So will the GPU be relegated to things like Photoshop filters, or will it be able to help crunch other things - like encoding an H.264/AVC video?

Snow Leopard is a pretty shitty name I can't believe apple is running with it.

I have been wondering why I never have access to see exactly how much of my GPU is being taxed. I hope this means I will be able to do so in the future.

This news just in: Ram and CPU power also speed up PC's and Macs too! who knew!

I wonder what the effect would be if we ran both Nvidia and ATI high end graphics cards in a system?

Which card would we connect the monitor to?

Would the system use both GPUs for processing?

It'd be kind of interesting to see what the results would be. Too bad we couldn't connect the cards together as you can with Crossfire or SLI.

gpu-z will show your gpu load under "sensors" tab. it does for my 4850.

wow!

not really, the gpu is not running x86 instruction set, but for certain tasks the software developer could have specially coded code that if the gpu are around it could offload those computations to. i recently wrote some code that does just that, microsoft and nvidia and ati make it very easy to do so. if there is not gpu then your cpu is stuck with the task, but if there is then it get it and the cpu is free to do other things. this will allow for some amazing windows eye candy tweaks, much like we see in games being available to the desktop programs rendering. i did a simple texture filter, and wow it worked, microsoft visual studio made is soo easy to do, just simple function. think of water, cload, smoke, ect effects and windows. wow

Snow Leopard is a pretty shitty name I can't believe apple is running with it.

Hahah well, there is a real leopard that is called a Snow Leopard, it's also known as an Ounce. The Toronto zoo had one when I was there last.

Now for my self I find the name Twitter pretty bad hahahah

Cool, so does that mean theoretically I only need either a gpu or cpu to boot up and use the most basic funtions of my computer? Well there are the physical limitations like no vga/dvi port...

ok so if ATI can also take use of it then y is the title talks only about Nvidia

is this going to be limited to cards specifically made for DirectX 11? Is my GTX 260 SLI setup going to be unable to take advantage of parallel processing?

Yeah, and Windows 7 is so brilliant. Who would have thunk?

At least Windows 7 doesn't sound like it wants to kill you...

Photoshop CS4 already uses CUDA (for some Nvidia cards) for some filters.

ROFL True!

@ok so if ATI can also take use of it then y is the title talks only about Nvidia

It's because ATi hasn't released something for OpenCL, yet. Anyone who has seen how to even initialize CAL would see that it isn't really ready for primetime yet, it needs a bit more polishing (and ease of use!). From what I read elsewhere, OpenCL is almost based on CUDA (or the code appears almost like it).

GPU's have a major advantage when it comes to anything having to do with math, thanks in part to a wider data bus (ie: a 256bit data bus for a GPU with 256bit registers can load 8 32-bit integers in a single operation, while a 64bit CPU can load only 2 32-bit integers in a single operation). Hence why PhysX runs so much faster on a GPU then on a CPU.

I wonder though, if dual cards will suddenly lead to a noticable performance increase in general computing as a result...

Heh I was wondering when an AMD afficianado would complain. Surprisingly, I had to scroll pretty far down the list to find the first one. Even the Nvidia rep quoted in the article says both Nvidia AND AMD products are going to be able to do this.

Who cares what the title said? You read the article so you got all the information, right?

So with a GTX 295 for example, how many cores will appear in the Task Manager, 480 cores? How about a 4870 X2, 1600 cores? I'm sure they'll figure out a better way than to utterly overcrowd that Task Manager window

Only 1 core per card or set of cards will appear in task manager.

Funniest thing is that Apple will release Snow Leopard well , in the summer . Hope it will be able to cope with the heat. Maybe they should have waited for winter

OpenCL is a step in the right direction. It will be pretty amazing if the various players can come to a compromise and give us access to the GPUs for tasks other than graphics rendering. CUDA is nice but I would much rather develop against OpenCL and grab Mac, Linux, Windows, ATI and NVIDIA in one go. I want to use it for astronomical image processing.

"Snow Leopard"???
Just more proof that Apple is the Fisher-Price of the computer industry.

It already does both. A lot of Adobe CS4 programs are optimized for the NVIDIA GPUs (Video editing, video compositing, encoding, photo editing, etc). Nero MoveIt, Cyberlink PowerDirector, Elemental's Badaboom all use CUDA (and sometimes ATI Stream) for video effects/rendering. There are a ton more for other computing tasks too (take a look at NVIDIA's CUDA page). Folding@home for example.

CrosSLI?

seriously atleast with vista you need the same display drivers to run aero, and the main monitor (primary) is whichever you select (after the priority of the bios etc)

For crying out loud! Where's ATI's solution! What about all those laptops that have the HD 3200 and 3100, why can't ATI do something that would make those who already own one... Happy! All I know is when I buy a product and that product is DEAD within 4-6 months and has no more support.. I just don't buy from that company any more.

I have a GeForce 260 GTX and a GeForce 8 8800 GTS (Which I still actively use at home with Folding@Home), all I have to say is that I am very happy with the fact that when nVidia introduced hardware accelerated CUDA starting from th 8 series, I was very happy that I bought nVidia and not ATI (I own an HD 3870 too and not a single word from ATI on supporting the 3xxx series with the new STREAM tech. ).

ATI just lost a costumer... sad.

wow I give you my doofus ofm thw week award .. dropign a company becaue they can't add backward support for all their old products .. damn i guess this eliminates half teh tech companies on your list. seriously not every thing can be made to flow baackwards sure it is nice when it does but sometimes it is just impossible to ahve every thing this way. i guess you should drop intel and AMD cpu's now too since neither the i7 or the phenom 2 (am3) chips are backward compatible with older main boards ????. or maybe you should drop nvida since CUDA and phyx is n ot possible on a GF 6000 series gpu ??? my point beign this is about the dumbest reason i ever heard some one say "they lost a customer" over.

if the 3xxx series can handle the "stream" tech then i'm sure ati will eventually add it to that series assumign it is a worth while buisness venture for tehm to do so (keep in mind amd's 4xxx series is quite inexpensive as far as gpu's go not to mention the 3xxx is starting to end it's production cycle)


now don't get me wrong I'm NOT a fanboy, I am actually currently useing a gf 9600 gt. I just had to say something cos i ahte to see fanboyism born out of simple stupidity. you wont go see me rantign and gettign huffy should nvidia invent some new tech next year that isnt compatible with my card ... this is the way of the world 90% of the time NEW tech will NOT be backward compatible with OLD tech. it is great when it is but history proves that it is a rare thing , it's certainly not a good reason to denounce a company and publicly name your self a FANBOY.

They sould have done this befor with tweeking
GPU speed at 1.5GHZ that runs like CPU speed it will blow it away
and the funny thing is that OS have a ways of tweaking speed up the processing of applications speed it what theya re trying to say. no wonder our kennel processing is too slow for alota things that CPU was going fast as they can't keep up the old kennel processing and GPU can't keep up that the kennel processing being slow that it holding our true speed of computing. might want to relook at how it made from our lacking slow OS.

For the people being retarded and saying "Why doesn't ATi have a solution!" It's called Stream. And people should be developing for OpenCL not CUDA. Why are people so fond of this random proprietary crap? The only voice of reason was the guy who realized it's better to get everyone in one go, not develop for one thing and hope people use it.