Jump to content

Nvidia GPU Technology Conference Coverage


Bosco

Recommended Posts

  • Replies 62
  • Created
  • Last Reply

Top Posters In This Topic

Wow, nice to finally see official specs on the Fermi/GT300... decent step up in specs, and the L2 cache and 64-bit addressing will no doubt push GPGPU/CUDA forward at an even faster rate :)

Share this post


Link to post
Share on other sites

I smell big Trouble for ATI :mellow:

me too

 

8800 memory bus is back too ...:huh:

 

but since it is GDRR5 i think it will be fast , but why did not ati increase the memory bus ...?

 

i just got a thought , 360/375 will be 384-bit but the high end will be 448-bit ..

 

what do you think..?

Edited by N.E.A

Share this post


Link to post
Share on other sites

Wow, nice to finally see official specs on the Fermi/GT300... decent step up in specs, and the L2 cache and 64-bit addressing will no doubt push GPGPU/CUDA forward at an even faster rate :)

Yeah the plugin for C++ looks very interesting. I see Fermi being a direct competitor to Larrabee in the GPGPU computing area when it launches.

 

But they did not talk much about gaming performance, did they?

Share this post


Link to post
Share on other sites

...but why did not ati increase the memory bus ...?

Wider memory buses are *very* expensive both on the die and on the physical card. GDDR5 is about twice as fast per clock as GDDR3/GDDR4 so a wider bus isn't needed for monstrous amounts of bandwidth.

 

 

The thing I'm excited about is the C++ compiler for it! I bet Intel just . it's pants. :lol:

Edited by Waco

Share this post


Link to post
Share on other sites

Won't need a motherboard soon, the graphics card will do it all! :lol:

 

Cheers

 

Paul

 

The term Co processing is being badied about. The GPU is not being looked at as a replacement for the CPU snce the CPU is great at serial functions while the GPU is better on the parallel functions . Meaning the two together do the work faster

Share this post


Link to post
Share on other sites


×
×
  • Create New...