Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
cellocello: belvdr's point was that the app is waiting for your hard drive. No matter how much you're doing in parallel, if you're loading from disk you'll almost certainly be limited by that.

Alright ... I guess I just don't understand the nature of what's going to get "tuned up" in this next release.

I mean, my CPU rarely reaches 100% (unless I'm specifically encoding something, or what have you - and even then it's not always a hard 100) - so, Snow Leopard will do nothing for users like me? :confused:
 
Snow Leopard is not GPU processing; one is an operating system, the other is a technology that operating system contains. There's plenty of things that can be changed (like, literally many millions of things) that influence performance. Most in minor ways (such as inlining childNodes() in WebKit) , some in major ways (such as the complete rewrite of the text layout system in Leopard), some in ways that only come into play in fairly specialized cases (such as GPU processing).
 
I know Snow Leopard isn't GPU processing. ;)

So you don't believe Snow Leopard will leverage new Macs with GPUs to speed up/smooth out/generally improve "everyday" apps and functions on the Mac? Stuff like Expose, preview, dashboard and what not? Things that aren't necessarily CPU killers?

You're just making it seem like the GPU leveraging is only going to happen in specific, specialized (presumably rare) instances.

(Am I using all those crazy words right ;))
 
Expose and Preview already use the GPU on Leopard. Dashboard does a little bit, but most of it is going through WebKit which doesn't (and won't on Snow Leopard either as far as I know). What OpenCL provides in Snow Leopard is a way to use the GPU for certain things that aren't graphics related.
 
I just red a very interesting press release from Nvidia.

http://www.nvidia.com/object/io_1228825271885.html

Quote from the middle of the press release:

"CUDA has had a tremendous reception from the world’s research community with scientists seeing up to a 20-200x speed-up in their applications with CUDA over a CPU."

Now, this is HIGHLY INTERESTING and will breath new live into this thread.

Of course this is about Cuda and not Open CL, but these are really just two different languages to do the same and they both run flawlessly on Nvidia GPUs which are in most Macs today.

seeing up to a 20-200x speed-up

WOW!
 
OP:"Snow Leopard may make even 2006 low end MBP's suddenly MUCH faster than even the latest MacPro WITHOUT Snow Leopard"

lol i think u need to do some reasearch on software/hardware. No way this could happen...

edit: so let me take a more sensible guess: A low end macbook 2006 with snow leopard will be MUCH slower than the latest macpro with jaguar....
 
OP:"Snow Leopard may make even 2006 low end MBP's suddenly MUCH faster than even the latest MacPro WITHOUT Snow Leopard"

lol i think u need to do some reasearch on software/hardware. No way this could happen...

There might be a few special cases:
e.g. you add SL, mod the old MBP - remember that you can have external GPU added to a laptop if you really want. For certain tasks, if SL could utilise the external cards, it could give better results.

I'm skeptical about getting up to a dual C2D Xeon 3.2GHz Mac Pro performance though.
 
I have to agree with what most people are saying here.. I'm no expert, but it would be correct to assume:

Let's look at what we perceive as 'computer speed'.


1. Snappiness / Responsiveness of the User Interface / Menus / Opening Windows etc

2. Opening and saving files / I/O Speeds / Storage Speeds

3. Time it takes to complete a process / filter / convert etc


Really, they're the three things we perceive as real speed in a computer. Unfortunately, for the OP of this thread, only one of those things has a major of being sped up by GPGPU computing.. and that's #3. While #1 MIGHT also see a speed up.

Converting files, performing a filter in photoshop, compressing media, encrypting files etc... those processes should be able to be palmed off to the GPU, resulting in a decent speed up - possibly a significant speed up, depending on how powerful your GPU is.

Also, it may be possible for the computer to feel more snappy, but I think this is only going to apply in certain areas, for example, navigating images and graphics in photoshop or other visual applications. Unless I'm missing something here - correct me if I'm wrong - but the code streams that control the user interface of MacOS X aren't really in a format that would be conducive to being performed by the GPU??

Just my $0.02. Apologies if what I am saying isn't 100% accurate. Just assumptions.

Scottie
 
Will Apple include support for older models? The new MBs/MBPs have GPU acceleration for video decoding when the older MBPs supported this feature. I wouldn't be surprised if Apple left the older models out just for marketing purposes, they've done it on numerous occasions in the past.
 
No, they're going to require recent GPUs to use GPU calculations - basically anything using the unified shader technology and later

And as others have stated, those speeds are NOT across the board. They're in highly intensive sequentia simple math problems where these speeds occur. Basically shaders can function as vectors to do these types of calculations. This has been around a while: see folding@home. The thing is, the CPU isn't going anywhere - the x86 instruction set is there for a reason and it does things no GPU can still. For instance, really complex math problems, specialized instruction sets, etc.
 

So the "real details" emerged 2 months ago?
Nope, but you can see they're doing a lot of work in the background. (e.g. the multitouch /surface team working with the Winmobile team, concurrency work going on etc etc).
As long as the implementation is software, to utilise kit that'll be in windows 7 machines anyway, they won't be hampered by Vista 7 machines not having the right bits to even run Aero 7... ;)
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.