Search Results

Search found 21089 results on 844 pages for 'virtual memory'.

Page 373/844 | < Previous Page | 369 370 371 372 373 374 375 376 377 378 379 380  | Next Page >

  • Adding two different Objects by overloading operator+ C++

    - by lampshade
    Hello, I've been trying to figure out how to add a private member from Object A, to a private member from Object B. Both Cat and Dog Class's inheriate from the base class Animal. I have a thrid class 'MyClass', that I want to inheriate the private members of the Cat and Dog class. So in MyClass, I have a friend function to overload the + operator. THe friend function is defined as follows: MyClass operator+(const Dog &dObj, const Cat &cObj); I want to access dObj.age and cObj.age within the above function, invoke by this statement in main: mObj = dObj + cObj; Here is the entire source for a complete reference into the class objects: #include <iostream> #include <vld.h> using namespace std; class Animal { public : Animal() {}; virtual void eat() = 0 {}; virtual void walk() = 0 {}; }; class Dog : public Animal { public : Dog(const char * name, const char * gender, int age); Dog() : name(NULL), gender(NULL), age(0) {}; virtual ~Dog(); void eat(); void bark(); void walk(); private : char * name; char * gender; int age; }; class Cat : public Animal { public : Cat(const char * name, const char * gender, int age); Cat() : name(NULL), gender(NULL), age(0) {}; virtual ~Cat(); void eat(); void meow(); void walk(); private : char * name; char * gender; int age; }; class MyClass : private Cat, private Dog { public : MyClass() : action(NULL) {}; void setInstance(Animal &newInstance); void doSomething(); friend MyClass operator+(const Dog &dObj, const Cat &cObj); private : Animal * action; }; Cat::Cat(const char * name, const char * gender, int age) : name(new char[strlen(name)+1]), gender(new char[strlen(gender)+1]), age(age) { if (name) { size_t length = strlen(name) +1; strcpy_s(this->name, length, name); } else name = NULL; if (gender) { size_t length = strlen(gender) +1; strcpy_s(this->gender, length, gender); } else gender = NULL; if (age) { this->age = age; } } Cat::~Cat() { delete name; delete gender; age = 0; } void Cat::walk() { cout << name << " is walking now.. " << endl; } void Cat::eat() { cout << name << " is eating now.. " << endl; } void Cat::meow() { cout << name << " says meow.. " << endl; } Dog::Dog(const char * name, const char * gender, int age) : name(new char[strlen(name)+1]), gender(new char[strlen(gender)+1]), age(age) { if (name) { size_t length = strlen(name) +1; strcpy_s(this->name, length, name); } else name = NULL; if (gender) { size_t length = strlen(gender) +1; strcpy_s(this->gender, length, gender); } else gender = NULL; if (age) { this->age = age; } } Dog::~Dog() { delete name; delete gender; age = 0; } void Dog::eat() { cout << name << " is eating now.. " << endl; } void Dog::bark() { cout << name << " says woof.. " << endl; } void Dog::walk() { cout << name << " is walking now.." << endl; } void MyClass::setInstance(Animal &newInstance) { action = &newInstance; } void MyClass::doSomething() { action->walk(); action->eat(); } MyClass operator+(const Dog &dObj, const Cat &cObj) { MyClass A; //dObj.age; //cObj.age; return A; } int main() { MyClass mObj; Dog dObj("B", "Male", 4); Cat cObj("C", "Female", 5); mObj.setInstance(dObj); // set the instance specific to the object. mObj.doSomething(); // something happens based on which object is passed in dObj.bark(); mObj.setInstance(cObj); mObj.doSomething(); cObj.meow(); mObj = dObj + cObj; return 0; }

    Read the article

  • If terminating a hung thread is a good idea, how do I do it safely?

