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

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

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

viii .

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

In this case. multiple filenames are allowed.pgroup. C/C++ C/C++ language statements are shown in the test of this guide using a reduced fixed point size. In this case item1 is optional.. PGI provides a glossary of terms which you can access at www.nvidia.nvidia. Geneva. Available online at http://www. FORTRAN Fortran language statements are shown in the text of this guide using a reduced fixed point size.1. and for language statements in the text. Information Technology – Programming Languages – FORTRAN. arguments. filename . { item2 | item 3} braces indicate that a selection is required. September.com/cuda. The Portland Group.com/doc/pgiug. x . Terminology If there are terms in this guide with which you are unfamiliar. square brackets are required to specify a p/t-set. you must select either item2 or item3. NVIDIA. • NVIDIA CUDA Programming Guide. 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. 7/21/2010. • ISO/IEC 1539-1:1997. Version 3.com/support/definitions. NVIDIA. 1997 (Fortran 95).com/cuda. ellipsis indicate a repetition.pgroup. Bold is used for commands. Available online at http:// www. • NVIDIA CUDA Compute Unified Device Architecture Reference Manual. examples. In this example. In the context of p/tsets.1. options. 2009. There are a wide variety of releases and distributions of each of these types of operating systems.. June 2010.Terminology Constant Width is used for filenames. Version 3.pdf. Available online at http:// www. • PGI Users Guide. Zero or more of the preceding item may occur.htm Related Publications The following documents contain additional information related to CUDA Fortran programming. square brackets indicate optional items. [ item1 ] in general. Release 10.9. directories. including assembly language statements.1.

The programming model supports four key abstractions: cooperating threads organized into thread groups. A CUDA programmer must partition the program into coarse grain blocks that can be executed in parallel. Mac OS X and Windows. PGI 2011 includes support for CUDA Fortran on Linux.Chapter 1. a small set of extensions to Fortran that supports and is built upon the CUDA computing architecture. 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. Graphic processing units or GPUs have evolved into programmable. Each block is partitioned into fine grain threads. and tremendous potential for many applications. NVIDIA introduced CUDA™. CUDA Fortran includes a Fortran 2003 compiler and tool chain for programming NVIDIA GPUs using Fortran. with compilers and libraries to support the programming of NVIDIA GPUs. shared memory and barrier synchronization within thread groups. GPU designs are optimized for the computations found in graphics rendering. here called CUDA C. highly parallel computational units with very high memory bandwidth. which can cooperate using shared memory and barrier synchronization. Compared to the PGI Accelerator directives-based model and compilers. but are general enough to be useful in many data-parallel. A properly designed CUDA program will run on any CUDA-enabled GPU. regardless of the number of available processor cores. a general purpose parallel programming architecture. 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. allowing direct programming of the GPU from a high level language. CUDA comes with an extended C compiler. and coordinated independent thread groups organized into a grid. CUDA Fortran is an analog to NVIDIA's CUDA C compiler. Introduction Welcome to Release 2011 of PGI CUDA Fortran. compute-intensive programs. 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 .

2 .

x. each instance will be executed by a different CUDA thread. A call to a kernel specifies how many parallel instances of the kernel must be executed. 64>>>( n. a. a. “Reference. and a local thread index within its thread block. 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. dimension(*) :: x. The CUDA threads are organized into thread blocks. x. y ) real.Chapter 2. Programming Guide This chapter introduces the CUDA programming model through examples written in CUDA Fortran.” on page 13. refer to Chapter 3. dimension(*) :: x. For a reference for CUDA Fortran. y real :: a integer :: n ! call the kernel call ksaxpy<<<n/64. value :: n.y real. and each thread has a global thread block index. a. and a thread index accessed through threadidx. device. value :: a integer. 3 . the call to the kernel ksaxpy specifies n/64 thread blocks. In this example. 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. y ) real. each thread performs one iteration of the common SAXPY loop operation. y ) end subroutine In this case. each with 64 threads. Each thread is assigned a thread block index accessed through the built-in blockidx variable. 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 kernel is defined using the attributes(global) specifier on the subroutine statement. Such a subroutine is called a device kernel or kernel.

