{"id":128,"date":"2010-07-31T13:14:22","date_gmt":"2010-07-31T13:14:22","guid":{"rendered":"http:\/\/sugarandcyanide.com\/blog\/?p=128"},"modified":"2010-07-31T13:14:22","modified_gmt":"2010-07-31T13:14:22","slug":"first-steps-in-opencl","status":"publish","type":"post","link":"http:\/\/sugarandcyanide.com\/blog\/2010\/07\/31\/first-steps-in-opencl\/","title":{"rendered":"first steps in openCL"},"content":{"rendered":"

I’m going to post something about my trip to Italy in a bit, after I get my pics uploaded properly to Flickr. But I wanted to put this up so I don’t forget about it later.<\/p>\n

A conversation at work got me interested enough in GPGPU calculations to start doing a bit of research into the topic. Seems that at a base level both the ATI and NVidia APIs behave similarly, and that OpenCL is a nice, understandable alternative if you don’t want to be locked in to a single vendor. (Not that I use ATI cards.) I tried getting CUDA working on my Mac, but between weirdness caused by the drivers and the fact that it doesn’t support 64-bit on OSX, I’ve decided to go with OpenCL<\/a> instead for my tests. Functionally everything will be the same, and after a nice tutorial on ATI’s site detailing some introductory points<\/a> I have code that compiles and does what I want it to do, which is more than I can say for CUDA.<\/p>\n

Anyway, here’s my first test. I wanted something simple for getting device information off my laptop.<\/p>\n

#include <iostream><\/span>
\nusing<\/span> namespace<\/span> std;<\/span><\/p>\n

#include <stdio.h><\/span>
\n#include <stdlib.h><\/span><\/p>\n

#include <OpenGL\/gl.h><\/span>
\n#include <OpenGL\/CGLDevice.h><\/span>
\n#include <OpenCL\/cl.h><\/span>
\n#include <OpenCL\/cl_gl.h><\/span>
\n#include <OpenCL\/cl_gl_ext.h><\/span>
\n#include <OpenCL\/cl_ext.h><\/span>
\n 
\n#define MSG_SIZE 4096<\/span>
\n#define MAX_DEVICE_IDS 16<\/span><\/p>\n

#define RT_STRING       0<\/span>
\n#define RT_UINT         1<\/span>
\n#define RT_BOOL         2<\/span>
\n#define RT_PLATFORMID   3<\/span>
\n#define RT_SIZET        5<\/span>
\n#define RT_SIZET_ARR    6<\/span><\/p>\n

\nvoid<\/span> printDeviceInfo(<\/span>cl_device_id deviceID)<\/span> {<\/span>
\n   
\n    cl_device_info param_list[<\/span>]<\/span> =<\/span> {<\/span>
\n        CL_DEVICE_NAME, CL_DEVICE_VENDOR, CL_DRIVER_VERSION, CL_DEVICE_VERSION,
\n        CL_DEVICE_PROFILE,
\n        CL_DEVICE_PLATFORM, CL_DEVICE_VENDOR_ID,
\n        CL_DEVICE_MAX_COMPUTE_UNITS, CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS,
\n        CL_DEVICE_MAX_WORK_ITEM_SIZES, CL_DEVICE_MAX_WORK_GROUP_SIZE,
\n       
\n        NULL<\/span>
\n    }<\/span>;<\/span>
\n   
\n    int<\/span> param_return_types[<\/span>]<\/span> =<\/span> {<\/span>
\n        RT_STRING, RT_STRING, RT_STRING, RT_STRING,
\n        RT_STRING, RT_PLATFORMID, RT_UINT,
\n        RT_UINT, RT_UINT, RT_SIZET_ARR, RT_SIZET,
\n       
\n        NULL<\/span>
\n    }<\/span>;<\/span>
\n   
\n    char<\/span> *<\/span>param_strings[<\/span>]<\/span> =<\/span> {<\/span>
\n        "Device Name"<\/span>, "Vendor"<\/span>, "Driver Version"<\/span>, "Device Version"<\/span>,
\n        "Device Profile"<\/span>, "Device Platform"<\/span>, "Device Vendor ID"<\/span>,
\n        "Max Compute Units"<\/span>, "Max Work Item Dimensions"<\/span>,
\n        "Max Work Item Sizes"<\/span>, "Max Work Group Size"<\/span>,
\n   
\n        NULL<\/span>
\n    }<\/span>;<\/span>
\n   
\n    int<\/span> i =<\/span> 0<\/span>;<\/span>
\n    int<\/span> maxWorkItemDimensions =<\/span> 3<\/span>;<\/span>
\n   
\n    while<\/span>(<\/span>param_strings[<\/span>i]<\/span> !<\/span>=<\/span> NULL<\/span>)<\/span> {<\/span>
\n        char<\/span>    msg[<\/span>MSG_SIZE]<\/span>;<\/span>
\n        size_t<\/span>  param_sizeT;<\/span>
\n        size_t<\/span>  param_sizeT_array[<\/span>MSG_SIZE]<\/span>;<\/span>
\n        cl_uint param_uint;<\/span>
\n        cl_bool param_bool;<\/span>
\n        size_t<\/span>  param_value_ret;<\/span>
\n        cl_platform_id param_platformID;<\/span>
\n       
\n        cl_int error;<\/span>
\n       
\n        switch<\/span> (<\/span>param_return_types[<\/span>i]<\/span>)<\/span> {<\/span>
\n            case<\/span> RT_STRING:<\/span>
\n                error =<\/span> clGetDeviceInfo(<\/span>deviceID,
\n                                        param_list[<\/span>i]<\/span>,
\n                                        MSG_SIZE,
\n                                        msg,
\n                                        &<\/span>param_value_ret
\n                                        )<\/span>;<\/span><\/p>\n