    - by Steve
    My Delphi program relies heavily on Outlook automation. Outlook versions prior to 2007-SP2 tend to get stuck in memory due to badly written addins and badly written Outlook code. If Outlook is stuck, calling CreateOleObject('Outlook.Application') or GetActiveObject ... doesn't return and keeps my application hanging till Outlook.exe is closed in the task manager. I've thought of a solution, but I'm unsure whether it's good practice or not. I'd start Outlook with CreateOleObject in a separate thread, wait 10 seconds in my main thread and if Outlook hangs (CreateOleObject doesn't return), offer the user to kill the Outlook.exe process from my program. But since I don't want to force the user to kill the Outlook.exe process, as an alternative I also need a way to kill the new thread in my program which keeps hanging now. Is this good practice? How can I terminate a hanging thread in Delphi without leaking memory?

    Read the article

  • Flashing and closing PyS60 SIS

    - by Matrich
    I created a py script for S60 2nd Edition FP3 with a background image, database, sound files which are stored in a folder on c drive. After several hours, I have managed to convert it into sis which I sent to my mobile phone. First, I installed the sis on the external memory and it failed to work. I read online that it should be on the same drive as python runtime. So I installed it on phone memory but it just flashes and then closes. I made another sis with a database and it worked so I suspect it could be because it cant find the background image and sound files in the folder. How can I know the cause of this? Also, how can I include the background image and sound files with the sis so that they are automatically installed when installing the sis file?

    Read the article

  • Is there a way to change the maximum width of a window without using the WM_GETMINMAXINFO message?

    - by David
    I want to change the imposed Windows maximum width that a window can be resized to, for an external application's window (not my C#/WinForms program's window). The documentation of GetSystemMetrics for SM_CXMAXTRACK says: "The default maximum width of a window that has a caption and sizing borders, in pixels. This metric refers to the entire desktop. The user cannot drag the window frame to a size larger than these dimensions. A window can override this value by processing the WM_GETMINMAXINFO message." Is there a way to modify this SM_CXMAXTRACK value (either system wide or for one particular window), without processing the WM_GETMINMAXINFO message? Maybe an undocumented function, a registry setting, etc.? (Or: The documentation for MINMAXINFO.ptMaxTrackSize says: "This value is based on the size of the virtual screen and can be obtained programmatically from the system metrics SM_CXMAXTRACK and SM_CYMAXTRACK." Maybe there is a way to change the size of the virtual screen?) Thank you

    Read the article

  • pushScene and popScene or replaceScene . which should we use and when ?

    - by srikanth rongali
    I am using push scene to get the next scene. But, I read that for each PushScene the scene is stored in stack. The memory usage is more. So, I am using the replaceScene in place of pushScene. But, with replace scene I am getting the memory-bad-access message in debugger. So, I want to popScene after using it, so that the retain count is zero. But, I am confused in using popScene. If I have a Scene1 and Scene2. I used the following to go in to Scene2. Now I need to remove Scene1 from stack. [[CCDirector sharedDirector] pushScene:Scene2]; Where should I write the popScene to popScene1. How to get the previous scene in current running scene ? Thank you/

    Read the article

  • Cleaning up a dynamic array of Objects in C++

    - by Dr. Monkey
    I'm a bit confused about handling an array of objects in C++, as I can't seem to find information about how they are passed around (reference or value) and how they are stored in an array. I would expect an array of objects to be an array of pointers to that object type, but I haven't found this written anywhere. Would they be pointers, or would the objects themselves be laid out in memory in an array? In the example below, a custom class myClass holds a string (would this make it of variable size, or does the string object hold a pointer to a string and therefore take up a consistent amount of space. I try to create a dynamic array of myClass objects within a myContainer. In the myContainer.addObject() method I attempt to make a bigger array, copy all the objects into it along with a new object, then delete the old one. I'm not at all confident that I'm cleaning up my memory properly with my destructors - what improvements could I make in this area? class myClass { private string myName; public unsigned short myAmount; myClass(string name, unsigned short amount) { myName = name; myAmount = amount; } //Do I need a destructor here? I don't think so because I don't do any // dynamic memory allocation within this class } class myContainer { int numObjects; myClass * myObjects; myContainer() { numObjects = 0; } ~myContainer() { //Is this sufficient? //Or do I need to iterate through myObjects and delete each // individually? delete [] myObjects; } void addObject(string name, unsigned short amount) { myClass newObject = new myClass(name, amount); myClass * tempObjects; tempObjects = new myClass[numObjects+1]; for (int i=0; i<numObjects; i++) tempObjects[i] = myObjects[i]); tempObjects[numObjects] = newObject; numObjects++; delete newObject; //Will this delete all my objects? I think it won't. //I'm just trying to delete the old array, and have the new array hold // all the objects plus the new object. delete [] myObjects; myObjects = tempObjects; } }

