On 9 Aug 2006 at 4:48, Darcy James Argue wrote: > On 09 Aug 2006, at 4:10 AM, Tyler Turner wrote: > > >> Oh fercrissakes. We aren't comparing a machine with > >> NO video card > >> versus a machine with a video card. I never said > >> that the video card > >> NEVER makes any difference to 2D acceleration under > >> ANY circumstances. > > > > That's not what turning down the acceleration on > > Windows does. It doesn't disable your video card. It > > disables features, like anti-aliasing, etc. > > 2D anti-aliasing is not handled by the graphics card! It's handled by > the *CPU*, . . .
How do you know this "fact?" Is your knowledge based on how Apple's hardware works? Or an investigation of both Mac and Windows? My imperfect understanding on the Windows side is that the graphics standards (DirectX and so forth) are supported in the hardware of the graphics cards so that certain kinds of drawing operations are entirely handed off to the graphics processor. I don't know exactly which operations, but if it works the way the printing subsystem of Windows works, the CPU might calculate the Windows abstraction for requesting a shape be drawn, and the processor on the graphics card would then do the calculations to put the appropriate pixels in the right location. In reaching for an analogy, I'd say that In a sense it's like the CPU calculates what vector is needed, sends the vector definition to the graphics card and the graphics card does the work of rasterizing the vector into actual pixels. Indeed, it may be that an application is requesting the appropriate vector and the CPU does no work *except* handing off the request to the graphics card. Now, obviously, on low-end PCs, a lot of the graphics work is done by the CPU, and a lot of cheap graphics cards use system RAM for their calculations. But that's not what we're talking about here, so far as I understand. > . . . which calculates everything itself and then tells the > graphics card what to draw. Of course disabling anti-aliasing means > you can scroll around faster -- if the CPU doesn't have to calculate > the anti-aliasing, then it can tell the video card which pixels to > draw much more quickly. This is at odds with my understanding of how PC graphics cards work. That doesn't mean you are wrong, but it would mean that I've misinterpreted about 15 years of reading about graphics hardware in the PC press. > Graphics cards do have hardware-supported anti-aliasing, but only for > 3D applications like games. Are you certain this is correct? Are you sure the graphics cards are doing *none* of the 2D calculations? I don't believe that's so at all, at least for PCs. > If you upgrade your processor but keep the same video card, you will > get faster Finale redraws. If you upgrade your video card, you will > see little to no improvement in your Finale redraws, because Finale > redraws, being a 2D task, are CPU-bound. All the graphics card > benchmarks bear this out. This is only true if you're correct about 2D drawing *never* being handed off to the graphics subsystem by the CPU. > > I just performed a Finale test that puts more focus on > > the video card. By dragging the screen around via > > right-click drag, I find that the screen redraws MUCH > > more smoothly when I turn the acceleration on my video > > card all the way down. Why? Because it isn't having to > > calculate all of the special effects. > > No, this is because the *CPU* isn't having to calculate all the > special effects. The GPU only handles stuff like transparency, video > playback, lighting, caching windows, and 3D effects like (in Mac OS > X) fast user switching, Dashboard, and Exposé. Redrawing Finale > windows is handled by the CPU. You can make the CPU's job easier by > reducing anti-aliasing (thus reducing the number of calculations), > but this has nothing to do with the video card. Do you know for a fact that all of the above applies to the way Windows works? Or are you just assuming that Windows works the same way as OS X? > MM *could* theoretically offload the drawing of Finale screens to the > graphics card by making the entire scroll view score a massive 3D > texture, in which case the GPU would handle scrolling and zooming. I > have actually suggested this idea to MM, but this is not something > they have done so far. In Windows, these kinds of things are not determined by applications, but by the OS graphics subsystem. > Seriously, please just go look at some 2D benchmarks for graphics > cards I don't doubt that you know how Apple hardware works. But I have some doubt about how applicable that knowledge is to PCs. -- David W. Fenton http://dfenton.com David Fenton Associates http://dfenton.com/DFA/ _______________________________________________ Finale mailing list Finale@shsu.edu http://lists.shsu.edu/mailman/listinfo/finale