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

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

4.8 1. 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. the variable will be allocated in the normal host paged memory. A pinned variable must be an allocatable array.2. When a pinned variable is allocated.2. and may only be accessed within that subprogram. Attributes(shared) A variable with the shared attributed is called a device shared variable or a shared variable. Device constant data may not be assigned or modified in any device subprogram. if the allocation in page-locked memory fails. an allocatable array.4. though a write in one thread is only guaranteed to be visible to other threads after the next call to the SYNCTHREADS() intrinsic. Device constant variables are allocated in the device constant memory space. Device constant variables may not be allocatable. but may be modified in host subprograms. or by other device subprograms to which it is passed as an argument. If declared in a module. or size of page-locked memory. availability. or an assumed-shape dummy array. Datatypes in Device Subprograms The following intrinsic datatypes are allowed in device subprograms and device data: Table 2. the variable may be accessed by that subprogram or subprograms contained in that subprogram.Datatypes in Device Subprograms • If declared in a host subprogram.8 4. An allocatable device variable has a dynamic lifetime. Other device variables have a lifetime of the entire application. Some operating systems or installations may restrict the use. and has a lifetime of the thread block. and have a lifetime of the entire application.1. A shared variable is allocated in the device shared memory for a thread block. it will be allocated in host pagelocked memory. A shared variable may only be declared in a device subprogram. Attributes(pinned) A variable with the pinned attributes is called a pinned variable. the variable may be accessed by any subprogram in that module and by any subprogram that uses the module. A device array may be an explicit-shape array. Attributes(constant) A variable with the constant attributes is called a device constant variable. from when it is allocated until it is deallocated.8 equivalent to real(kind=8) 6 . Intrinsic Datatypes Type integer logical real double precision Type Kind 1. It can be read or written by all threads in the block. A shared variable may not be data initialized.

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

and will only be initiated when all previous operations on that queue have been completed. Kernel Loop Directive CUDA Fortran allows automatic kernel generation and invocation from a region of host code containing one or more tightly nested loops. • The value of streamid must be an integer greater than or equal to zero. and may execute concurrently with each other. the host program should synchronize with the device to avoid erroneous results. As with any kernel. across the thread blocks and grid. and no subsequent operations on any queue begin until the stream zero operation is complete. used when no stream identifier is specified. For these operations. it is the programmer's responsibility to ensure that parallel execution is legal and produces the correct answer. the launch is asynchronous.2. Programmers must be careful when using concurrent host and device execution. in cases where the host program reads or modifies device or constant data. including kernel execution and data copies to and from device memory. An application can manage more concurrency by using multiple streams.Asynchronous concurrent execution assumed-size shared variables in the thread block. If not specified. the compiler handles the generation of the final reduction kernel. Asynchronous concurrent execution There are two components to asynchronous concurrent execution with CUDA Fortran. where the launch actually corresponds to placing the kernel on a queue for execution by the device. the host can continue executing. the host program can synchronize and wait for all previously launched or queued kernels. The work in the loops specified by the directive is executed in parallel. see Section 3. inserting synchronization into the kernel as appropriate. operations on different stream queues may execute out-of-order with respect to when they were placed on the queues. the call actually returns to the host program before the kernel subroutine begins execution. including calling or queueing more kernels for execution on the device. 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. are implemented using stream queues. its value is treated as zero. The default stream. An operation is placed at the end of the stream queue. The one exception to this rule is a scalar reduction operation. The call can be treated as a kernel launch operation. Concurrent Stream Execution Operations involving the device. such as summing the values in a vector or matrix. stream zero is special in that operations on the stream zero queue will begin only after all preceding operations on all queues are complete. In this way. Each user-created stream manages its own queue. The program can use cudaThreadSynchronize() or CUDA Events to wait for the completion of the kernel. it specifies the stream to which this call is associated. By calling the runtime routine cudaThreadSynchronize. is stream zero. Concurrent Host and Device Execution When a host subprogram calls a kernel subroutine. 8 .3.

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

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

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

12 .

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

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

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

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

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

or if it is exhausted. device :: b(:) allocate(b(5024). allocatable. Fortran interfaces to the CUDA memory management runtime routines are provided. 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. Allocating Pinned Memory Allocatable arrays with the pinned attribute are dynamically allocated using the Allocate statement. the compiler allocates the array in normal paged host memory. These functions return memory which will bypass certain Fortran allocatable properties such as automatic deallocation.stat=istat) … if(allocated(b)) deallocate(b) Scalar variables can be allocated on the device using the Fortran 2003 allocatable scalar feature. The compiler will generate code to allocate the array in host page-locked memory. Otherwise. These arrays will also have a lifetime of the subprogram as is usual with the Fortran language: subroutine vfunc(a.Allocating Device and Pinned Arrays real. and thus the arrays are treated more like C malloc’ed areas. allocatable. real. The cudaMalloc function can be used to allocate single-dimensional arrays of the supported intrinsic datatypes. pinned allocatable arrays work and act like any other allocatable array on the host.n) real. refer to“Memory Management.stat=istat) … if(allocated(p)) deallocate(p) 18 . If no such memory space is available. allocatable. pinned :: p(:) allocate(p(5000). device :: adev(n) real. declare and initialize the scalar on the host as: integer.” on page 32. device :: v(:) istat = cudaMalloc(v. and cudaFree can be used to free it: real. 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. To use these. if available. allocatable.c. Mixing standard Fortran allocate/deallocate with the runtime Malloc/Free for a given array is not supported. 100) … istat = cudaFree(v) For a complete list of the memory management runtime routines.