    Read the article

  • How do I use Eval() to reference values in a SortedDictionary in an asp Repeater?

    - by MatthewMartin
    I thought I was clever to switch from the memory intensive DataView to SortedDictionary as a memory efficient sortable data structure. Now I have no idea how get the key and value out of the datasource in the <%# or Eval() expressions. SortedDictionary<int, string> data = RetrieveNames(); rCurrentTeam.DataSource = data; rCurrentTeam.DataBind(); <asp:Repeater ID="rNames" runat="server"> <ItemTemplate> <asp:Label ID="lblName" runat="server" Text='<%# Eval("what?") %>' /> </ItemTemplate> </asp:Repeater> Any suggestions?

    Read the article

  • Size of Objects in Java Heap w/ Regards to Methods

    - by Eric
    I know about primitives and objects living on the heap, but how does the number of methods effect heap size of the object? For example: public class A { int x; public getX() { return x; } } public class B { int x; public getX() { return x; } public getXString() { return String.valueOf(x); } public doMoreInterestingStuff() { return x * 42; } //etc } When instantiated, both objects live on the heap, both have memory allocated to their primitive x, but is B allocated more heap space due to having more method signatures? Or are those ONLY on the classLoader? In this example its trivial, but when there are 100,000+ of these objects in memory at any given time I imagine it could add up.

    Read the article

  • How to specify execution time of x86 and PowerPC instructions?

    - by Goofy
    Hello! I have to approximate execution time of PowerPC and x86 assembler code.I understand that I cannot compute exact it dependson many problems (current processor state - x86 processor dicides internal instructions in microinstructions, memory access time obtainig code from cache of from slower memory etc.). I found some information in Intel Optimizaction reference (APPENDIX C), but it does not provide information about all general purpose instructions. Is there any complete reference about it? What about PowerPC processors? Where can I fund such information?

    Read the article

  • Returning a C++ reference in a const member functionasses

    - by Chris Kaminski
    A have a class hierarchy that looks somethign like this: class AbstractDataType { public: virtual int getInfo() = 0; }; class DataType: public AbstractDataType { public: virtual int getInfo() { }; } class Accessor { DataType data; public: const AbstractDataType& getData() const { return(data); } } Well, GCC 4.4 reports: In member function ‘const AbstractDataType& Accessor::getData() const’: error: invalid initialization of reference of type ‘const AbstractDataType&’ from expression of type ‘const DataType’ Where am I going wrong - is this a case where I MUST use a pointer?

    Read the article

  • How does jQuery store data with .data()?

