12.1.16 Today at AWS re:Invent, Amazon Web Services, Inc. (AWS), an Amazon.com company (NASDAQ:AMZN), announced the next generation of Amazon Elastic Compute Cloud (EC2) Memory Optimized, Compute Optimized, and High input/output (I/O) instances, and added two new hardware acceleration options to its range of compute services. The new F1 instance is the cloud’s first customer-programmable, hardware-accelerated compute instance with Field Programmable Gate Arrays (FPGAs). Amazon EC2 Elastic GPUs allow customers to easily attach low-cost, professional grade graphics acceleration to Amazon EC2 instances. AWS also announced Amazon Lightsail, a new way to get started with AWS that makes it easy to spin up powerful virtual private servers (VPS) that have bundled storage and networking with simple, monthly pricing. To get started with Amazon EC2, visit https://aws.amazon.com/new/reinvent/compute/.
“With ten instance families, AWS offers customers, by far, the broadest and deepest range of compute functionality to support an incredibly wide range of applications. From very low cost, burstable T2 general-purpose instances for websites and M4 instances for business applications like SAP, to our new P2 instances for machine learning and HPC applications – and everything in between – AWS delivers the right compute option for virtually any workload today,” said Matt Garman, Vice President, Amazon EC2. “Most companies have many diverse applications with varying requirements. They’ve told us having the ability to choose the right instance for the right workload allows them to have optimal price performance and move faster. Today, we expanded this range of options even further by adding a way to help customers get started quickly with Amazon Lightsail, introducing the next generation of our Memory, Compute, and Storage Optimized instance families, and delivering hardware acceleration with Elastic GPUs and FPGA-enabled F1 instances.”
New Memory Optimized, Compute Optimized and High I/O instances, and larger T2 instance sizes
Optimized to deliver the right combination of CPU, memory, storage, and networking capacity for a wide range of applications, the next-generation of AWS’s Memory Optimized, Compute Optimized, and High I/O instances include the latest processors and storage technologies. They also feature enhanced networking with AWS’s Elastic Network Adapter (ENA), a custom network interface optimized to deliver high Packet Per Second (PPS) performance, consistent low latency, and low network jitter. With ENA, larger instance sizes deliver total bandwidth above 20 Gbps, and smaller instance sizes offer 10 Gpbs peak network throughput. And, new larger sizes of T2 burstable instances enable web applications requiring larger memory footprints.
Available today are two larger sizes of T2 Burstable Performance Instances, which provide a baseline level of CPU performance with the ability to burst to full core performance when needed. The new t2.xlarge offers 16 GiB of memory and 4 vCPU, and the new t2.2xlarge offers 32 GiB of memory and 8 vCPU. Customers with existing T2 workloads can now scale up to the larger T2 sizes if desired.
Available today, R4 instances are designed for high performance databases, distributed memory caches, in-memory analytics, genome assembly and analysis, and other enterprise applications. They feature a larger L3 cache that is twice the size of the previous generation (R3), a new 16xlarge size that offers twice the memory as the previous generation with 488 GiB of fast, DDR4 memory, and 64 vCPUs (two times as many as the largest R3) – all for 20 percent less per GiB of RAM than the previous generation R3 instances.
Coming in early 2017, C5 instances include the next generation of the Intel Xeon Processor family (code named Skylake) with AVX 512 and up to 72 vCPUs (twice that of previous generation C4 compute-optimized instances), and 144 GiB of memory, making them the best price to compute performance of any Amazon EC2 instance. C5 instances also feature new AWS hardware acceleration that delivers three times the Amazon EBS bandwidth of C4 instances for workloads that require high amounts of IOPs. C5 instances are ideal for batch processing, distributed analytics, high performance science and engineering applications, ad serving, massively multiplayer online (MMO) gaming, and video encoding.
Coming in early 2017, I3 instances are ideal for the most demanding I/O-intensive relational databases, NoSQL databases, transactional systems, and analytics workloads. I3 instances feature twice the memory and vCPUs as the previous generation high I/O instance family (I2), and over two times the storage as the previous generation with 15.2 TB of fast, low latency locally attached storage backed by Non Volatile Memory Express (NVMe)-based SSDs. Additionally, I3 instances deliver up to nine times the IOPs as the previous generation with 3.3 million random IOPS at 4 KB block size, and offer total I/O throughput of 16 GB per second.
F1 instances – custom FPGA hardware acceleration for all
A growing set of applications such as genomics research, financial analysis, and video processing push the boundaries of what customers can do with a CPU. For these applications, developers can program acceleration algorithms that are custom-tailored for their specific workloads directly into silicon by using FPGAs. Until now, this type of custom hardware acceleration has been prohibitively expensive, time consuming, and limited to those who work at companies that can afford the significant fixed cost at scale. The new Amazon EC2 F1 instance is the first cloud instance with programmable hardware for FPGA application acceleration. Customers can program their own FPGAs to increase performance by as much as 30x over general-purpose CPUs. The F1 instance FPGA Developer AMI and Hardware Developer Kit include everything a developer needs to develop, simulate, debug, and compile hardware acceleration code. Once the FPGA design is complete, developers can save it as an Amazon FPGA Image (AFI) and deploy it to an F1 instance with just a few clicks. Developers can also bring their own FPGA designs, or go to the AWS Marketplace to find pre-built AFIs that include common hardware accelerations. FPGAs are connected to F1 instances through a dedicated, isolated network fabric and are not shared across instances, users, or accounts, making them fast and secure.
