:) = B(1:M. 1 ) call mmul_kernel<<<dimGrid.ty) = A(i.Bdev. Cdev.1:L) ! Create the grid and block dimensions dimGrid = dim3( N/16.L). Cdev(N.Cdev ! dim3 variables to define the grid and block shapes type(dim3) :: dimGrid.:) :: A. C ) real. dimension(:.ty) = B(kb+tx-1. Bdev. L/16.j) = Cij end subroutine mmul_kernel ! The host routine to drive the matrix multiplication subroutine mmul( A. 16 ! Fill the submatrices ! Each of the 16x16 threads in the thread block ! loads one element of Asub and Bsub Asub(tx.0 ! Do the k loop in chunks of 16.dimBlock>>>( Adev. 2 ) L = size( B.1:M) Bdev(:. B.M).:) :: Adev. Cdev ) end subroutine mmul end module mmul_mod N 44 .L) ) ! Copy A and B to the device Adev = A(1:N.

• It steps through the arrays A and B in blocks of size 16. This routine computes the dot product of A(i. • It fills dimGrid and dimBlock to hold the grid and thread block sizes. • It calls mmul_kernel to compute Cdev on the device. • It synchronizes to make sure both submatrices are loaded by all threads in the block. • Finally.Chapter 5. • It initializes a scalar in which it will accumulate the dot product.:)*B(:. A and B. it stores the computed value into the correct element of C. passed as assumed-shape arrays. • It allocates device memory arrays Adev. Because the data copy operations are synchronous. • It synchronizes again to make sure all threads are done reading the submatrices before starting the next block. A and B. The thread executing this routine is one of 16x16 threads cooperating in a thread block. • It frees the device memory arrays. MMUL This host subroutine has two input arrays. • It copies the arrays A and B to Adev and Bdev using array assignments. The routine performs the following operations: • It determines the size of the matrices in N. The host subroutine mmul is a wrapper for the kernel routine mmul_kernel. for which element of C(i. and L. and one output array. MMUL_KERNEL This kernel subroutine has two device memory input arrays. it does the following steps: • It loads one element of the submatrices of A and B into shared memory. C. no extra synchronization is needed between the copy operations and the kernel launch. depending on the block and thread index.j) for a particular value of i and j.j) it is computing. and three scalars giving the array sizes. 45 . • For each block. Examples Source Code Description This source code module mmul_mod has two subroutines. • It determines the i and j indices. M. C. one device memory output array. • It copies Cdev back from device memory to C. and Cdev. It performs the following operations: • It determines the thread indices for this thread. • It accumulates the dot product of its row and column of the submatrices. Bdev.

46 .

com trs@pgroup. Many questions and problems can be resolved at our frequently asked questions (FAQ) site at www.com All technical support is by email or submissions using an online form at www. OR 97035 USA The PGI User Forum is monitored by members of the PGI engineering and support teams as well as other PGI customers.com/resources/docs. Contact Information You can contact The Portland Group at: The Portland Group STMicroelectronics.com/support/faq. 47 .htm. Two Centerpointe Drive Lake Oswego.Chapter 6. Phone support is not currently available.php Or contact us electronically using any of the following means: Fax Sales Support WWW +1-503-682-2637 sales@pgroup. Inc.pgroup.com/userforum/index.com/support.pgroup.pgroup. The forum newsgroups may contain answers to commonly asked questions.com www.htm or in your local copy of the documentation in the release directory doc/index. PGI documentation is available at www. Log in to the PGI website to access the forum: www.pgroup.htm.pgroup.

48 .

Sign up to vote on this title
UsefulNot useful