    - by TK
    I am a little confused how jQuery stores data with .data() functions. Is this something called expando? Or is this using HTML5 Web Storage although I think this is very unlikely? The documentation says: The .data() method allows us to attach data of any type to DOM elements in a way that is safe from circular references and therefore from memory leaks. As I read about expando, it seems to have a rick of memory leak. Unfortunately my skills are not enough to read and understand jQuery code itself, but I want to know how jQuery stores such data by using data(). http://api.jquery.com/data/

    Read the article

  • C++ Class Inheritance architecture - preventing casting

    - by Some One
    I have a structure of base class and a couple of inherited classed. Base class should be pure virtual class, it should prevent instantiation. Inherited classes can be instantiated. Code example below: class BaseClass { public: BaseClass(void); virtual ~BaseClass(void) = 0; }; class InheritedClass : public BaseClass { public: InheritedClass1(void); ~InheritedClass1(void); }; class DifferentInheritedClass : public BaseClass { public: DifferentInheritedClass(void); ~DifferentInheritedClass(void); }; I want to prevent the following operations to happen: InheritedClass *inherited1 = new InheritedClass(); DifferentInheritedClass *inherited2 = new DifferentInheritedClass (); BaseClass *base_1 = inherited1; BaseClass *base_2 = inherited2; *base_1 = *base_2;

    Read the article

  • SDL_BlitSurface() not displaying image?

    - by Christian Gonzalez
    So I'm trying to display a simply image with the SDL library, but when I use the function SDL_BlitSurface() nothing happens, and all I get is a black screen. I should also note that I have the .bmp file, the source, and the executable file all in the same directory. //SDL Header #include "SDL/SDL.h" int main(int argc, char* args[]) { //Starts SDL SDL_Init(SDL_INIT_EVERYTHING); //SDL Surfaces are images that are going to be displayed. SDL_Surface* Hello = NULL; SDL_Surface* Screen = NULL; //Sets the size of the window (Length, Height, Color(bits), Sets the Surface in Software Memory) Screen = SDL_SetVideoMode(640, 480, 32, SDL_SWSURFACE); //Loads a .bmp image Hello = SDL_LoadBMP("Hello.bmp"); //Applies the loaded image to the screen SDL_BlitSurface(Hello, NULL, Screen, NULL); //Update Screen SDL_Flip(Screen); //Pause SDL_Delay(2000); //Deletes the loaded image from memory SDL_FreeSurface(Hello); //Quits SDL SDL_Quit(); return 0; }

    Read the article

  • C++: warning: '...' declared with greater visibility than the type of its field '...::<anonymous>'

    - by Albert
    I'm getting these two warnings (with GCC 4.2 on MacOSX): /Users/az/Programmierung/openlierox/build/Xcode/../../src/main.cpp:154:0 /Users/az/Programmierung/openlierox/build/Xcode/../../src/main.cpp:154: warning: 'startMainLockDetector()::MainLockDetector' declared with greater visibility than the type of its field 'startMainLockDetector()::MainLockDetector::<anonymous' /Users/az/Programmierung/openlierox/build/Xcode/../../src/main.cpp:154:0 /Users/az/Programmierung/openlierox/build/Xcode/../../src/main.cpp:154: warning: 'startMainLockDetector()::MainLockDetector' declared with greater visibility than its base 'Action' In this code: struct Action { virtual ~Action() {} virtual int handle() = 0; }; static void startMainLockDetector() { /* ... */ struct MainLockDetector : Action { bool wait(Uint32 time) { /* ... */ } int handle() { /* ... */ } }; /* ... */ } I'm not exactly sure what these warnings mean (what visibility?) and how to fix them. (I really want the class MainLockDetector to be local for that function only.) I have already compiled the same code with a lot of other compilers (clang, GCC 3.*, GCC 4.0, GCC 4.4, etc) and never got any warning for this code.

    Read the article

  • C# SendMessage to C++ WinProc

    - by jws
    I need to send a string from C# to a C++ WindowProc. There are a number of related questions on SO related to this, but none of the answers have worked for me. Here's the situation: PInvoke: [DllImport("user32", CharSet = CharSet.Auto)] public extern static int SendMessage(IntPtr hWnd, uint wMsg, IntPtr wParam, string lParam); C#: string lparam = "abc"; NativeMethods.User32.SendMessage(handle, ConnectMsg, IntPtr.Zero, lparam); C++: API LRESULT CALLBACK HookProc (int code, WPARAM wParam, LPARAM lParam) { if (code >= 0) { CWPSTRUCT* cwp = (CWPSTRUCT*)lParam; ... (LPWSTR)cwp->lParam <-- BadPtr ... } return ::CallNextHookEx(0, code, wParam, lParam); } I've tried a number of different things, Marshalling the string as LPStr, LPWStr, also tried creating an IntPtr from unmanaged memory, and writing to it with Marshal.WriteByte. The pointer is the correct memory location on the C++ side, but the data isn't there. What am I missing?

