Drawing tilemap is lowering my fps to a crawl

[CyanPrime]

Okay, I'm trying to get my fps to 60, but right now it's at around 20. What can I do to this code to speed it up? Note: this is c++ using sfml.

Code: [Select]

       App.Clear();
        for(int x = 0; x < lv.width; x++){
            for(int y = 0; y < lv.height; y++){
 
                int tileXCoord = 0;
                int tileYCoord = 0;
                int tileSheetWidth = tilemapImage.GetWidth() / lv.tileSize;
 
                if (lv.tile[x][y] != 0)
                {
                    tileXCoord = lv.tile[x][y] % tileSheetWidth;
                    tileYCoord = lv.tile[x][y] / tileSheetWidth;
                }
 
                tilemap.SetSubRect(sf::IntRect(tileXCoord * lv.tileSize, tileYCoord * lv.tileSize, (tileXCoord * lv.tileSize) + lv.tileSize, (tileYCoord * lv.tileSize) + lv.tileSize));
                tilemap.SetPosition(x * lv.tileSize, y * lv.tileSize);
                App.Draw(tilemap);
            }
        }
 
        playerSprite.SetSubRect(sf::IntRect(player.width * player.frame, player.height * player.state,
                                       (player.width * player.frame) + player.width, (player.height * player.state) + player.height));
        playerSprite.SetPosition(player.x, player.y);
 
        App.Draw(playerSprite);
 
        if(player.walking){
            if(player.frameDelay >= 0)
                player.frameDelay--;
 
            if(player.frameDelay <= 0){
 
                player.frame++;
                player.frameDelay = 10;
 
                if(player.frame >= 4)
                    player.frame = 0;
            }
        }
 
 
        for(int x = 0; x < lv.width; x++){
            for(int y = 0; y < lv.height; y++){
 
                int tileXCoord = 0;
                int tileYCoord = 0;
                int tileSheetWidth = tilemapImage.GetWidth() / lv.tileSize;
 
                if (lv.ftile[x][y] != 0)
                {
                    tileXCoord = lv.ftile[x][y] % tileSheetWidth;
                    tileYCoord = lv.ftile[x][y] / tileSheetWidth;
                }
 
                tilemap.SetSubRect(sf::IntRect(tileXCoord * lv.tileSize, tileYCoord * lv.tileSize, (tileXCoord * lv.tileSize) + lv.tileSize, (tileYCoord * lv.tileSize) + lv.tileSize));
                tilemap.SetPosition(x * lv.tileSize, y * lv.tileSize);
                App.Draw(tilemap);
            }
        }
 
 
        App.Display();

[Elffus]

If tiles aren't being constantly changed wouldn't it be easier to store them in an sf::Image and only perform changes when necessary? Showing one big image is much faster than loads of smaller ones I believe.

[CyanPrime]

If tiles aren't being constantly changed wouldn't it be easier to store them in an sf::Image and only perform changes when necessary? Showing one big image is much faster than loads of smaller ones I believe.

How would I go about doing this?

[Groogy]

If you are using SFML2 then you can use sf::RenderImage otherwise sf::Image::Copy.

http://www.sfml-dev.org/documentation/2.0/classsf_1_1RenderImage.htm
http://www.sfml-dev.org/documentation/1.6/classsf_1_1Image.htm#408c40c8675363adfa96a8b9a9576bc2

[CyanPrime]

That worked! Thank you very much!!!!

[Gibgezr]

The thread below has example code that demonstrates one way of doing it:
http://www.sfml-dev.org/forum/viewtopic.php?t=4376

[drakelord]

Just out of curiosity, what makes the engine run faster displaying one image versus a bunch of smaller ones?  

One reason I can think of in my head is only having to perform all of the extra manipulations such as scaling and what not once, but other than that, I don't know.  I don't have much OpenGL experience to know what goes on behind the scenes.