                cout<\/span> <<<\/span> param_strings[<\/span>i]<\/span> <<<\/span> ": "<\/span> <<<\/span> msg <<<\/span> "\\n<\/span>"<\/span>;<\/span>
\n                break<\/span>;<\/span><\/p>\n

            case<\/span> RT_UINT:<\/span>
\n                error =<\/span> clGetDeviceInfo(<\/span>deviceID,
\n                                        param_list[<\/span>i]<\/span>,
\n                                        sizeof<\/span>(<\/span>cl_uint)<\/span>,
\n                                        &<\/span>param_uint,
\n                                        NULL<\/span>
\n                                        )<\/span>;<\/span><\/p>\n

                cout<\/span> <<<\/span> param_strings[<\/span>i]<\/span> <<<\/span> ": "<\/span> <<<\/span> param_uint <<<\/span> "\\n<\/span>"<\/span>;<\/span>
\n                break<\/span>;<\/span>
\n               
\n               
\n            case<\/span> RT_BOOL:<\/span>
\n                error =<\/span> clGetDeviceInfo(<\/span>deviceID,
\n                                        param_list[<\/span>i]<\/span>,
\n                                        sizeof<\/span>(<\/span>cl_bool)<\/span>,
\n                                        &<\/span>param_bool,
\n                                        NULL<\/span>
\n                                        )<\/span>;<\/span><\/p>\n

                cout<\/span> <<<\/span> param_strings[<\/span>i]<\/span> <<<\/span> ": "<\/span> <<<\/span> (<\/span>param_bool ?<\/span> "True"<\/span> :<\/span> "False"<\/span>)<\/span> <<<\/span> "\\n<\/span>"<\/span>;<\/span>
\n                break<\/span>;<\/span>
\n               
\n            case<\/span> RT_PLATFORMID:<\/span>
\n                error =<\/span> clGetDeviceInfo(<\/span>deviceID,
\n                                        param_list[<\/span>i]<\/span>,
\n                                        sizeof<\/span>(<\/span>cl_platform_id)<\/span>,
\n                                        param_platformID,
\n                                        NULL<\/span>
\n                                        )<\/span>;<\/span><\/p>\n

                cout<\/span> <<<\/span> param_strings[<\/span>i]<\/span> <<<\/span> ": "<\/span> <<<\/span> param_platformID <<<\/span> "\\n<\/span>"<\/span>;<\/span>
\n                break<\/span>;<\/span>
\n           
\n            case<\/span> RT_SIZET:<\/span>
\n                error =<\/span> clGetDeviceInfo(<\/span>deviceID,
\n                                        param_list[<\/span>i]<\/span>,
\n                                        sizeof<\/span>(<\/span>size_t<\/span>)<\/span>,
\n                                        &<\/span>param_sizeT,
\n                                        NULL<\/span>
\n                                        )<\/span>;<\/span><\/p>\n

                cout<\/span> <<<\/span> param_strings[<\/span>i]<\/span> <<<\/span> ": "<\/span> <<<\/span> param_sizeT <<<\/span> "\\n<\/span>"<\/span>;<\/span>
\n                break<\/span>;<\/span>
\n               
\n            case<\/span> RT_SIZET_ARR:<\/span>
\n                error =<\/span> clGetDeviceInfo(<\/span>deviceID,
\n                                        param_list[<\/span>i]<\/span>,
\n                                        sizeof<\/span>(<\/span>size_t<\/span>)<\/span> *<\/span> maxWorkItemDimensions,
\n                                        param_sizeT_array,
\n                                        &<\/span>param_value_ret
\n                                        )<\/span>;<\/span><\/p>\n

                cout<\/span> <<<\/span> param_strings[<\/span>i]<\/span> <<<\/span> ": "<\/span> <<<\/span> param_sizeT_array[<\/span>0<\/span>]<\/span> <<<\/span> ", "<\/span>
\n                    <<<\/span> param_sizeT_array[<\/span>1<\/span>]<\/span> <<<\/span> ", "<\/span>
\n                    <<<\/span> param_sizeT_array[<\/span>2<\/span>]<\/span> <<<\/span> "\\n<\/span>"<\/span>;<\/span>
\n                break<\/span>;<\/span>
\n               
\n               
\n               
\n            default<\/span>:<\/span>
\n                break<\/span>;<\/span>
\n        }<\/span>
\n       
\n        i++<\/span>;<\/span>
\n    }<\/span>
\n}<\/span><\/p>\n

