Habana Labs, an Intel company, and Genesis Cloud are collaborating to deliver a new class of cloud instances with Habana® Gaudi®2 accelerators to enable high-performance, high-efficiency deep learning training and inference workloads in the cloud. This week the Genesis Cloud team will demonstrate the performance of Gaudi2-based instances at Collision 2023.
Habana Gaudi2 accelerators bring to Genesis Cloud AI infrastructure proven training and inference performance that delivers roughly two times that of Nvidia A100 GPUs on large language and multi-modal models: inference that’s 1.42x on 176 billion1 and 2.89x on 7 billion parameter BLOOMZ2, 2.44x for fine-tuning T5-3B3, and 2.84x for inference on Stable Diffusion4. With robust performance, efficiency and near-linear scaling6, Gaudi2 helps Genesis Cloud offer customers highly compelling performance-per-price, enabling more training and inference compute for less
Through its collaboration with Habana to offer Gaudi2-based instances, Genesis Cloud aims to help customers accelerate their innovation with affordable and easy to use deep learning model training and deployment. Gaudi2-based instances enable customers to run a wide array of language and vision-based AI applications as well as the newly emergent generative AI applications like large language models, increasingly used in numerous industries—such as retail, construction, healthcare, security, manufacturing—to name a few.
Habana has made it easy to build and deploy new models or migrate existing models with a few lines of code to get started. Our SynapseAI® Software Stack is designed to optimize performance, efficiency and ease of use with Gaudi hardware. SynapseAI integrates PyTorch and TensorFlow frameworks and Kubernetes orchestration and provides a host of tools and how-to content to make getting started easy. To support a vast set of customer AI applications, we’ve implemented the Habana Optimum Library on the Hugging Face hub, making it easy for customers to enable more than 50,000 AI models.
Habana and Genesis Cloud share a commitment to energy-efficiency to mitigate environmental impacts. Genesis Cloud infrastructure is powered by 100% renewable energy and evaluation by server manufacturer, Supermicro, shows 2 times the performance-per-Watt of the Gaudi2 server employed in Genesis Cloud as compared to the comparable Supermicro server based on the Nvidia A100 GPU.5 This translates to Gaudi2-based instances operating on half the power as consumed by competitor solutions to train or deploy a variety of models.
Genesis Cloud will demonstrate Gaudi2-based instances at the Collision Conference, June 27-29 at the Enercare Centre in Toronto Visit in booth E309.
To learn how your organization can run AI and deep learning workloads on Genesis Cloud with instances based on Gaudi2 accelerators, visit the Genesis Cloud booth at the Collision Conference 2023 and register to test Gaudi2 on the Genesis Cloud at https://gnsiscld.co/gaudi_2. Alternatively, reach out to [email protected] with questions or requests.
[1, 2] https://huggingface.co/blog/habana-gaudi-2-bloom#latency
[3, 4] https://huggingface.co/blog/habana-gaudi-2-benchmark
 Power performance measurements performed by Supermicro in their lab (April 2023). System configurations below. Results may vary. For test details see: https://habana.ai/habana-claims-validation/
[6 For details, see: https://habana.ai/habana-claims-validation/]