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

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

1.2.8 4(default). DB >>> ( A ) The function call will fail if the grid or block arguments are greater than the maximum sizes allowed. They may have any of the intrinsic datatypes in the following table. refer to “Shared data. The system module cudafor includes definitions of the derived type dim3. Reference • bytes is optional.4(default).2. it must be an integer. it must be a scalar integer. defined as type(dim3) integer(kind=4) :: x. and have a value of zero. if present.8 1. For instance. It specifies the stream to which this call is enqueued. constant.5 on page 41.8 1 (default) Additionally.8 equivalent to real(kind=8) 4(default). if present. such as the function arguments and execution configuration arguments. the value zero is used. Device code Datatypes allowed Variables and arrays with the device. If not specified. they may be of derived type. and specifies the number of bytes of shared memory to be allocated for each thread block to use for assumed-size shared memory arrays.z end type 21 . Device Code Intrinsic Datatypes Type integer logical real double precision complex character(len=1) Type Kind 1. or if bytes is greater than the shared memory available. See Section 4. allowing for static shared memory declared in the kernel and for other dedicated uses. or a value returned by a call to cudaStreamCreate.4(default). a kernel subroutine attributes(global) subroutine sub( a ) can be called like: call sub <<< DG. or another allowed derived type.Chapter 3. For more information. are limited to the types described in this section. or shared attributes. • stream is optional. Table 3. or declared in device subprograms. where the members of the derived type have one of the allowed intrinsic datatypes.” on page 16.y.

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

complex real real.4. complex integer. Reference Fortran Mathematical Intrinsics Table 3.Chapter 3. complex real real Fortran Numeric Inquiry Intrinsics Table 3. complex integer (integer. real.real) real. complex real real. Fortran Mathematical Intrinsics Name acos asin atan atan2 cos cosh exp Argument Datatypes real real real (real. complex real real. 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 .3.5.integer) real Fortran Bit Manipulation Intrinsics Table 3. complex Name log log10 sin sinh sqrt tan tanh Argument Datatypes real. Fortran Numeric Inquiry Intrinsics Name bit_size digits epsilon huge maxexponent minexponent Argument Datatypes integer integer. real integer. real real integer. real real real Name precision radix range selected_int_kind selected_real_kind tiny Argument Datatypes real.

Fortran Reduction Intrinsics Name all any count maxloc maxval Argument Datatypes logical logical logical integer. it has no arguments: void syncthreads() Sometimes threads within a block access the same addresses in shared or global memory. thus creating potential read-after-write. use syncthreads()to specify synchronization points in the kernel. real Name minloc minval product sum Argument Datatypes integer.7.6. real. real integer. refer to “Thread Management. real. To avoid these potential issues. real integer. syncthreads syncthreads_count syncthreads_and syncthread_or threadfence threadfence_block threadfence_system For detailed information on these functions. Synchronization Functions The synchronization functions control the synchronization of various threads during execution of thread blocks.Device code Fortran Reduction Intrinsics Table 3. This 24 .” on page 32. complex Fortran Random Number Intrinsics Table 3. 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. real integer. write-after-read. complex integer. . or write-after-write hazards for some of these memory accesses. SYNCTHREADS The syncthreads intrinsic subroutine acts as a barrier synchronization for all threads in a single thread block.

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

Device code threadfence() is one method to enforce a specific order.true.” on page 32. as described in “cudaGetDeviceProperties.true. THREADFENCE_SYSTEM void threadfence_system() threadfence_system acts as a memory fence.true. 26 . only if the value of the argument is .2 and higher. 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. The clock frequency can be determined by calling cudaGetDeviceProperties. GPU_TIME The gpu_time intrinsic returns the value of the clock cycle counter on the GPU.0 or higher. The value of the function is . 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. for all threads in the warp. Each of these functions has a single argument. 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. Warp-Vote Operations Warp-vote operations are only supported by devices with compute capability 1. 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. ALLTHREADS The allthreads function is a warp-vote operation with a single scalar logical argument: if( allthreads(a(i)<0. The function allthreads evaluates its argument for all threads in the current warp.0) ) allneg = .

