Wednesday, November 9th 2016

Qualcomm Starts Shipping 48-Core Centriq 2400 Processors

At a press conference held today in San Jose, Calif., Qualcomm Datacenter Technologies, Inc., a subsidiary of Qualcomm Incorporated (NASDAQ: QCOM), officially announced commercial shipment of the world's first and only 10 nanometer server processor series: the Qualcomm Centriq 2400 processor family. The Qualcomm Centriq 2400 processor family is the first high-performance Arm-based processor series designed to offer groundbreaking throughput performance for cloud workloads running in today's datacenters. Purpose built for cloud, the Qualcomm Centriq 2400 server processor family delivers exceptional performance-per-watt and performance-per dollar.

"Today's announcement is an important achievement and the culmination of more than four years of intense design, development and ecosystem enablement effort," said Anand Chandrasekher, senior vice president and general manager, Qualcomm Datacenter Technologies, Inc. "We have designed the most advanced Arm-based server processor in the world that delivers high performance coupled with the highest energy efficiency, enabling our customers to realize significant cost savings."
Qualcomm Centriq 2400 Processor Family Technical Specs
The Qualcomm Centriq 2400 processor family is a single chip platform-level solution built using Samsung's 10 nanometer FinFET process with 18 billion transistors on only 398 mm². It contains up to 48 high-performance, 64-bit, single-thread cores, running at up to 2.6 GHz frequency. The cores are connected with a bi-directional segmented ring bus with 250GB/s of aggregate bandwidth to avoid performance bottlenecks under full load. To maximize performance under various use cases, the design has 512KB of shared L2 cache for every two cores, and 60 MB of unified L3 cache distributed on the die. It has 6 channels of DDR4 memory and can support up to 768 GB of total DRAM capacity with 32 PCIe Gen3 lanes and 6 PCIe controllers. The Qualcomm Centriq 2400 processor family also supports Arm's TrustZone secure operating environment, and supports hypervisors for virtualization. The Qualcomm Centriq 2400 is able to achieve exceptional performance, while consuming less than 120 watts.

With a list price of $1,995, the 48 core Qualcomm Centriq 2460 processor offers greater than 4X better performance per dollar and up to 45% better performance per watt versus Intel's highest-performance Skylake processor, the Intel Xeon Platinum 818.

A number of cloud service providers and technology companies participated in the launch event and demonstrated applications of an ARM-based datacenter running on the Qualcomm Centriq 2400 series. These included Alibaba, LinkedIn, Cloudflare, American Megatrends Inc., Arm, Cadence Design Systems, Canonical, Chelsio Communications, Excelero, Hewlett Packard Enterprise, Illumina, MariaDB, Mellanox, Microsoft Azure, MongoDB, Netronome, Packet, Red Hat, ScyllaDB, 6WIND, Samsung, Solarflare, Smartcore, SUSE, Synopsys, Uber, and Xilinx.

Optimized for common cloud workloads, the Qualcomm Centriq 2400 processor family has been designed to deliver throughput performance for highly threaded cloud native applications that are developed as micro-services and deployed for scale-out. Some of the key applications and use cases demonstrated today include:
  • Web front end with HipHop Virtual Machine
  • NoSQL databases including MongoDB, Varnish, Scylladb
  • Cloud orchestration and automation including Kubernetes, Docker, metal-as-a-service
  • Data analytics including Apache Spark
  • Deep learning inference
  • Network function virtualization
  • Video and image processing acceleration
  • Multi-core electronic design automation
  • High throughput compute bioinformatics
  • Neural class networks
  • OpenStack Platform
  • Scaleout Server SAN with NVMe
  • Server-based network offload
Add your own comment

8 Comments on Qualcomm Starts Shipping 48-Core Centriq 2400 Processors

This is the real bomb, GPU are a side story. Intel is getting pushed for the first time in well over a decade on the top end, Power9 & SPARC M8, as well as mid & bottom tier, with EPYC & now QC server chips.
Posted on Reply
R0H1T said:
This is the real bomb, GPU are a side story. Intel is getting pushed for the first time in well over a decade on the top end, Power9 & SPARC M8, as well as mid & bottom tier, with EPYC & now QC server chips.
Posted on Reply
"Hip Hop Virtual Machine"? Really?

I thought Power and Sparc always kicked Intel's ass.
Posted on Reply
StrayKAT said:
"Hip Hop Virtual Machine"? Really?

I thought Power and Sparc always kicked Intel's ass.
Huge difference in CPU architecture, Intel had to offer a "Complex Instruction Set/Out of Order Execution" CPU since the primary target was running more than just one application or database, Power and Sparc were platform specific with "Reduced Instruction Set/Out-In Order Execution" where specialized hardware within the chip allowed common tasks to run on dedicated hardware VS desktop computers ran any application and had to use general purpose computing architecture where sheer clock rate and latency combined with breaking a large command (like multiplication, division etc...) down may have required 8 cycles to complete instead of 2 on dedicated hardware. Microsoft and Apple were the fruition of the speed increases allowed by Intel and the microchip industry that was emerging at the time, only server software really existed until machines were fast enough to give most people a GUI. /tangent
Posted on Reply
48 cores and only six memory channels? I hope those caches are well designed.
Posted on Reply
It's honestly hilarious how many markets Intel has given up by sitting on it's laurels. If Intel had never screwed AMD over back in the day with it's "rebate" program, they would have likely been forced to continue pushing their CPU designs. Instead, they stuck with 4 core in the mainstream until they were forced by AMD to go higher. Ditto goes for letting ARM gain mass market adoption and now it looks like they are going after Intel's bread and butter, the server market. They also completely missed the AI learning market, which is now owned by Nvidia. Intel seems to only be able to win when they are able to bully or cheat the competition out of the market. Just look at how they recently poached Raja Kudori from AMD. Intel makes me sick and they deserve all that's coming to them.
Posted on Reply
If the layout of the actual die is anywhere near the diagram in the picture, that is an greatly symmetrical and efficient design.
Posted on Reply
Could be. ARM has a very modern and clean design compared to all that x86 junk needed for legacy and compatibility.
Posted on Reply