Maxeler’s Multiscale Dataflow Computing (MDC) technology delivers the next generation of algorithms at a cost comparable to lower fidelity approaches. “Maxeler helps our clients leverage our dataflow computing infrastructure to rapidly create application accelerators running on FPGAs. In the dataflow approach to processing, pipelines of tasks exploit temporal and spatial parallelism, offering dramatic performance improvements,” said Oskar Mencer, Chief Executive Officer, Maxeler Technologies. “With a strong research division and many Dataflow applications developed using our tools, the time and expense of deploying hardware appliances on-premises becomes the bottleneck. With Amazon FPGA Images and F1 instances integrated into the AWS Marketplace, we can now publish our Dataflow applications to over a million AWS customers who can deploy them with one click. This substantially simplifies bringing new innovations to market and enables us to focus on new algorithms and supporting customers.”
NGCodec is a provider of video compression technology. “Cloud video encoding is exploding, but traditional software approaches need massive CPU resources and cannot deliver the video quality and latency necessitated by new applications. We have developed a next generation video encoder that, using FPGA hardware acceleration, offers low latency and low cost while maximizing video quality to keep up with today’s broadcasting standards,” said Oliver Gunasekara, Chief Executive Officer and co-founder, NGCodec. “AWS is the first cloud provider to add FPGA instances, which provide massive acceleration. In just three weeks, we ported our RealityCodec™ H.265/HEVC encoder to the new F1 instance type to enable AWS Cloud customers to leverage the benefits of higher video quality, lower latency, and lower cost for live H.265/HEVC video encoding. Updating our install base globally will be as simple as publishing a new Amazon FPGA image and re-launching F1 instances.”
CME Group is one of the world’s leading derivatives marketplace offering risk management solutions, including futures and options, equity indexes, foreign exchange, energy, agricultural products, and metals. “Large-scale financial clearing risk management is essential to delivering value for our customers,” said Kevin Kometer, Chief Information officer, Chief Information Officer, CME Group. “CME Group has long been an innovator in the use of accelerated computing, for the clearning risk management of increasingly complex instruments, including extensive research into FPGAs. Amazon EC2 F1 instances will allow us to substantially accelerate rate of innovation of risk analysis for our customers, while delivering greater cost efficiency relative to using traditional IT infrastructure.”
Elastic GPUs – high performance graphics acceleration for Amazon EC2 Instances
Amazon EC2 provides two families of GPU-based instances today: G2 instances for graphics-intensive applications like 3D application streaming, video encoding, and other server-side graphics workloads, and P2 instances for general-purpose GPU compute applications like machine learning, high performance databases, and other server-side workloads requiring massive parallel floating point processing power. While these instances are very popular, there are times when customers need a small amount of GPU for graphics acceleration, or have applications that could benefit from some GPU, but also require high amounts of compute, memory, or storage, making them less suited to the G2 or P2 instances. Elastic GPUs is a new Amazon EC2 feature that provides flexible, workstation-quality GPUs that customers can attach to existing Amazon EC2 instance types in much the same way as customers can attach Amazon EBS volumes. With Amazon Elastic GPUs, customers can configure the right amount of graphics acceleration for their particular workload without being constrained by fixed hardware configurations and limited GPU selection. Amazon Elastic GPUs are planned to support OpenGL and offer up to 8 GiB of GPU memory, making them ideally suited for any workload that needs additional GPUs such as gaming, industrial design, HPC visualization, 3D modeling, rendering, or virtual desktops.
ANSYS is a global leader in engineering simulation. “ANSYS Enterprise Cloud delivers a virtual simulation data center optimized for AWS. It delivers a rich interactive graphics experience critical to supporting the end-to-end engineering simulation processes that allow our customers to deliver innovative product designs. With Amazon EC2 Elastic GPUs, ANSYS will be able to more easily deliver this experience right-sized to the price and performance needs of our customers,” said Ray Milhem, Vice President, Enterprise Solutions & Cloud, ANSYS, Inc. “We are certifying ANSYS applications to run on Elastic GPUs to enable our customers to innovate more efficiently on the cloud.”
Siemens PLM Software is a world-leading provider of product lifecycle management and manufacturing operations management software. “Amazon EC2 Elastic GPUs is a game-changer for Computer Aided Design (CAD) in the cloud,” said Bob Haubrock, Vice President, NX Product Management, Siemens PLM. “With Elastic GPUs, our customers can now run Siemens PLM NX on Amazon EC2 with professional-grade graphics, and take advantage of the flexibility, security, and global scale that AWS provides. Siemens PLM is excited to certify NX on the Amazon EC2 Elastic GPUs platform to help our customers push the boundaries of design and engineering innovation.”
Amazon Lightsail – Powerful VPS made easy, starting at $5 per month
Sometimes, developers just want a server in the cloud to deploy their code or run their software, and Amazon Lightsail provides the easiest way for them to launch a VPS with just a few clicks. Amazon Lightsail includes everything customers need for projects such as web sites, blogs, custom applications, or development servers. Customers can get started with just three clicks; they simply choose a virtual private server image from a menu of over 10 images that includes Linux, WordPress, Drupal, and others, and select one of five server sizes. Amazon Lightsail automatically launches the VPS, attaches the SSD-based storage, sets up the security groups, creates a Virtual Private Cloud (VPC), and then makes it simple for developers to set up the DNS and IP addresses. With Amazon Lightsail, customers can use preconfigured operating system templates or pre-installed popular application setups. Amazon Lightsail runs on the AWS Cloud, making it easy for customers to connect to the more advanced services when they are ready. Amazon Lightsail is available immediately, and customers pay a simple monthly fee for each virtual private server, starting at $5 per month.