NVIDIA Announces GTX 580

tey think april fools
Fermi architecture has its SMs arranged clusters of 32 shaders, making that the smallest unit of granularity. For the particularly slow, 580 does not evenly divide into 32, thus making this card an impossibility.

Comments on that are the first place I've seen the name "Thermi". I like it

Hahaha, Thermi.

nvidia is desperate. fermi is a HUGE flop and they know it. It hasn't even hit the market yet and people have cast it off as krap. ATI has their act together and will continue to dominate the market if Nvidia can't create a card that can compete. I told everyone that ATI would one day hit the nail on the head and boom it appears they have. When ATI gets drivers to match (though they have been doing much better) ATI will be unstoppable

nightelph said:

Hahaha, Thermi.

Click to expand...

Thats what I thought too

jarlmaster47 said:

nvidia is desperate. fermi is a HUGE flop and they know it. It hasn't even hit the market yet and people have cast it off as krap. ATI has their act together and will continue to dominate the market if Nvidia can't create a card that can compete. I told everyone that ATI would one day hit the nail on the head and boom it appears they have. When ATI gets drivers to match (though they have been doing much better) ATI will be unstoppable

Click to expand...

Again, you're assuming that Nvidia's chief market is gaming. That assumption is wrong. Nvidia is gearing themselves towards High Performance Computing, General Purpose Computing, and Hollywood | Expensive special effects.

Simply look at gross margins of a graphics card for a PC vs. a Tesla card for a server, HPC cluster. A 500$ GTX480 vs. a 3000$ Tesla. Which would YOU want to sell more of?

You have to expand your view past just gaming for GPU's these days and look at the bigger picture. And trust me, ATI is way behind in the general purpose GPU computing realm.


+1 to Thermi!

Shiggity said:

You have to expand your view past just gaming for GPU's these days and look at the bigger picture. And trust me, ATI is way behind in the general purpose GPU computing realm.

Click to expand...

ATI has separate gaming (Radeon) and GPGPU (Firestream) product lines. nVidia used to (GeForce and Tesla), but they seem to have lost any boundary between the two, and GPGPU sat on gaming it until it suffocated. I can't find any comparison benchmarks between CUDA/Tesla and Firestream, though. Got any links?