Re: mmx vs. floating point - C / C++ / MFC Discussion Boards

Re: Reading text file using ifstream

toxcct6-Dec-05 1:10

toxcct

6-Dec-05 1:10

you have to either wass it a char* string and to change the function prototype !!!

TOXCCT >>> GEII power
[toxcct][VisualCalc 2.20]

Re: Reading text file using ifstream

grero6-Dec-05 1:29

grero

6-Dec-05 1:29

Sorry about that, with the function prototype changed to char*, the code compiles of course, but I still can't open the file...

Roger

Re: Reading text file using ifstream

kakan6-Dec-05 0:02

kakan

6-Dec-05 0:02

"wchar_t *fname"?
You might considder "char * fname"?

Or a UNICODE build?

Re: Reading text file using ifstream

S. Senthil Kumar6-Dec-05 4:10

S. Senthil Kumar

6-Dec-05 4:10

Does the path to the file look like this ?

const char *path = "C:\Test\TestFile.txt";

If that's true, then you need to escape the \ character by addding another \. Like

const char *path = "C:\\Test\\TestFile.txt";

Regards
Senthil
_____________________________
My Blog | My Articles | WinMacro

Re: Reading text file using ifstream

grero6-Dec-05 8:28

grero

6-Dec-05 8:28

Hi,
Sorry, I'm not using the full path. The file I'm trying to read is in the same directory as the program itself. I tried using full path with the double \\, but to no use. The file still won't open.

Roger

mmx vs. floating point

9ine5-Dec-05 23:38

9ine

5-Dec-05 23:38

I need to get my floating points calculation to run faster, the floating point numbers stored in instances of some class in vector<class> array I just need to multiply them and add result:

class{
float val1;
float val2;
};

for(i=0; i<size; i++)
="" res="" +="class[i]-">val1 * class[i]->val2;

will it be faster to store floating numbers from vector massive to plain floating array

float *ar1, *ar2;
for(i=0; i<size; i++)
="" ar1[i]="class[i]-">val1;
ar2[i] = class[i]->val2;

and then perform mmx calculations or memory moving ops will take also some time so there will be no significant speed increase? Or will it be faster to use compound intrinsics to load floating numbers to float[4] and then perform multiplication?

If some one encountered similar problem?

9ine

Re: mmx vs. floating point

C^hris L^osinger6-Dec-05 2:11

C^hris L^osinger

6-Dec-05 2:11

9ine wrote:
perform mmx calculations

MMX instructions are all integer-based, so they can't help you at all with floating point calcs.

Cleek | Image Toolkits | Thumbnail maker

Re: mmx vs. floating point

9ine6-Dec-05 3:02

9ine

6-Dec-05 3:02

I know about mmx, by this word I mean all multemedia extensions, it is SSE with floating points which I wonder about for my application if you need me to be more exact.
I can do it my self easely but dont want to waste time if some one done it before an got speed improvement, which I mean:

timemmx = [mem move from classes in vector<> array] + [SSE ops]
timefloat = [standart floating multiplications]

timemmx << timefloat! ??

Otherwise I will need to make much bigger changes in my code to get floating values from my vector<class> to align them in memory, which I'm reluctant.

9ine

Re: mmx vs. floating point

C^hris L^osinger6-Dec-05 4:03

C^hris L^osinger

6-Dec-05 4:03

i'm not sure anyone could say for sure without actually trying.

but, items in a vector are stored contiguously in memory, so you can treat them as a plain C++ array. (this is part of the 2003 C++ standard, IIRC - and all implementations store data that way regardless). so, you should be able to load the SSE registers directly from the contents of your vector.

Cleek | Image Toolkits | Thumbnail maker

Re: mmx vs. floating point

9ine7-Dec-05 3:40

9ine

7-Dec-05 3:40

actually not the floating points are stored in vector but class instances with floating numbers

class CTest
{
public:
CTest();
~CTest();

float val1;
float val2;
}

and we get this vector<CTest *> array and we need to

for(i=0; i<array.size(); i++)
res += array[i]->val1 * array[i]->val2;

if we can try it keeping in mind that data in vector is stored lineary hence val1 and val2 stored with the same stride in memory?

