Union Serverless Broadens Support for NVIDIA Accelerated Computing

We are excited to announce that Union Serverless users can now harness the power of NVIDIA A100 Tensor Core GPUs and NVIDIA L4 Tensor Core GPUs, in addition to our existing support for NVIDIA T4 GPUs, with NVIDIA H100 Tensor Core GPUs coming soon.

Modern machine learning development requires access to production-grade GPU hardware optimized for parallelized computation and high throughput. Union Serverless makes it easier than ever to develop and scale your ML applications with expanded GPU access that now includes NVIDIA A100 GPUs. Use the Union Serverless platform to first develop locally in Python and then seamlessly orchestrate complex jobs and efficiently run distributed training in the cloud.

The Power of Union Serverless

We created Union Serverless to make building, developing, and shipping your AI workflows simple and intuitive. With Union Serverless, we power a diverse set of AI workloads while handling the provisioning of underlying compute infrastructure, efficiently streamlining data flow and optimizing performance with a pay-as-you-go pricing model. Focus on creating value for your business by training and developing AI models, without having to worry about the hassle of infrastructure provisioning and management.

To take advantage of Union Serverless, simply start off by writing and developing ML workflows in Python on your machine. Declare your infrastructure requirements natively in your code and, with one command, seamlessly switch to running your workflows on real datasets at scale in the cloud. Union Serverless securely and efficiently manages orchestrating your workflow. No infrastructure expertise necessary, no messing with configurations, and no procuring and managing hardware or refactoring your code to work in the cloud. 

Today’s announcement about support for NVIDIA A100 GPUs follows our recent integration with NVIDIA NIM microservices in the open-source Flytekit SDK. Developers can easily run generative AI models accelerated by NVIDIA NIM microservices broadly across the open-source Flyte platform and Union BYOC. 

NVIDIA A100 GPU: A Key Tool

NVIDIA A100 GPUs, with their large 40GB on-chip memory and small energy footprint, are workhorses for modern machine learning applications. A100 GPUs are optimized for deep learning and AI workloads like building out inference workflows, complex neural networks, running massive simulations, parallel processing, and so much more. Given their versatility and power, A100 GPUs are often in high demand. That’s why we’re excited to announce that Union users now have instant access to A100 GPUs on Union Serverless. Your first $30 of compute is free, and afterwards, you only pay for what you use.

Using NVIDIA A100 GPUs on Union Serverless

Using a GPU with Union Serverless is as simple as specifying a GPU in your Python function where you define your training code:

@task(
    limits=Resources(gpu="1")
)
def training_task():
  ...

Specify an A100 GPU with just one more line of Python

@task(
    requests=Resources(gpu="1"),
    accelerator=GPUAccelerator("nvidia-tesla-a100"),
)
def training_my_task():
    ...

Pay for What You Use 

Union offers flexible pricing options tailored to your computational needs. Whether you require CPU, GPU, or memory resources, we’ve got you covered. And the best part — you only pay for what you use. Additionally, we now offer access to NVIDIA L4 Tensor Core GPUs and already offer NVIDIA T4 GPUs, with support for NVIDIA H100 GPUs coming soon.

Getting Started

Get started with a free trial to begin developing on NVIDIA A100 GPUs on Union Serverless. Sign up at signup.union.ai with your existing GitHub account – no credit card required. Check out the Serverless docs here and learn more about using accelerators here.

For specific examples on how to leverage NVIDIA A100 GPUs to build out rich ML applications, see:

• Credit Default Prediction with XGBoost & NVIDIA RAPIDS
• Forecasting with GluonTS & PyTorch on GPUs
• HDBSCAN Soft Clustering With Headline Embeddings on GPUs

If you have any questions or need assistance, our support team is here to help. Contact us through the chat option on serverless.union.ai or reach out to our team: support@union.ai 

Happy workflow building!