    Read the article

  • Oracle Professional Developer, Ver. 6.0.33 for MS-DOS

    - by Frank Developer
    Hi, I'm trying to install this fossil in DOS 6.22, which is running in Microsoft Virtual PC 2007. I'm hoping that some of you veteran Oracle developers have any recollection of in what order do these diskette images need to be loaded, since I don't have any documentation available for 6.0.33. A friend emailed me the following 720K diskette images: INSTALLE.IMG RDBMS1.IMG RDBMS2.IMG RDBMS3.IMG RSF.IMG SQLLDR1.IMG SQLLDR2.IMG UTIL1.IMG UTIL2.IMG UTIL3.IMG Within Virtual PC, I capture a diskette image, one at a time, into Drive A: VPC's floppy disk emulation supports FAT12, so the supported diskette formats can be: 720K or 1.44MB. If anyone happens to have documentation for this legacy version, I would greatly appreciate a copy of it! Thank You.

    Read the article

  • android set hidden the keybord on press enter (in a EditText)

    - by user119333
    Hi, When my user press "ENTER" on the virtual android "user validate entry!" keybord my keybord stay visible ! (Why?) Here my Java code... private void initTextField() { entryUser = (EditText) findViewById(R.id.studentEntrySalary); entryUser.setOnKeyListener(new OnKeyListener() { public boolean onKey(View v, int keyCode, KeyEvent event) { if (event.getAction() == KeyEvent.ACTION_DOWN) { switch (keyCode) { case KeyEvent.KEYCODE_DPAD_CENTER: case KeyEvent.KEYCODE_ENTER: userValidateEntry(); return true; } } return true; } }); } private void userValidateEntry() { System.out.println("user validate entry!"); } ... here my View <LinearLayout xmlns:android="http://schemas.android.com/apk/res/android" android:layout_width="wrap_content" android:layout_height="wrap_content"> <EditText android:id="@+id/studentEntrySalary" android:text="Foo" android:layout_width="wrap_content" android:layout_height="wrap_content" /> </LinearLayout> Maybe something wrong on my virtual device ?

    Read the article

  • internet explorer, google chrome injection

    - by Volim Te
    I wrote code that injects a function in Internet Explorer/Chrome but it doesn't work with these processes. Basically, it fills one big structure with all the APIs my function needs, strings, and other data, then it opens a process to get a handle, virtualallocex to allocate enough memory to store a function and structure there, and it writes the function and the structure in allocated memory. It then runs createremotethread there with the function as a starting address and structure as parameter. It works all great with calc/notepad/winamp processes but I have problems with browser injection. I'm wondering what could it be, I'm using these APIs. x.xCreateFile x.xWriteFile x.xCloseHandle x.xSleep x.xVirtualAlloc x.xVirtualFree x.xMessageBox x.xLoadLibrary x.xShellExecute Is it because browsers are protected now and they're running with lowest privileges?

