Nvidia Kepler & Tegra stuff...
View Single Post
November 28, 2011, 07:13 AM
HardwareCanuck Review Editor
Join Date: Feb 2007
I think you guys have to think of it in a different perspective.
NVIDIA can take their time on this one because of GF110's strength in the current lineup. AMD's high end Tahiti cores have to come up with at least a 20% performance increase over Cayman just to draw even with the GTX 580, let alone overclocked versions thereof. That is no small feat.
Plus, if you don't think AMD is having some serious issues with 28nm fabrication, you'd be pretty far off base. Remember that with Fermi, NVIDIA went through the process of completely redesigning their architecture for a DX11 environment. AMD meanwhile has essentially been using the same architecture since RV600 with a few bits and pieces added or tacked on for basic DX11 functionality. Tahiti (and possibly all of the Southern Islands family) is their first true ground up DX11 part and they are surely going through the same issues as NVIDIA did when designing Fermi: low yields, massive die size, etc.
To make matters even worse, TSMC's cancellation of the 32nm process has led directly to a ton of pent up demand for 28nm wafer production. This means everyone from Texas Instruments to Qualcomm to AMD to NVIDIA are going to be fighting for 28nm fab space. You can only imagine what that is going to do for GPU production quantities...
My System Specs
View Public Profile
Send a private message to SKYMTL
Find all posts by SKYMTL