| 1 | | stub |
| | 1 | = Building BOINC applications with Cuda and OpenCL = |
| | 2 | |
| | 3 | [[PageOutline]] |
| | 4 | |
| | 5 | We have built sample BOINC-Cuda and BOINC-OpenCL applications as templates for making your own app. |
| | 6 | |
| | 7 | == Windows == #windows |
| | 8 | |
| | 9 | |
| | 10 | == Linux == #linux |
| | 11 | |
| | 12 | |
| | 13 | == Mac OS X == #mac |
| | 14 | |
| | 15 | * Cuda (BOINC-Cuda sample app for Mac OS X can be found at "/boinc/samples/nvcuda/") |
| | 16 | {{{ |
| | 17 | Makefile_mac; common_mac.mk; cuda_kernel_mac.cu; cuda_mac.c; readme_mac.txt |
| | 18 | }}} |
| | 19 | |
| | 20 | Unlike Windows and Linux, the 'nvcc' cuda compiler for Mac has some trouble compiling .cu files that |
| | 21 | contain both BOINC and Cuda codes. Any attempt to compile such .cu files might result in errors like |
| | 22 | these: |
| | 23 | {{{ |
| | 24 | /usr/lib/gcc/i686-apple-darwin9/4.0.1/include/mmintrin.h(55): error: identifier "__builtin_ia32_emms" is undefined |
| | 25 | /usr/lib/gcc/i686-apple-darwin9/4.0.1/include/mmintrin.h(68): error: identifier "__builtin_ia32_vec_init_v2si" is undefined |
| | 26 | /usr/lib/gcc/i686-apple-darwin9/4.0.1/include/mmintrin.h(111): error: identifier "__builtin_ia32_vec_ext_v2si" is undefined |
| | 27 | /usr/lib/gcc/i686-apple-darwin9/4.0.1/include/mmintrin.h(150): error: identifier "__builtin_ia32_packsswb" is undefined |
| | 28 | /usr/lib/gcc/i686-apple-darwin9/4.0.1/include/mmintrin.h(165): error: identifier "__builtin_ia32_packssdw" is undefined |
| | 29 | ... |
| | 30 | ... |
| | 31 | }}} |
| | 32 | The solution that we came up with is to compile BOINC and Cuda seperately, meaning that compile .c files that have BOINC codes with |
| | 33 | gcc and compile .u files that have Cuda codes such as kernel definitions with nvcc (This can simply be done by putting .cu and .c |
| | 34 | file names at corresponding "CUFILES := " and "CCFILES := " entries in Makefile_mac). To handle the results computed by the kernels, |
| | 35 | you will need to write external functions. The function definitions that make kernel calls will be put in .cu file while the function headers with "extern" prefix are put in .c file and can be called in main function or elsewhere. |
| | 36 | |
| | 37 | What if your machine doesn't have an NVIDIA Cuda-enabled GPU? Then, you can compile and run your app in emulation mode. As far as we know, NVIDIA Cuda SDK 3.1 no longer supports emulation mode. Thus, you will need to install an older Cuda SDK version, like version 3.0, on your machine. To make an executable file in emulation mode, simply type "make -f Makefile_mac emu=1". |
| | 38 | |
| | 39 | Also, note that you will need to define some environment variables before running the executable file. |
| | 40 | {{{ |
| | 41 | export PATH=/usr/local/cuda/bin:$PATH |
| | 42 | export DYLD_LIBRARY_PATH=/usr/local/cuda/lib:$DYLD_LIBRARY_PATH |
| | 43 | }}} |
| | 44 | |
| | 45 | One last thing to mention, the makefile for Mac is made with the assumption that the NVIDIA Cuda SDK is installed at the default path: "$ROOT/Developer/GPU Computing/" on Mac. If it is installed at a different path on your machine, then you will need to edit "common_mac.mk" file. Do so by looking for "ROOTDIR ?= /DeveloperGPU\ Computing" in "common_mac.mk" file and replace this path with an appropriate path on your machine. |