Arithmetic and Bitwise Atomic Functions Function atomicadd( mem.1 and higher.2 or higher is required if the first argument has the shared attribute.false. The function anythread evaluates its argument for all threads in the current warp.8.0) ) allneg = . which must be a variable or array element in shared memory (with the shared attribute) or in device global memory (with the device attribute). for all threads in the warp.0. 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. This function is only supported by devices of compute capability 2. value ) atomicsub( mem. and store the combined value back to the first argument location. Both arguments must be of type integer(kind=4). Note The return value for each of these functions is the first argument. Atomic Functions The atomic functions read and write the value of their first operand. They also combine that value with the value of the second argument. Reference ANYTHREAD The anythread function is a warp-vote operation with a single scalar logical argument: if( anythread(a(i)<0. only if the value of the argument is . Arithmetic and Bitwise Atomic Functions These atomic functions read and return the value of the first argument. The value of the function is . These functions are: Table 3.false. Atomic functions are only supported by devices with compute capability 1. value ) Additional Atomic Update mem = mem + value mem = mem – value 27 .true.Chapter 3. Compute capability 1. mem. depending on the function. Example: if( ballot(int_value) ) allneg = . 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.true.

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

integer(kind=4) :: i. OPEN. • Fortran intrinsic functions not listed in Section 3. returns the number of bytes required to hold the value of A. • Subroutine and function calls are supported only if they can be inlined. character substrings are not supported.comp. PRINT. Compare and Swap Atomic Function Function atomiccas(mem. where A is a variable or expression. BACKSPACE. Host code Host subprograms may use intrinsic functions. • Recursive subroutines and functions are not allowed.3 are not supported. • Objects with the Pointer and Allocatable attribute are not allowed.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. • Cray pointers are not supported. WRITE.6. ENDFILE. • Input/Output statements are not allowed: READ. such as the new sizeof intrinsic function. j j = sizeof(i) ! this assigns the value 4 to j 29 . SIZEOF Intrinsic A call to sizeof(A). Reference Table 3. • Optional arguments are not allowed. NAMELIST. REWIND. • Floating point exception handling is not supported. INQUIRE.Chapter 3. • Objects with character type must have LEN=1. CLOSE. FORMAT. • STOP and PAUSE statements are not allowed. • Alternate return specifications are not allowed. • ENTRY statements are not allowed. • Automatic arrays must be fixed size.10.

30 .

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

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

should be of benefit. an allocatable device array as the second argument. devptr may be any allocatable. cudaMallocPitch may pad the data. cudaMallocPitch integer function cudaMallocPitch(devptr. Similarly. devptr may also be of TYPE(C_DEVPTR). c_f_pointer will take a TYPE(C_DEVPTR) as the first argument. height) cudaMallocPitch allocates data on the device. cudaFreeArray integer function cudaFreeArray(carray) type(cudaArrayPtr) :: carray 33 . width. cdesc. as well as consistency checks between Fortran device arrays and host arrays. cudaFree integer function cudaFree(devptr) cudaFree deallocates data on the device. in which case the integer values are expressed in bytes. count) cudaMalloc allocates data on the device. devptr may be of TYPE(C_DEVPTR). cudaMalloc integer function cudaMalloc(devptr. Or. 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. Both of these features are subject to change when.1. “Device Code Intrinsic Datatypes. “Device Code Intrinsic Datatypes. The height is an integer.Chapter 4.1. one-dimensional device array of a supported type specified in Table 3. The count is in terms of elements. a shape as the third argument. width.1. Runtime APIs TYPE(C_DEVPTR). pitch. height cudaMallocArray allocates a data array on the device. height) type(cudaArrayPtr) :: carray type(cudaChannelFormatDesc) :: cdesc integer :: width. devptr may be any allocatable. devptr may be of TYPE(C_DEVPTR). 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. and the padded width is returned in the variable pitch. and in effect transfer the allocation to the Fortran array. two-dimensional device array of a supported type specified in Table 3. devptr may be any allocatable device array of a supported type specified in Table 3. Or. proper Fortran pointers for device data are supported. “Device Code Intrinsic Datatypes.” on page 21. Use the functions in this section for memory management. in the future.” on page 21. Currently. cudaMallocArray integer function cudaMallocArray(carray. The width is in terms of number of elements. Under CUDA Fortran.” on page 21. in which case the count is in bytes.