\nint<\/span> main(<\/span>int<\/span> argc, char<\/span> **<\/span>argv)<\/span> {<\/span>
\n    cl_device_id deviceID;<\/span>
\n    cl_uint numDevices;<\/span><\/p>\n

    cl_uint err =<\/span> clGetDeviceIDs(<\/span>
\n                NULL<\/span>,                  
\n                CL_DEVICE_TYPE_CPU,    
\n                MAX_DEVICE_IDS,        
\n                &<\/span>deviceID,             
\n                &<\/span>numDevices
\n                )<\/span>;<\/span>
\n   
\n    cout<\/span> <<<\/span> "Number of CPU devices: "<\/span> <<<\/span> numDevices <<<\/span> "\\n<\/span>"<\/span>;<\/span>
\n   
\n    for<\/span> (<\/span>int<\/span> i =<\/span> 0<\/span>;<\/span> i <<\/span> numDevices;<\/span> i++<\/span>)<\/span> {<\/span>
\n        printDeviceInfo(<\/span>deviceID)<\/span>;<\/span>
\n    }<\/span><\/p>\n

    cout<\/span> <<<\/span> "\\n<\/span>\\n<\/span>"<\/span>;<\/span><\/p>\n

   
\n    <\/p>\n

    err =<\/span> clGetDeviceIDs(<\/span>
\n                NULL<\/span>,
\n                CL_DEVICE_TYPE_GPU,
\n                MAX_DEVICE_IDS,
\n                &<\/span>deviceID,
\n                &<\/span>numDevices
\n                )<\/span>;<\/span>
\n   
\n    cout<\/span> <<<\/span> "Number of GPU devices: "<\/span> <<<\/span> numDevices <<<\/span> "\\n<\/span>"<\/span>;<\/span>
\n   
\n    for<\/span> (<\/span>int<\/span> i =<\/span> 0<\/span>;<\/span> i <<\/span> numDevices;<\/span> i++<\/span>)<\/span> {<\/span>
\n        printDeviceInfo(<\/span>deviceID)<\/span>;<\/span>
\n    }<\/span>
\n   
\n    cout<\/span> <<<\/span> "\\n<\/span>\\n<\/span>"<\/span>;<\/span>
\n   
\n}<\/span>
\n <\/div>\n<\/p>\n

(Hope that copy-and-pasted correctly.) To compile it on your Mac:<\/p>\n

g++ test2.cpp -o test2 -framework OpenCL<\/div>\n<\/p>\n

Here are the results on my laptop, a 2010 Core i7 MacBook Pro 15″:<\/p>\n

Number of CPU devices: 1\nDevice Name: Intel(R) Core(TM) i7 CPU       M 620  @ 2.67GHz\nVendor: Intel\nDriver Version: 1.0\nDevice Version: OpenCL 1.0\nDevice Profile: FULL_PROFILE\nDevice Platform: 0\nDevice Vendor ID: 16909312\nMax Compute Units: 4\nMax Work Item Dimensions: 3\nMax Work Item Sizes: 1, 1, 1\nMax Work Group Size: 1

\nNumber of GPU devices: 1\nDevice Name: GeForce GT 330M\nVendor: NVIDIA\nDriver Version: CLH 1.0\nDevice Version: OpenCL 1.0\nDevice Profile: FULL_PROFILE\nDevice Platform: 0\nDevice Vendor ID: 16918016\nMax Compute Units: 6\nMax Work Item Dimensions: 3\nMax Work Item Sizes: 512, 512, 64\nMax Work Group Size: 512\n<\/pre>\n
Next step: getting the GPU to do some calculations.<\/p>\n","protected":false},"excerpt":{"rendered":"
I’m going to post something about my trip to Italy in a bit, after I get my pics uploaded properly to Flickr. But I wanted to put this up so I don’t forget about it later. A conversation at work got me interested enough in GPGPU calculations to start doing a bit of research into […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[15],"tags":[30,36],"class_list":["post-128","post","type-post","status-publish","format-standard","hentry","category-programming","tag-gpgpu","tag-opencl"],"_links":{"self":[{"href":"http:\/\/sugarandcyanide.com\/blog\/wp-json\/wp\/v2\/posts\/128","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/sugarandcyanide.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/sugarandcyanide.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/sugarandcyanide.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/sugarandcyanide.com\/blog\/wp-json\/wp\/v2\/comments?post=128"}],"version-history":[{"count":0,"href":"http:\/\/sugarandcyanide.com\/blog\/wp-json\/wp\/v2\/posts\/128\/revisions"}],"wp:attachment":[{"href":"http:\/\/sugarandcyanide.com\/blog\/wp-json\/wp\/v2\/media?parent=128"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/sugarandcyanide.com\/blog\/wp-json\/wp\/v2\/categories?post=128"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/sugarandcyanide.com\/blog\/wp-json\/wp\/v2\/tags?post=128"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}