Announcing OCI Compute with NVIDIA L40S GPUs
Oracle Cloud Infrastructure (OCI) is excited to announce the forthcoming launch of the BM.GPU.L40S.4 instance with the NVIDIA L40S GPU based on the NVIDIA Ada Lovelace architecture. The NVIDIA L40S GPU includes fourth-generation Tensor Cores, aTransformer Engine, and support for the 8-bit floating-point (FP8) data format, delivering over 1.45 petaflops of tensor processing power to enable next-generation generative AI, LLM training, and inferencing.
The NVIDIA L40S is the highest-performing universal GPU, with 1.2-2x improved price performance, up to 1.2x more generative AI inference performance, and up to 1.7x better training performance compared to the NVIDIA A100 Tensor Core GPU.
The L40S GPU also provides powerful graphics and video capabilities andis flexible enough to support the following range of workloads:
- Large language model (LLM) training, fine-tuning, and inference
- Multimodal generative AI
- Omniverse and industrial digitalization
- 3D graphics and rendering
- Virtualized workloads
- FP32 HPC
- Streaming encoding and decode, AV1
OCI combines the L40S GPU with up to 112 cores of the latest-generation Intel Sapphire Rapids CPU, 1 TB of system memory, 15.36 TB of low-latency NVMe local storage for caching data, and 400Gb/sec of cluster network bandwidth.
Want to know more?
For more information on OCI’s capabilities, please visit us at GPU compute and AI infrastructure.