Today at the Intel Vision conference, Habana Labs, an Intel company, announced its second-generation deep learning processors, the Habana® Gaudi®2 Training and Habana® Greco™ Inference processors. The processors are purpose-built for AI deep learning applications, implemented in 7nm technology and build upon Habana’s high-efficiency architecture to provide customers with higher-performance model training and inferencing for computer vision and natural language applications in the data center. At Intel Vision, Habana Labs revealed Gaudi2’s training throughput performance for the ResNet-50 computer vision model and the BERT natural language processing model delivers twice the training throughput over the Nvidia A100-80GB GPU.
“The launch of Habana’s new deep learning processors is a prime example of Intel executing on its AI strategy to give customers a wide array of solution choices – from cloud to edge – addressing the growing number and complex nature of AI workloads. Gaudi2 can help Intel customers train increasingly large and complex deep learning workloads with speed and efficiency, and we’re anticipating the inference efficiencies that Greco will bring.”
Sandra Rivera, Intel executive vice president and general manager of the Datacenter and AI Group
Habana® Gaudi®2 Training Processor
Habana Gaudi2 significantly increases training performance, building on the same high-efficiency first-generation architecture that enables customers with up to 40% better price performance in the AWS cloud with Amazon EC2 DL1 Instances and on-premises with the Supermicro X12 Gaudi Training Server. In addition to its 7nm leap from 16nm, Gaudi2 also features 24 Tensor Processor Cores, an increase from eight cores in the first Gaudi and designed expressly for large deep learning workloads. The new Gaudi2 AI training processor integrate media processing on-chip, triple on-board memory to 96 GB and double SRAM to 48 MB. These contribute to Gaudi2’s performance – up to three times the training throughput over first-generation Gaudi.
“Compared with the A100 GPU, implemented in the same process node and roughly the same die size, Gaudi2 delivers clear leadership training performance as demonstrated with apples-to-apples comparison on key workloads,” said Eitan Medina, chief operating officer at Habana Labs. “This deep-learning acceleration architecture is fundamentally more efficient and backed with a strong roadmap.”
Customer Benefits
Gaudi2 gives customers a high-performance deep learning training processor alternative for computer vision workloads, such as object detection in autonomous vehicles, object detection in medical imaging, and defect detection in manufacturing. It enables customers to train natural language processing workloads like subject matter analysis to identify fraud detection in sensitive documents such as insurance claims and grant submissions.
Networking Capacity, Flexibility, Efficiency
Habana has made it cost-effective and easy for customers to scale out training capacity by amplifying training bandwidth on Gaudi2. With the integration of 24 100 Gigabit RDMA over Converged Ethernet (RoCE) ports on chip, customers can easily scale and configure Gaudi2 systems to suit their deep learning cluster requirements. With system implementation on widely used industry-standard Ethernet connectivity, Gaudi2 enables customers to choose from a wide array of Ethernet switching and related networking equipment, resulting in overall cost savings. Avoiding proprietary interconnect technologies in the data center is important for IT decision makers who want to avoid single vendor “lock-in.” The on-chip integration of the networking interface controller (NIC) ports also lowers component costs.
Habana provides customers with the HLS-Gaudi®2 server, featuring eight Gaudi2 processors and dual-socket Intel® Xeon® Ice Lake CPUs. One thousand Gaudi2 processors have been deployed in the Habana Gaudi2 data center in Haifa, Israel to support research and development for Gaudi2 software optimization and to inform further advancements in forthcoming solutions.
Simplified model build and migration
The Habana SynapseAI® Software Suite is designed for deep learning model development and to ease migration of existing GPU-based models to Gaudi platform hardware. The SynapseAI stack features integration of TensorFlow and PyTorch frameworks and 30+ computer vision and natural language processing models. Developers are supported with documentation and tools, how-to content, a community support forum on the Habana Developer Site, and with reference models and model roadmap on the Habana GitHub.
Availability of Gaudi2 training solutions
Habana is partnering with Supermicro to bring the Supermicro Gaudi®2 Training Server to market in 2H 2022. Habana is also teamed with DDN® to deliver a turnkey server featuring the Supermicro Gaudi server with augmented AI storage capacity with the pairing of the DDN AI400X2 storage solution.