The competition in the field of deep learning chips has never stopped. In 2018, a deep learning hardware war will be launched. In this battle, NVIDIA, AMD, and Intel can laugh at the end.
With the release of NVIDIA TItan V, we have entered the turbulent period of deep learning hardware development. Whether NVIDIA can maintain the position of the leading supplier of deep learning hardware in 2018 is not known, AMD and Intel Nervana still have opportunities.
So for the consumer who wants to buy hardware, the most savvy choice is to wait for 3 to 9 months, and then decide after this uncertain state has passed.
NVIDIA decided to realize its monopoly position before the competition began to emerge. In this way, they hope to ensure industry leadership in the next 1-2 years, so their TItan V is priced at $3,000!
Although TItan V's deep learning core Tensor Core has unique performance, but the price/performance ratio is too bad, it makes the market less attractive. At this stage, there is no other choice, so at least what is currently used? .
AMD's hardware level has already surpassed NVIDIA, and they plan to develop matching deep learning software. If this step is achieved, its price/performance ratio will easily surpass NVIDIA and become a new benchmark in this field. At that time, NVIDIA will fight for the market with its strong financial strength, so we may see very cheap NVIDIA products in the future. Note that this situation is based on AMD's launch of high-quality software - if AMD jumps, it loses the chance to grab the crown, and NVIDIA's products will remain at high prices.
There is another new contender on the market: Intel Nervana's Neural Network Processor (NNP). It is still relatively competitive with several unique features that cater to the needs of CUDA developers. The NNP processor can solve most of the problems in the CUDA kernel that optimizes deep learning. This chip is truly the first deep learning chip.
In general, for a single chip ranking, we will follow the order of Nervana AMD NVIDIA, because NVIDIA chips have to maintain balance in games, deep learning and high-performance computing, AMD also needs to consider both games and Deep learning, only Nervana can focus on deep learning, which is a huge advantage, making their chips less useless structural design than the other two.
However, the winners often do not depend on pure performance or cost-effectiveness, but rather on a cost-effective + peripheral ecology + deep learning framework.
Let's take a closer look at the pros and cons of the three companies' products and see where they are.
Plastic Capillary Pressure Gauge
Plastic Capillary Pressure Gauge,Mini-Sized Capillary Manometer,D28 Plastic Tube Pressure Gauge,Pipe Manometer
ZHOUSHAN JIAERLING METER CO.,LTD , https://www.zsjrlmeter.com