AWS unveils R8g EC2 Instances with 4th-Generation Graviton Processors

At the latest re:Invent, AWS unveiled a preview of R8g instances optimized for memory and fueled by the fourth-generation Graviton processors. These new instances cater to memory-intensive tasks such as databases, in-memory caches, and real-time big data analytics.

As per the cloud provider's statement, Graviton processors offer comprehensive encryption for all high-speed physical hardware interfaces, boasting a remarkable 30% improvement in compute performance, 50% more cores, and 75% greater memory bandwidth compared to Graviton3 counterparts. The forthcoming R8g instances from this new generation will feature up to 96 Neoverse V2 cores, 2 MB of L2 cache per core, and 12 DDR5-5600 channels. Jeff Barr, Vice President and Chief Evangelist at AWS, further emphasizes, "Graviton4 processors also support all of the security features from the previous generations and include some important new ones including encrypted high-speed hardware interfaces and Branch Target Identification (BTI)."

As per the official press release, the statement declares that Graviton4 instances exhibit a potential 45% speed improvement for large Java applications compared to their Graviton3 counterparts. However, the announcement does not offer additional details or benchmarks. Liz Fong-Jones, Field CTO at Honeycomb, had the chance to assess the performance of these new instances. She comments," Our Go-based OpenTelemetry data ingestion workload required 25% fewer replicas on the Graviton4-based R8g instances compared to Graviton3-based C7g/M7g/R7g instances—and additionally achieved a 20% improvement in median latency and 10% improvement in 99th percentile latency."

The community's reaction has been mixed, with some users questioning the performance claims and others expressing concerns about the absence of pricing information. This is significant as Graviton3 instances have typically been more expensive than Graviton2 ones. User LunaSea comments on HackerNews, "Since Graviton3 still isn't available in most regions, especially on the RDS side, I'm really not holding my breath."

Graviton4 wasn't the only custom-built hardware showcased by AWS at the conference. Trainium2, the second-generation accelerator tailored for deep learning training of FMs and LLMs, is crafted to provide a substantial improvement with up to 4 times faster training performance and 3 times more memory capacity compared to the preceding generation chips. Trainium2 will be featured in Trn2 instances. Adam Selipsky, CEO at AWS, shares insights on X (formerly Twitter). "AWS’s chips were front and center at re:Invent where we announced Graviton4 and Trainium2. Graviton4 is the most powerful and energy-efficient chip we have ever built. Trainium2 will allow our customers to train their generative AI models even more quickly and with enhanced price performance. It was ten years ago that Amazon first decided to start developing its chips. I love to present the challenges for software developers dealing with AI services using a simple example and service on AWS."

Numerous conference sessions delved into Graviton processors, and you can now access the recorded sessions on YouTube. The c8g instances are currently in preview.