Posted on 10/01/2009 9:17:53 AM PDT by Ernest_at_the_Beach
The graph below is one of transistor count, not die size. Inevitably, on the same manufacturing process, a significantly higher transistor count translates into a larger die size. But for the purposes of this article, all I need to show you is a representation of transistor count.
See that big circle on the right? That's Fermi. NVIDIA's next-generation architecture.
NVIDIA astonished us with GT200 tipping the scales at 1.4 billion transistors. Fermi is more than twice that at 3 billion. And literally, that's what Fermi is - more than twice a GT200.
At the high level the specs are simple. Fermi has a 384-bit GDDR5 memory interface and 512 cores. That's more than twice the processing power of GT200 but, just like RV870 (Cypress), it's not twice the memory bandwidth.
(Excerpt) Read more at anandtech.com ...
HOLY MOLY!!!!
That’s insane.
3 billion transistors seems a little low for that many cores though. A current quad core i7 CPU from intel has almost a billion transistors. The six and eight core i9 CPUs will probably have 2 billion or more transistors. I would think.
I’m having a hard time imagining what software an average joe is going to run that will make full use of computer power 5 years from now.
5.56mm
When I read (last year I guess) that the Folding@Home software had been ported to the GPUs, and that they were smokin’ all the other categories combined (including the PS/3, which had already previously smoked all the previous categories combined in about seven months’ time), I wondered how long before the GPU-makers parlayed that into success with mobile devices and whatnot. This is some cool whatnot, eh? ;’)
>> Can programs such as Photoshop CSx or Gimp be easily recompiled to run under CUDA or will special versions have to be written?.
Not really. But computationally intensive algorithms could be cast off to the GPU while the CPU is handling other aspects of application processing.
Although Photoshop would not run from the GPU, things like filter effects, video encoding, and image scaling subsystems and plugins could leverage the GPU’s parallel processing technology resulting in total performance improvement. In other words, certain blocks of application functionality could execute on the GPU while the CPU services the bulk of the application’s general I/O and GUI requirements.
Interesting talks on the new generation GPUs. As you noticed I do not keep abreast on new developments in the IC world or computers in general. But find it interesting reading.
NVIDIA's "Fermi" Architecture White Paper
****************************EXCERPT*************************
For those of you that are wanting a better look inside of Fermi, we have republished NVIDIA's Fermi white paper here for your reading pleasure.
******************snip**********************
NVIDIA has told us today that it will be a "few short months" before we see Fermi GPUs on retail video cards for gaming. They have also told us that Fermi will be close to the same power envelope as its current GPUs.
There are HardForums threads raging in the Video Card section for those of you wishing to jump into the fray.
******************************************
This is not going to be just a higher powered video card.....
Jump to the White paper....and .... the conclusion.....
************************************************
Perhaps I’ just purchase a new super high end computer loaded with one of these GPU cards and replace the oil filled space heater in my basement with it. Talk about carbon foot prints. heheh.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.