“Add GPU acceleration to any Amazon EC2 instance for faster inference at much lower cost (up to 75% savings)”
So you’ve just kicked off the training phase of your multilayered deep neural network. The training phase is leveraging Amazon EC2 P3 instances to keep the training time to a minimum, but it’s still going to take a while. With time in hand, you begin to contemplate what infrastructure you’ll use to run your inferences.
You’re already familiar with the merits of using GPUs for the training phase. GPUs have the ability to parallelize massive amounts of simple math computations, which makes them perfect for training neural networks. GPUs are more expensive to run than CPUs, but because they can parallelize the number crunching, you don’t need to run them as long as you would the equivalent training performed on CPUs. In fact, training on GPUs can be orders-of-magnitude quicker. So it may cost you more per hour to run a GPU, but you won’t need to run it anywhere nearly as long when on a CPU. Besides factoring in cost, training your models faster allows you to get them into production quicker to perform inferences. So in terms of the training phase, it makes complete sense to go with GPUs.
So your contemplation now focuses on whether to use GPU or CPU infrastructure to perform inferencing once the training completes and your model is ready. We know that GPUs cost more per hour to run. Performing inferences through a trained neural network are far less taxing in terms of required computation and data volume that needs to be ingested and processed. Therefore, CPUs seem to be the way to go. However, you know from past experiences that over time, your CPU hosted inferencing tends to bottleneck due to overwhelming demand and this makes you reconsider running the inferencing on GPUs, but you now need to budget in the extra cost as a project consideration. This dilemma of whether to use GPUs versus CPUs for inferencing, with respect to both cost and performance is all too familiar for many organizations. The choice of using a GPU or CPU was a fairly mutually exclusive upfront decision made when using EC2. As of today, this is no longer the case.
Amazon Elastic Inference is a new service from AWS which allows you to complement your EC2 CPU instances with GPU acceleration, which is perfect for hosting your inferencing models. You can now select the appropriate CPU sized EC2 instance and boost its number crunching ability with GPU processing. Like with many other AWS services, you only pay for the actual accelerator hours you use. What this means is that you can get full GPU processing power but being up to 75% cheaper than running an equivalent GPU sized EC2 instance.
(You might also want to read up on this year’s announcements from re:Invent, particularly our blog post on how Amazon FSx for Lustre Makes High Performance Computing More Accessible.)
For starters, Amazon Elastic Inference is launching with 3 types of Teraflop mixed precision powered accelerators: eia1.medium, eia1.large, and the eia1.xlarge
Amazon Elastic Inference has been seamlessly integrated into both the AWS EC2 console and the AWS CLI. In the following EC2 console screenshot, attaching GPU acceleration, is as simple as enabling the “Add an Elastic Inference accelerator” option:
The equivalent AWS CLI command looks like the following, noting that the existing API has been extended with a new optional elastic-inference-accelerator parameter:
aws ec2 run-instances \ --image-id ami-00ffbd996ef2211e3 \ --key-name DNN_Key --security-group-ids sg-12345678 \ --subnet-id subnet-12345678 \ --instance-type c5.xlarge \ --elastic-inference-accelerator Type=eia1.large --iam-instance-profile Name="InferenceAcceleratorProfile"
The following list itemizes several prerequisites that need to be in place to leverage Amazon Elastic Inference:
- A Private Link endpoint configured for Elastic Inference must be present
- An IAM role with the necessary policies to connect to the Elastic Inference accelerator
- Build your models using TensorFlow, Apache MXNet, and/or ONNX
- Use the latest AWS Deep learning AMIs, which have been updated with Amazon Elastic Inference support baked directly into the TensorFlow, Apache MXNet deep learning frameworks
As you can see with a few extra configuration options in place you can have the best of both worlds, CPU hosted inferencing with GPU acceleration. You no longer need to spend time contemplating CPUs over GPUs – take both!!
Another game changer in the machine learning space from AWS – give it a try and check out our Lab on Analyzing CPU vs GPU Performance for AWS Machine Learning.
New on Cloud Academy: AWS Solution Architect Lab Challenge, Azure Hands-on Labs, Foundation Certificate in Cyber Security, and Much More
Now that Thanksgiving is over and the craziness of Black Friday has died down, it's now time for the busiest season of the year. Whether you're a last-minute shopper or you already have your shopping done, the holidays bring so much more excitement than any other time of year. Since our...
Understanding Enterprise Cloud Migration
What is enterprise cloud migration? Cloud migration is about moving your data, applications, and even infrastructure from your on-premises computers or infrastructure to a virtual pool of on-demand, shared resources that offer compute, storage, and network services at scale. Why d...
6 Reasons Why You Should Get an AWS Certification This Year
In the past decade, the rise of cloud computing has been undeniable. Businesses of all sizes are moving their infrastructure and applications to the cloud. This is partly because the cloud allows businesses and their employees to access important information from just about anywhere. ...
AWS Regions and Availability Zones: The Simplest Explanation You Will Ever Find Around
The basics of AWS Regions and Availability Zones We’re going to treat this article as a sort of AWS 101 — it’ll be a quick primer on AWS Regions and Availability Zones that will be useful for understanding the basics of how AWS infrastructure is organized. We’ll define each section,...
Application Load Balancer vs. Classic Load Balancer
What is an Elastic Load Balancer? This post covers basics of what an Elastic Load Balancer is, and two of its examples: Application Load Balancers and Classic Load Balancers. For additional information — including a comparison that explains Network Load Balancers — check out our post o...
Advantages and Disadvantages of Microservices Architecture
What are microservices? Let's start our discussion by setting a foundation of what microservices are. Microservices are a way of breaking large software projects into loosely coupled modules, which communicate with each other through simple Application Programming Interfaces (APIs). ...
Kubernetes Services: AWS vs. Azure vs. Google Cloud
Kubernetes is a popular open-source container orchestration platform that allows us to deploy and manage multi-container applications at scale. Businesses are rapidly adopting this revolutionary technology to modernize their applications. Cloud service providers — such as Amazon Web Ser...
AWS Internet of Things (IoT): The 3 Services You Need to Know
The Internet of Things (IoT) embeds technology into any physical thing to enable never-before-seen levels of connectivity. IoT is revolutionizing industries and creating many new market opportunities. Cloud services play an important role in enabling deployment of IoT solutions that min...
Which Certifications Should I Get?
As we mentioned in an earlier post, the old AWS slogan, “Cloud is the new normal” is indeed a reality today. Really, cloud has been the new normal for a while now and getting credentials has become an increasingly effective way to quickly showcase your abilities to recruiters and compan...
How to Go Serverless Like a Pro
So, no servers? Yeah, I checked and there are definitely no servers. Well...the cloud service providers do need servers to host and run the code, but we don’t have to worry about it. Which operating system to use, how and when to run the instances, the scalability, and all the arch...
AWS Security: Bastion Hosts, NAT instances and VPC Peering
Effective security requires close control over your data and resources. Bastion hosts, NAT instances, and VPC peering can help you secure your AWS infrastructure. Welcome to part four of my AWS Security overview. In part three, we looked at network security at the subnet level. This ti...
Top 13 Amazon Virtual Private Cloud (VPC) Best Practices
Amazon Virtual Private Cloud (VPC) brings a host of advantages to the table, including static private IP addresses, Elastic Network Interfaces, secure bastion host setup, DHCP options, Advanced Network Access Control, predictable internal IP ranges, VPN connectivity, movement of interna...