OnQ

Bringing acceleration technology closer to the CPU

2016年5月23日

Qualcomm products mentioned within this post are offered by Qualcomm Technologies, Inc. and/or its subsidiaries.

Acceleration technology has proven to bring value in addressing today’s challenging data center workloads, driven by the incessant demand to optimize performance relative to cost and power. At the same time, ARM-based solutions from Qualcomm Technologies are bringing a new breed of processors, driven by the same demands of optimizing performance relative to power and its impact on cost.

This combination of processing and acceleration is bringing new levels of performance to emerging applications.

Cache Coherent Interconnect for Accelerators (CCIX) brings acceleration technology closer to the CPU. 

The standard interconnect for processors and peripheral devices today is on its current evolutionary trajectory to higher bandwidth as it moves to ever higher signaling rates. This will enable the next generation of devices such as NICs (Network Interface Cards) and HBAs (Host Bus Adapters), which demand higher throughput. However, acceleration technology has a potential for optimum performance when data movement is minimized and requires a different solution, one that brings to the table the benefits of coherency. While proprietary interconnect technologies exist to connect to accelerators, we’ve seen the adoption constrained by the practical hurdles imposed by proprietary technology. 

A new acceleration framework is needed, one that will provide the benefits of cache coherency and high performance in an open framework. Representing a milestone in the industry, a single interconnect  technology is being developed that will provide exactly that open framework, bringing the benefits of cache coherency to the developers of next generation server solutions. 

At Qualcomm, we're excited to be working with other recognized industry leaders in catalyzing change. We envision the proliferation of accelerated solutions in areas such as big data analytics to machine learning to Network Function Virtualization, as open standards and choice in vendors allow for rapid development without the risk of being locked into a specific solution or vendor.

Learn more about CCIX Consortium here

Engage with us on

and