    Read the article

  • Bad_alloc exception when using new for a struct c++

    - by bsg
    Hi, I am writing a query processor which allocates large amounts of memory and tries to find matching documents. Whenever I find a match, I create a structure to hold two variables describing the document and add it to a priority queue. Since there is no way of knowing how many times I will do this, I tried creating my structs dynamically using new. When I pop a struct off the priority queue, the queue (STL priority queue implementation) is supposed to call the object's destructor. My struct code has no destructor, so I assume a default destructor is called in that case. However, the very first time that I try to create a DOC struct, I get the following error: Unhandled exception at 0x7c812afb in QueryProcessor.exe: Microsoft C++ exception: std::bad_alloc at memory location 0x0012f5dc.. I don't understand what's happening - have I used up so much memory that the heap is full? It doesn't seem likely. And it's not as if I've even used that pointer before. So: first of all, what am I doing that's causing the error, and secondly, will the following code work more than once? Do I need to have a separate pointer for each struct created, or can I re-use the same temporary pointer and assume that the queue will keep a pointer to each struct? Here is my code: struct DOC{ int docid; double rank; public: DOC() { docid = 0; rank = 0.0; } DOC(int num, double ranking) { docid = num; rank = ranking; } bool operator>( const DOC & d ) const { return rank > d.rank; } bool operator<( const DOC & d ) const { return rank < d.rank; } }; //a lot of processing goes on here; when a matching document is found, I do this: rank = calculateRanking(table, num); //if the heap is not full, create a DOC struct with the docid and rank and add it to the heap if(q.size() < 20) { doc = new DOC(num, rank); q.push(*doc); doc = NULL; } //if the heap is full, but the new rank is greater than the //smallest element in the min heap, remove the current smallest element //and add the new one to the heap else if(rank > q.top().rank) { q.pop(); cout << "pushing doc on to queue" << endl; doc = new DOC(num, rank); q.push(*doc); } Thank you very much, bsg.

    Read the article

  • How to Sort List Collection

    - by Muhammad Akhtar
    I have class like public class ProgressBars { public ProgressBars() { } private Int32 _ID; private Int32 _Name; public virtual Int32 ID {get { return _ID; } set { _ID = value; } } public virtual Int32 Name { get { return _Name; } set { _Name = value; }} } here is List collection List<ProgressBars> progress; progress.Sort //I need to get sort here by Name how can I sort this collection? Thanks

    Read the article

  • is my function correct?

    - by sbsp
    This is part of an assignment so please dont post solutions, just point me in the right direction if possible? I am passing a pointer to a char array to my method, as well as a value for the actual height of the char array. I am looping through to see if all values are 0, if they are then return 0, esle return one The method is used as a test to see if i should free memory or not and set the pointer to null if it is full of 0's. The issue i am having is that the programme should have "some unfree" memory at the end, so i have no idea whether or not its doing it correctly - and gdb i struggle with immensley. Thanks for reading int shouldBeNull(char *charPointer, int sizeOfCharArray) { int isIn = 0; int i = 0; while(i < sizeOfCharArray){ if(*charPointer != '0'){ isIn = 1; break; } i++; charPointer++; } return isIn; }

    Read the article

  • Can this way of storing typed objects be improved?

    - by Pindatjuh
    This is an "can it be improved"-question. Topic: Storing typed objects in memory. Background information: I'm building a compiler for the x86-32 Windows platform for my language. My goal includes typed objects. Idea: Every primitive is a semi-class (it can be used as if it was a normal class, but it's stored more compact). Every class is represented by primitives and some meta-data (containing class-properties, inheritance stuff, etc.). The meta-data is complex: it doesn't use fields but instead context-switches. For primitives, the meta-data is very small, compared to a "real" class, which is alot bigger. This enables another idea that "primitives are objects", in my language, which I found nessecairy. Example: If I have an array of 32 booleans, then the pure content of this array is exactly 4 byte (32 bits of booleans). The meta-data will contain flags that the type is an array of booleans, which contains 32 entries. The meta-data is very compacted, on bit-level: using a sort of "packing" mechanism, which is read by a FSM at runtime, when doing inspection of the type (like when passing the object to methods for checking, etc.) For instance (read from left to right, top to bottom, remember vertical position when going to the right, and check nearest column header for meaning of switch): Primitive? Array? Type-Meta 1 Byte? || Size (1 byte) 1 1 [...] 1 [...] done 0 2 Bytes? || Size (2 bytes) 1 [...] done || Size (4 bytes) 0 [...] done Integer? 1 Byte? 2 Bytes? 0 1 0 1 done 1 done 0 done Boolean? Byte? 0 1 0 done 1 done More-Primitives 0 .... Class-Stuff (Huge) 0 ... (After reaching done the data is inserted. || = byte alignment. [...] is variable sized. ... is not described here, for simplicity. And let's call them cost-based-data-structures.) For an array of 32 booleans containing all true values, the memory for this type would be (read top-down): 1 Primitive 1 Array 1 ArrayType: Primitive 0 Not-Array 0 Not-Integer 1 Boolean 0 Not-Byte (thus bit) 1 Integer Size: 1 Byte 00100000 Array size 01010101 01010101 01010101 01010101 Data (user defined) Thus, 8 bytes represent 32 booleans in an array: 11100101 00100000 01010101 01010101 01010101 01010101 How can I improve this? (Both performance- and memory-consumption wise)

