NVIDIA’s Kepler based Compute Monster Features 7 Billion Transistors, Unveiling at GPU Technology Conference in May
NVIDIA would be announcing various new technologies at its upcoming GPU Technology Conference in May, They would also showcase various new features of a new GPU which would hold 7 Billion transistors on a single die. From NVIDIA:
"Individuals from the GPU architecture and CUDA software groups will dive into the features of the compute architecture for" Kepler "- NVIDIA's new 7-Billion transistor GPU. From the reorganized processing cores with new instructions and processing capabilities, Thu an Improved Memory System with faster atomic processing and low-overhead ECC, we will explore how the Kepler GPU achieves World Leading Performance and efficiency, and How It Enables wholly new types of parallel To Be Solved problems. "
Seven Billion transistors on a single die is quite a huge deal, For comparison AMD's Radeon HD 7970 graphics card has 4.3 billion transistors fused inside a 365mm2 die. This would mean that the compute unit would have a die size beyond 550mm2 which hints at the GK110. GK110 is currently a flagship chip slated for launch in Q3/Q4 2012, exact date is unknown. The GPU based on GK110 would hold 2000+ Cores and 4GB memory through a 512-bit interface.
Though it isn't known if this compute unit is actually based on the GK110 or some other chip. One thing's for sure, Offering twice the transistors of the current flagship GTX 680 would mean the cost would be quite high. Possibilities remain that this could be a Quadro/Tesla part. The GPU Technology Conference would be held in San Jose, United States from May 14 - 17.