Rumor: NVIDIA Kepler GK104 Transistor Density Revealed
The NVIDIA Kepler GK104 GPU transistor counts have been revealed. The GTX 680, which is based on GK104, will pack 3540 million transistors into a die size of 294mm^2. With the rumors saying the NVIDIA GTX 680 is faster than AMD’s Radeon HD 7970, this would make the NVIDIA graphics card more efficient than their rivals competing product. The die size being smaller should also result in a cheaper to manufacturer GPU and hopefully lower prices. It is interesting to note however that the smaller die size is uncharacteristic of NVIDIA, whom are usually well known for their large, monolithic GPUs. The fact they are producing a smaller, faster graphics card, with a much lower TDP, shows what appears to be a change in design ideals. It could also be that NVIDIA plans to release a faster gpu sometime in the future, which may very well be monolithic in size.
As expected, despite penetrate NDA lock until 22 March Info smaller bites to Kepler nVidia chip GK104 outside. According to (the usually well-informed) user napoleon from Chiphell forum is the GK104-chip bring 3.54 billion transistors, which are located at only 294mm die area, the TDP of the GK104-based GeForce GTX 680 is now at 195 watts . are fixed This achieves nVidia first time the high transistor-packing densities of the AMD chips: While in the 40nm generation yet continuous a noticeable poorer packing density was (ie, fewer transistors on the same chip area housed), it is the 28nm generation with respect to the packing density GK104 of the chips now on the same level as in R1000/Tahiti-Chip.
Comments are closed.