LOL ... 800 stream processors for ATi are not their terminology it's consumers reality in last 14 month. And some fast food journalist that writes hardware news for the last 10 years should know better.
William Dally, chief scientist of Nvidia Corp., said at the Design Automation Conference that in order to continue rapid evolution of graphics processing units (GPUs), Nvidia and other designers will need new electronic design automation (EDA) tools to optimize power consumption and improve design efficiency. He also made several predictions regarding the future of GPUs.
“Chip designers will need tool and techniques to optimize power, interconnect and locality. We are really looking to EDA to give us power tools. I want high-level tools that allow you to gain insights into power architectures very early in design,” Bill Dally said, reports EETimes web-site.
More advanced tools are required to enable power exploration and analysis at the architectural levels. Interconnect is a dominant factor in power consumption, Mr. Dally indicated. To succeed further, the semiconductor industry needs computer aided design (CAD) tools that capture high-level design and intent, as well as new architectures and compilation to expose locality.
Mr. Dally refers to graphics processing units as “throughput processors” and claims that they will continue to evolve rather rapidly in the following years. Mr. Dally predicts that in 2015 graphics chips will have 5000 stream processing engines, will have about 20TFLOPs of computing power and will be made using 11nm fabrication process.
"In 2015, the 11nm-generation will have 5,000 floating point arithmetic cores, a performance of 20TFLOPS per chip with an operating frequency of 3GHz, and a memory bandwidth of 1.2TB/s," said Mr. Dally according to a report from Nikkei.
The GeForce GTX 285 (G200b) – the current flagship GPU from Nvidia – has 240 stream processors, whereas ATI’s most powerful Radeon HD 4890 chip (RV790) features 160 5-way SIMD units, or 800 stream processors in ATI’s terminology.
According to Nvidia’s chief scientist, advanced EDA tools can enable performance leaps greater than those enabled by Moore’s Law.