    Read the article

  • Marshal managed string[] to unmanaged char**

    - by Vince
    This is my c++ struct (Use Multi-Byte Character Set) typedef struct hookCONFIG { int threadId; HWND destination; const char** gameApps; const char** profilePaths; } HOOKCONFIG; And .Net struct [StructLayout(LayoutKind.Sequential, CharSet = CharSet.Auto)] public struct HOOKCONFIG { public int threadId; public IntPtr destination; // MarshalAs? public string[] gameApps; // MarshalAs? public string[] profilePaths; } I got some problem that how do I marshal the string array? When I access the struct variable "profilePaths" in C++ I got an error like this: An unhandled exception of type 'System.AccessViolationException' occurred in App.exe Additional information: Attempted to read or write protected memory. This is often an indication that other memory is corrupt. MessageBox(0, cfg.profilePaths[0], "Title", MB_OK); // error ... Orz

    Read the article

  • segmented reduction with scattered segments

    - by Christian Rau
    I got to solve a pretty standard problem on the GPU, but I'm quite new to practical GPGPU, so I'm looking for ideas to approach this problem. I have many points in 3-space which are assigned to a very small number of groups (each point belongs to one group), specifically 15 in this case (doesn't ever change). Now I want to compute the mean and covariance matrix of all the groups. So on the CPU it's roughly the same as: for each point p { mean[p.group] += p.pos; covariance[p.group] += p.pos * p.pos; ++count[p.group]; } for each group g { mean[g] /= count[g]; covariance[g] = covariance[g]/count[g] - mean[g]*mean[g]; } Since the number of groups is extremely small, the last step can be done on the CPU (I need those values on the CPU, anyway). The first step is actually just a segmented reduction, but with the segments scattered around. So the first idea I came up with, was to first sort the points by their groups. I thought about a simple bucket sort using atomic_inc to compute bucket sizes and per-point relocation indices (got a better idea for sorting?, atomics may not be the best idea). After that they're sorted by groups and I could possibly come up with an adaption of the segmented scan algorithms presented here. But in this special case, I got a very large amount of data per point (9-10 floats, maybe even doubles if the need arises), so the standard algorithms using a shared memory element per thread and a thread per point might make problems regarding per-multiprocessor resources as shared memory or registers (Ok, much more on compute capability 1.x than 2.x, but still). Due to the very small and constant number of groups I thought there might be better approaches. Maybe there are already existing ideas suited for these specific properties of such a standard problem. Or maybe my general approach isn't that bad and you got ideas for improving the individual steps, like a good sorting algorithm suited for a very small number of keys or some segmented reduction algorithm minimizing shared memory/register usage. I'm looking for general approaches and don't want to use external libraries. FWIW I'm using OpenCL, but it shouldn't really matter as the general concepts of GPU computing don't really differ over the major frameworks.

    Read the article

< Previous Page | 369 370 371 372 373 374 375 376 377 378 379 380  | Next Page >