Hardware vs Software

[SMcNeill]

I had a couple of people on Discord message me asking about if I could toss up a performance difference between software and hardware images for them.  I swear, I know I had some already here somewhere, but I couldn't find them when I snooped for them earlier, so -- by request -- here's a quick and dirty demo of the difference:

Code: (Select All)

Screen _NewImage(1280, 720, 32)



box = _NewImage(320, 180, 32) 'make a screen to just hold a box

Cls , &HFFFF0000&&, box 'make that box red



count = 0: t# = Timer + 1

Do

    count = count + 1

    Cls

    If Timer > t# Then

        box_moves = count

        count = 0

        t# = Timer + 1

    End If

    bx = (bx + 1) Mod 960

    _PrintString (0, 0), "The box moved" + Str$(box_moves) + " times per second with software images."

    _PutImage (bx, 100), box

    _Display

Loop Until _KeyHit





_Delay .5

_KeyClear



box2 = _CopyImage(box, 33) 'make us a hardware image of that same box

_DisplayOrder _Hardware , _Hardware1 'and we're only going to display hardware screens

t# = Timer + 1: count = 0

Do

    count = count + 1

    If Timer > t# Then

        hardbox_moves = count

        count = 0

        t# = Timer + 1

        Cls

        _PrintString (0, 0), "The box moved" + Str$(box_moves) + " times per second with software images."

        _PrintString (0, 20), "The box moved" + Str$(hardbox_moves) + " times per second with hardware images."

        If temp <> 0 Then _FreeImage temp

        temp = _CopyImage(0, 33)

    End If

    bx = (bx + 1) Mod 960

    _PutImage , temp 'place this image on the hardware screen

    _PutImage (bx, 100), box2, 1 'place this image on the hardware screen 1 (the 2nd hardware screen)

    _Display

Loop Until _KeyHit

System

Run that. Open your program manager and watch your CPU usage. Let it go for several seconds and you can see how quickly it moves a simple red box across the screen for you.

Then hit the space bar... Give it a couple of seconds to swap over from software to hardware images, and watch and see what the difference in those numbers might be.

Note that in the second portion of the program, where we're tossing out our hardware image, that I'm STILL drawing to a software image in the background. (Well, clearing it and printing to it.) Then I'm making a hardware copy of that image to display. While ALSO using a second hardware image for the box itself...

And yet, it still runs just a *wee* bit faster for us, with lesser CPU usage and all the other benefits that hardware brings over software.

Now tell me, is it perhaps a *wee* bit faster on your machines as well?