We have been reporting on the GeForce GTX 970 memory allocation issue since Saturday. If you are not aware the card seemed to only be able to access 3.5 GB of its supposed 4 GB VRAM. NVIDIA has already responded to these claims and now they have actually gone ahead and changed the official specifications for the card.
Senior Vice President of Engineering at NVIDIA Jonah Alben sent PC Perspective the below diagram of the reason behind the GTX 970 acting weirdly.
Looking at the diagram we can see that there are four key things greyed out. Three SMM blocks and a single L2 cache block. NVIDIA has said that, “Despite initial reviews and information from NVIDIA, the GTX 970 actually has fewer ROPs and less L2 cache than the GTX 980. NVIDIA says this was an error in the reviewer’s guide and a misunderstanding between the engineering team and the technical PR team on how the architecture itself functioned. That means the GTX 970 has 56 ROPs and 1792 KB of L2 cache compared to 64 ROPs and 2048 KB of L2 cache for the GTX 980”.
So is the GTX 970 really a 4GB card? Well we know that it has 4 GB of VRAM and you can access the 3.5 GB just fine, but when you go to access the last 0.5 GB it is far slower than the first 3.5 GB. How slow? From the tests that have been performed it looks like it is at 1/7th the speed of the first 3.5 GB.
Source: PC Perspective | News Archive