First, it was bookstores, then virtually every brick-and-mortar retailer, then the cloud. And now Amazon has Intel in its gunsights. At its annual re:Invent conference yesterday in Las Vegas, Amazon Web Services unveiled the Graviton2, a next-generation datacenter processor that Amazon says will power AWS services including Amazon EMR, Elastic Load Balancing, and Amazon ElastiCache. The Graviton2 is a custom AWS design built using a 7 nanometer manufacturing process based on 64-bit Arm Neoverse cores. AWS said the Graviton2 is seven times faster than its first-generation A1 Graviton processors “with four times the compute cores and five times faster memory.” By using the Graviton2 to power its datacenters, Amazon is hoping to lessen its reliance on processors made by Intel and AMD. Intel holds a 90 percent share of the datacenter processor market, with AMD controlling most of the rest, according to Reuters.
Unveiling the processor at his keynote address at re:Invent, AWS chief executive Andy Jassy said customers can “run virtually all your instances on this, with a 40 percent better price / performance than x86 instances.”
What does this mean for AWS customers? For starters, it should mean higher performance and lower costs. AWS chief evangelist Jeff Barr said, “All of these performance enhancements come together to give these new instances a significant performance benefit over the 5th generation (M5, C5, R5) of EC2 instances.” AWS released some eye-popping initial benchmarks that recorded the following per-vCPU performance improvements over M5 instances:
- SPECjvm 2008: +43 percent (estimated).
- SPEC CPU 2017 integer: +44 percent (estimated).
- SPEC CPU 2017 floating point: +24 percent (estimated).
- HTTPS load balancing with Nginx: +24 percent.
- Memcached: +43 percent performance, at lower latency.
- X.264 video encoding: +26 percent.
- EDA simulation with Cadence Xcellium: +54 percent.
Barr noted that AWS is working on three types of Graviton2-powered EC2 instances:
- General purpose: 1-64 vCPUs and up to 256GiB of memory.
- Compute-optimized: 1-64 vCPUs and up to 128GiB of memory.
- Memory-optimized: 1-64 vCPUs and up to 512GiB of memory.
The instances will have up to 25Gbps of network bandwidth, 18Gbps of EBS-optimized bandwidth, and will also be available in bare metal form.
Barr gave no time frame for when the Graviton2 will be rolled out except to say, “I will have more information to share with you in 2020.”
And what does this mean for Intel? Because Amazon and other major cloud operators have mostly been using Intel Xeon chips, Graviton2 could eat into Intel’s lucrative server processor market.
Featured image: Shutterstock