AMD is re-boosting its server business with the launch of a new 32-core "Naples" chip. The Naples chip is based on Zen architecture and will be launched in Q2.
According to PC World, the new Naples chip is an attempt by AMD to regain its foothold back in the server market, currently dominated by competitor brand Intel. Intel reportedly has a 90 percent market share, and it will be a big test for AMD in the market. The price of AMD chipsets might be of some help in this respect, giving customers bargaining power over the highly priced chipsets made by Intel.
The Naples chip is based on Zen architecture and is a mega chip with 32 cores which will not be of use for everyone. So the company has already planned to release more chips with fewer cores for common usage. AMD's top management claims that the Naples chip will have more input/output (I/O) than comparable Intel chips. AMD hopes to pair the Naples chip with its upcoming Vega GPU, which will be used for machine learning.
According to Anandtech, the Naples chip will be infused with eight channels and two DIMMs (Dual In-line Memory Module) per channel. This will enable each Naples chip to support up to 2TB of DRAM (Dynamic Random Access Memory) per socket, making it 4TB in a single server. This will be a great boost for any server and Intel is surely going to face stern competition.
Reportedly, the Naples chip will also be using four of AMD's Zeppelin dies (Ryzen dies) in a single package. This will lead to the Naples chip having around 19.2 billion transistors, which is far bigger than any other chip currently present in the market. Besides, the "Infinity fabric" going through the Naples chip is also a major point of discussion in the technological sphere.
As per experts, the Naples chip by AMD will be taking the "Skylake chip" by Intel head on, scheduled to be used by Google in its cloud servers. The Skylake chip will have AVX-512 to run "vectorized" applications, while Naples chip is limited at AVX-128, which might be a detriment for the AMD chipset.
