Im guessing that in about half a year im gonna want a new computer. do u think the imacs will ever get 256mb graphics cards before apple goes x86 or should i get a g5 before they go intel?
I think you don’t have to be concerned about whether your Mac is PPC- or x86-powered, unless you’re using Mac OS (Classic) apps or you seriously depend on the performance of Mac OS X apps that will not be updated as Universal Binaries.
I don’t think it’s very likely that they will get to 256 MB before they become Intelerons. Apple just revamped the line, and I’m guessing that we’ll be seeing these new iMac G5 2.0 models for quite some time.
I think you don’t have to be concerned about whether your Mac is PPC- or x86-powered, unless you’re using Mac OS (Classic) apps or you seriously depend on the performance of Mac OS X apps that will not be updated as Universal Binaries.
Do you have any specific information about Classic app performance on Intel?
That’s right. It’s a line in the sand. There is no Classic on Intel Macs.
Now, I suspect the iMac will be one of the last Macs to go Intel. I think the switch-over will go down the notebooks and then back up the desktops. That means the PowerBook will be first, then the iBook, then the Mac mini, then the iMac, then, finally, the Power Mac and Xserve. (I assume here, of course, that the lineup itself won’t radically change. It could. I make no predictions for that event.)
Therefore, since the transition has a target start of June 2006 and a target finish of June 2007, it may be 2007 before the iMac goes Intel. In that case, there will certainly be at least one more iMac G5 revision. In that revision, the iMac just might get 256 MB graphics.
I predict that Apple will combine all their computers into two distinct lines, the Desktop and the Laptop. They will only offer two specifications for each, a consumer model and a professional model. This will greatly reduce confusion as to which model you should get, and will make identifying Macs much easier.
I also predict that Microsoft will give all its money to charity and declare itself a nonprofit organization.
Im guessing that in about half a year im gonna want a new computer. do u think the imacs will ever get 256mb graphics cards before apple goes x86 or should i get a g5 before they go intel?
What do you want to do with this monstrosity of a g-card?
With a resolution of 1024x768 and 32-bit color we get
1024*768*32/8/1024/1024 = 3 MB RAM needed to keep that screen.
Yeah, but if you are using OpenGL, which is used heavily by OSX’s Quartz, you need significantly more for storage of textures and other effects that the interface uses.
How many textures need to be stored there at any given time? On average?
I think the textures are so few on average that just a few MB is in constant use.
This huge spec is only motivated for hi-end games. (Like Doom 3, I can’t come up with anything else).
How many textures need to be stored there at any given time? On average?
I think the textures are so few on average that just a few MB is in constant use.
This huge spec is only motivated for hi-end games. (Like Doom 3, I can’t come up with anything else).
Well… A texture for each window you have open, a texture for the desktop, a texture for every reflection, a texture for every widget, three buffers of each desktop, two buffers for windows… you get the idea.
How many textures need to be stored there at any given time? On average?
I think the textures are so few on average that just a few MB is in constant use.
This huge spec is only motivated for hi-end games. (Like Doom 3, I can’t come up with anything else).
Well… A texture for each window you have open, a texture for the desktop, a texture for every reflection, a texture for every widget, three buffers of each desktop, two buffers for windows… you get the idea.
Bob
Not to mention those are stored in bitmap form, and if you’ve seen the size of a bitmap for a 640x480 image, the average size of a window, you could imagine the effect of having several windows open at the same time would have on memory usage. You also need to consider how much more video memory you need for things like Photoshop and Final Cut, which use Quartz extensively, there you have several more kinds of textures being loaded, either for presentation or off-screen processing, sure you can put stuff in regular memory or in swap, but all that swapping can make your work slow to a crawl and as such limits what the developers will make for the Mac, knowing that common models of Apple hardware used by professionals typically have only 128MB of vram.
When I’ll some time in the future shop for a g-card then I’ll worry more about supported functions than about space in it.
Vectors don’t require much space, bitmaps may if the texture is complex.
Say you’ll paint a chessboard on the screen. The black & white squares will only require the color information of two dots (in theory about optimality), the pieces too even if you want a 3D-look on them, it can be rendered by the g-card.
If we work with a 32-bit 1024x768 resolution then the entire screen will take 3 MB.
If each window on average takes 1 MB then you’ll need 29 of them to fill VRAM of 32MB.
I think this is not the case. If the background in the windows is just one colored then you don’t need to buffer the entire windows for example.
I’m not sure how it works but this seems reasonable. I can play pretty advanced 3D games for example with a humble 32MB VRAM.
It appears to me that you have no idea how OS X renders its screen contents. I don’t really have time to explain as I understand it, but you can read what John Siracusa has to say about it over at Ars Technica. It’s much more complicated than you’re describing, as pretty much every window is rendered as an OpenGL polygon.
Anyway, read that and let me know if you’re still confuddled.
According to the article and diagrams is the CPU mostly sending commands, drawing is almost completely done by the GPU in QE. To avoid slow pipes is also frequently used bitmaps cached in VRAM which is virtualized like RAM.
Since drawing is done per app is also the amount of VRAM needed very small, usually much less than the 3 MB the entire screen fills in memory.
You will see lags when switching from app to app though on a heavy loaded system, or if the screen resolution is so big that you can see all the apps competing for the GPU at the same time (a covered window may not need to be drawn) - with which more VRAM will help of course, but not as much as you all seem to think, QE works with less than 32 MB VRAM.
You will see lags when switching from app to app though on a heavy loaded system, or if the screen resolution is so big that you can see all the apps competing for the GPU at the same time
The OS and any application that isn’t a game (or something as taxing as a game), or some experimental program, shouldn’t have any lags in a modern system.
