WikiChip Fuse
A Look at Spring Crest: Intel Next-Generation DC Training Neural Processor

A look at the microarchitecture of Intel Nervana next-generation data center training neural processor, codename Spring Crest.

Read more
Marvell Lays Out ARM Server Roadmap

Marvell outlines its current and future Arm server microprocessor roadmap, aiming at a 2-year cadence with greater than 2x performance gen-over-gen.

Read more
Nvidia Inference Research Chip Scales to Dozens of Chiplets

Nvidia recently presented a research chip comprising dozens of chiplets that enables them to scale from milliwatts to hundreds of watts in order to cater to different markets such as edge, mobile, automotive, and data center.

Read more
Intel’s Spring Crest NNP-L Initial Details

An initial look into Intel’s upcoming Nervana Neural Network Processor (NNP) accelerators.

Read more
A Look at NEC’s Latest Vector Processor, the SX-Aurora

A look at the NEC SX-Aurora, their latest vector processor – increasing compute while maintaining a high B/F through six HBM2 modules leveraging TSMC 2nd gen CoWoS technology. The SX-Aurora introduces a new form factor, system architecture, and execution model.

Read more
Ampere Ships First Gen ARM Server Processors

Partnering up with Lenovo, ARM server startup Ampere has announced their first-generation eMAG microprocessors are now shipping.

Read more
A Look at Cavium’s New High-Performance ARM Microprocessors and the Isambard Supercomputer

Reaching general availability, here is a look at Cavium’s new ThunderX2 family of high-performance ARM microprocessors for the data center.

Read more
Cambricon Reaches for the Cloud With a Custom AI Accelerator, Talks 7nm IPs

Cambricon has announced their first high-performance and high-power AI accelerator for the data center in an effort to gain market share in the growing Chinese AI market.

Read more
The 2,048-core PEZY-SC2 sets a Green500 record

PEZY-SC2 is PEZY’s newest 16nm a many-core processor featuring 2,048 cores that is powering the world’s most power-efficient supercomputer. The chip is the first to use a unique 3D packaging interconnect technology known as TCI in order to achieve a memory bandwidth of over 2 TB/s.

Read more