9ine

-- modified at 9:42 Wednesday 7th December, 2005

changing mouse pointer at button click in vc++

Member 15877415-Dec-05 23:29

Member 1587741

5-Dec-05 23:29

16:00 IST
Good Afternoon to all of you,

My problem is based on changing the mouse pointer at the button click .
Like in ms-paint on clicking the the line button the mouse icon gets changed.
I would be highly obliged for the early response

Thanks and regards

Re: changing mouse pointer at button click in vc++

ThatsAlok6-Dec-05 0:48

ThatsAlok

6-Dec-05 0:48

pankajgarg11 wrote:
Like in ms-paint on clicking the the line button the mouse icon gets changed.

Is SetCursor Api is of any help?

"Opinions are neither right nor wrong. I cannot change your opinion. I can, however, change what influences your opinion." - David Crow

cheers,
Alok Gupta
VC Forum Q&A :- I/ IV

LPTSTR to float

paper675-Dec-05 23:25

paper67

5-Dec-05 23:25

I am using VC 6.0. I am writing an MFC application where I use strings with the _T("") macro.(Character set neutral). Is there any way to convert from a LPTSTR to a float value. I noticed that the function _ttof does not exist in VC 6.0. Does anybody know how this conversion can be done.

Thx.

Re: LPTSTR to float

Cool Ju5-Dec-05 23:36

Cool Ju

5-Dec-05 23:36

Hi
Use _tstof()

Bye,
Cool Ju Cool | :cool:

Dream Ur Destiny

Re: LPTSTR to float

paper675-Dec-05 23:51

paper67

5-Dec-05 23:51

_tstof() doesn't work.

C:\Projects\PCnt\XLConvDlg.cpp(835) : error C2065: '_tstof' : undeclared identifier

Thx

Re: LPTSTR to float

Cool Ju5-Dec-05 23:58

Cool Ju

5-Dec-05 23:58

Hi,

Add math.h and stdlib.h

Bye,
Cool Ju Cool | :cool:

Dream Ur Destiny

Re: LPTSTR to float

paper676-Dec-05 0:08

paper67

6-Dec-05 0:08

Doesn't work. Same error.

Re: LPTSTR to float

Member 15877415-Dec-05 23:45

Member 1587741

5-Dec-05 23:45

try using atof function

Re: LPTSTR to float

paper675-Dec-05 23:53

paper67

5-Dec-05 23:53

Yes, but I want to use a character neutral set function.

Thx

Re: LPTSTR to float

ThatsAlok6-Dec-05 0:53

ThatsAlok

6-Dec-05 0:53

paper67 wrote:
Is there any way to convert from a LPTSTR to a float value.

what about _tcstod function

"Opinions are neither right nor wrong. I cannot change your opinion. I can, however, change what influences your opinion." - David Crow

cheers,
Alok Gupta
VC Forum Q&A :- I/ IV

-- modified at 6:54 Tuesday 6th December, 2005

Re: LPTSTR to float

paper676-Dec-05 2:49

paper67

6-Dec-05 2:49

That's the one I need.

Thx a lot.

Re: LPTSTR to float

ThatsAlok6-Dec-05 2:55

ThatsAlok

6-Dec-05 2:55

paper67 wrote:
Thx a lot.

My Pleasure

"Opinions are neither right nor wrong. I cannot change your opinion. I can, however, change what influences your opinion." - David Crow

cheers,
Alok Gupta
VC Forum Q&A :- I/ IV

Re: LPTSTR to float

Bob Ciora6-Dec-05 1:16

Bob Ciora

6-Dec-05 1:16

How about the old standby, sscanf? There is also a wide-character version, swscanf. If you've eliminated the whitespace around the string-represented floating-point value, then the following will work:

// input: CString sData

float fValue;

// sscanf will return the number of items pulled out of the string
int nResult = sscanf((LPCTSTR)sData, "%f", &fValue);
if( nResult != 1 )
{
    // Didn't find a floating-point value.  Take corrective action...
}

Bob Ciora

Re: LPTSTR to float

Rage6-Dec-05 1:32

Rage

6-Dec-05 1:32

wcstod() contained in <stdlib.h>

~RaGE();

RE: lib for usb cam video capture

9ine5-Dec-05 22:59

9ine

5-Dec-05 22:59

Is there ready lib or dll for VC++ to capture video from usb camera with available function to get a single captured frame to Memory Array in different types of byte stream (rgb,gray,yuv, etc...) or just in any one type?

9ine

Use Ctrl+Left/Right to switch messages, Ctrl+Up/Down to switch threads, Ctrl+Shift+Left/Right to switch pages.