and the lowest byte of value is used. otherwise the stream argument is optional and defaults to zero. src. devptr may be of TYPE(C_DEVPTR). “Device Code Intrinsic Datatypes. cudaMemcpy2D integer function cudaMemcpy2D(dst. and height are in terms of bytes.” on page 20.1. count.” on page 20. refer to “Data Transfer Using Runtime Routines. The count is in terms of elements. src. cudaMemcpyAsync integer function cudaMemcpyAsync(dst. Alternatively. Or. refer to “Data Transfer Using Runtime Routines. value.” on page 21. of a supported type specified in Table 3. scalar or array. cudaMemset2D integer function cudaMemset2D(devptr. dst and src may be any device or host. Alternatively. The pitch. devptr may be any device array of a supported type specified in Table 3. or cudaMemcpyDeviceToDevice.1. cudaMemset integer function cudaMemset(devptr.1. The copy can be associated with a stream by passing a non-zero stream argument. kdir) cudaMemcpy copies data from one location to another. width. value. and the lowest byte of value is used.1. width. in which case the count is in term of bytes. devptr may be of TYPE(C_DEVPTR).” on page 21. height. in which case the pitch. count. scalar or array.Memory Management cudaFreeArray frees an array that was allocated on the device. “Device Code Intrinsic Datatypes. dst and src may be of TYPE(C_DEVPTR) or TYPE(C_PTR). dst and src may be any device or host. If kdir is specified. “Device Code Intrinsic Datatypes. spitch. dpitch. devptr may be any device scalar or array of a supported type specified in Table 3. kdir.” on page 21. it must be one of the defined enums cudaMemcpyHostToDevice. height) cudaMemset2D sets an array to the specified value. src. cudaMemcpyDeviceToHost. or cudaMemcpyDeviceToDevice. width. for more information. stream) cudaMemcpy copies data from one location to another. The count is in terms of elements. it must be one of the defined enums cudaMemcpyHostToDevice. kdir) cudaMemcpy2D copies data from one location to another. dst and src may be any device or host array. count) cudaMemset sets a location or array to the specified value. The count is in terms of elements.1. pitch. The width 34 .” on page 21. and height are in terms of elements. The value must match in type and kind. cudaMemcpyDeviceToHost. in which case the count is in term of bytes. Or. If kdir is specified. of a supported type specified in Table 3. cudaMemcpy integer function cudaMemcpy(dst. This function operates on page-locked host memory only. in which case the count is in term of bytes. The value must match in type and kind. width. kdir may be optional. kdir may be optional.” on page 21. “Device Code Intrinsic Datatypes. dst and src may be of TYPE(C_DEVPTR) or TYPE(C_PTR). for more information. of a supported type specified in Table 3. “Device Code Intrinsic Datatypes.

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

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

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

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

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

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

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

42 .

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

Bdev.16 Cij = Cij + Asub(tx. 2 ) ! Allocate the device arrays allocate( Adev(N.ty) = A(i.L) ) ! Copy A and B to the device Adev = A(1:N.1:L) = Cdev deallocate( Adev.j) do k = 1.kb+ty-1) Bsub(tx. allocatable.:) :: A.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. 1 ) M = size( A. Cdev.L).dimBlock>>>( Adev. M. C ) real.Matrix Multiplication Example real :: Cij ! Get the thread indices tx = threadidx%x ty = threadidx%y ! This thread computes C(i. Bdev.j)) i = (blockidx%x-1) * 16 + tx j = (blockidx%y-1) * 16 + ty Cij = 0.k) * Bsub(k. C ! allocatable device arrays real. L/16.Cdev ! dim3 variables to define the grid and block shapes type(dim3) :: dimGrid. 1 ) dimBlock = dim3( 16. B. dimension(:.j) = Cij end subroutine mmul_kernel ! The host routine to drive the matrix multiplication subroutine mmul( A. device.0 ! Do the k loop in chunks of 16. the block size do kb = 1. Cdev ) end subroutine mmul end module mmul_mod N 44 .Bdev.:) = B(1:M.:) :: Adev.j) = sum(A(i. 16 ! Fill the submatrices ! Each of the 16x16 threads in the thread block ! loads one element of Asub and Bsub Asub(tx. 16. dimension(:. dimBlock ! Get the array sizes N = size( A. Cdev(N.1:L) ! Create the grid and block dimensions dimGrid = dim3( N/16. 2 ) L = size( B.ty) = B(kb+tx-1. B. ! Copy the results back and free up memory C(1:N.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. Bdev(M. 1 ) call mmul_kernel<<<dimGrid.M).1:M) Bdev(:.:) * B(:.

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

46 .

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

48 .

Sign up to vote on this title
UsefulNot useful