An email from my friendly NVIDIA rep called my attention to this recent blog post from Andy Keane, head GPU honcho at NVIDIA. In the post, Keane thoroughly pounds Intel for presenting a paper titled Debunking the 100x GPU vs. CPU Myth which, in its abstract, asserts that an older NVIDIA GPU (the GTX280) is only 2.5x faster than …
Intel needs to be slapped
Their support (or glaring lack thereof) of GMA500 or Poulsbo support for Linux AND Windows is SHAMEFUL.
Intel is all hype
Intel's believes their experience with Larrabee is ipso facto proof that accelerators are not all that that are hyped to be. It's because they live by Grove's law; if Intel can't do it, it can't be done.
Accelerators are all hype
If they're so good, why do they not benefit the good of the common people?
By common, I mean professionals who don't work in medical science or astrophysics.
I have a Quadro FX 3800 and while it accelerates some things, that's down to the drivers not CUDA or OpenCL. A Fermi card would be faster, but then Autodesk, for example, wouldn't support it, rendering my expensive support contract useless.
This is an industry full of figures and multipliers, and while they make interesting reading they still mean squat to most people.
But they do:
>> If they're so good, why do they not benefit the good of the common people?
But they do, just not in the same context. Remember that these are primarily graphics accelerators, used by the likes of young techheads like me to render impressive looking games.
There's also a number of CUDA accelerated video encoding tools, PhysX provides GPU accelerated physics processing for games and simulations, so on and so forth.
I love my GTX 480 :)
- Updated Zucker punched: Google gobbles Facebook-wooed Titan Aerospace
- Elon Musk's LEAKY THRUSTER gas stalls Space Station supply run
- Android engineer: We DIDN'T copy Apple OR follow Samsung's orders
- Microsoft: Windows version you probably haven't upgraded to yet is ALREADY OBSOLETE
- Oz bank in comedy Heartbleed blog FAIL