NVIDIA and Google Cloud Platform
NVIDIA Tesla T4 GPUs now available on Google Cloud Platform.
Watch this on-demand webinar to learn more about T4 GPUs on GCP.
Contact Us
Google Cloud Platform offers NVIDIA Tesla K80, P4, T4, P100, and V100 GPUs
Predicting our climate’s future. A new drug to treat cancer. Some of the world’s most important challenges need to be solved today, but require tremendous amounts of computing to become a reality.
Together, NVIDIA and Google Cloud are helping you achieve faster results to address these challenges — all without massive capital expenditures or complexity of managing the underlying infrastructure. The NVIDIA Tesla K80, P4, T4, P100, and V100 GPUs are all generally available on Google Cloud Platform. With NVIDIA GPUs on Google Cloud Platform, deep learning, analytics, physical simulation, video transcoding, and molecular modeling take hours instead of days. You can also leverage NVIDIA GRID virtual workstations on Google Cloud Platform to accelerate your graphics-intensive workloads from anywhere.
Benefits of using NVIDIA GPUs on Google Cloud Platform
Access some of the same hardware that Google uses to develop high-performance deep-learning products, without having to worry about the capital expenditures or IT operations of managing your own infrastructure. NVIDIA Tesla K80, P4, P100, T4, and V100 GPUs on Google Cloud Platform means the hardware is passed through directly to the virtual machine to provide bare metal performance.
Virtual workstations with NVIDIA GRID and Tesla P4, T4 and P100 GPUs enable creative and technical professionals to access demanding applications from the cloud.
NVIDIA GPUs available on Google Cloud Platform
NVIDIA Tesla K80 is generally available in the Google Cloud. It drastically lowers model-training times and HPC cost by delivering superior performance with fewer, more powerful server instances, engineered to deliver above 5–10x performance boost on real-world applications.
Over 550 industry-leading HPC applications already support NVIDIA GPUs, including all top 15 HPC applications and all deep-learning frameworks. With features like dual-GPU design and Dynamic GPU Boost, Tesla K80 is built to deliver superior performance for these applications.
NVIDIA Tesla P4 and T4 GPUs are now generally available on Compute Engine.
Inference platform
NVIDIA Tesla P4 and T4 GPUs boost the efficiency of scale-out servers running deep-learning workloads and enable responsive AI-based services. The P4 and T4 are designed to slash inference latency, while providing better energy efficiency. This helps unlock AI services that were previously impossible due to latency limitations.
Virtual workstations
NVIDIA Tesla P4 and T4 help provide designers, engineers, scientists, and graphic artists virtual workstations possessing the power to take on the biggest visualization challenges with immersive, interactive, photorealistic environments. With NVIDIA GRID, subject matter experts are now untethered from their desks and can access even the most demanding professional applications and data from nearly anywhere, on virtually any device.
Video transcoding
NVIDIA Tesla P4 can transcode and infer up to 18 Full HD video streams in real time and T4 up to 38 Full HD videos streams, powered by a dedicated hardware-accelerated decode engine that works in parallel with the GPU doing inference. By integrating deep learning into the video pipeline, customers can offer smart, innovative video services to users that were previously impossible to do.
NVIDIA Tesla P100 is generally available on Google Cloud Platform.
Unified supercomputing
Designed to boost throughput and save money for both HPC and ML applications. Powered by the NVIDIA Pascal architecture, each Tesla P100 delivers 4.7 and 9.3 teraflops of double-precision and single-precision performance for HPC and ML workloads.
Greater efficiency with CoWoS with HBM2
Applications often spend more time and energy waiting for data than processing it. The NVIDIA Tesla P100 tightly integrates compute and data on the same package by adding Chip on Wafer on Substrate (CoWoS) with HBM2 technology to deliver unprecedented computational efficiency. This integration provides a huge generational leap in application performance by delivering up to 3X memory bandwidth over prior-generation solutions.
Simplified parallel programming with the Page Migration Engine
Parallel programming just got a lot simpler with the Pascal architecture. The Page Migration Engine frees developers to focus more on tuning for computing performance and less on managing data movement. It also allows applications to scale beyond the physical memory size of the GPU, with support for virtual memory paging. With Unified Memory technology, developers see a single memory space for the entire instance to dramatically improve productivity.
NVIDIA Tesla V100 GPUs are now generally available on Compute Engine and Google Kubernetes Engine.
Today’s most demanding workloads and industries require the fastest hardware accelerators. Customers can now select as many as eight NVIDIA Tesla V100 GPUs, 96 vCPU, and 624 GB of system memory in a single VM, receiving up to 1,000 teraflops of mixed precision hardware acceleration performance. Next-generation NVIDIA NVLink interconnects deliver up to 300 GB/s of GPU-to-GPU bandwidth, 9X over PCIe, boosting performance on deep learning and HPC workloads by up to 40%.
Visit the GPU documentation to stay up to date on the latest in pricing and regional availability for NVIDIA GPUs.
NVIDIA GPUs in Google Kubernetes Engine (GKE) are generally available and ready to be used widely from the latest GKE release.
Using GPUs in GKE can turbocharge compute-intensive applications like machine learning, image processing, and financial modeling. By packaging your CUDA workloads into containers, you can benefit from the massive processing power of GKE’s NVIDIA GPUs whenever you need it, without having to manage hardware or even VMs.
NVIDIA Tesla P4, V100, P100, and K80 GPUs are now generally available.
Google Cloud Platform has now added support for NVIDIA GPU Cloud. NVIDIA GPU Cloud (NGC) provides simple access to GPU-accelerated software containers for deep learning, HPC applications, and HPC visualization. NGC containers are optimized and pre-integrated to run GPU-accelerated software that takes full advantage of NVIDIA Tesla V100 and P100 GPUs on Google Cloud Platform. By making it simple to access NVIDIA GPU-accelerated software and NVIDIA GPUs, Google Cloud Platform is now helping you deploy production-quality GPU-optimized software in just minutes.
Additional Resources
GPUs for Google Cloud Platform
Read the blog
Why Shazam uses GCP and NVIDIA GPUs
Read the blog
8 reasons to run your ML with NVIDIA GPUs on GCP
Watch the video
NVIDIA GPUs on Compute Engine
Learn more
GPUs in Google Kubernetes Engine now available in beta
Read the blog
NVIDIA GPU Cloud with Google Cloud Platform
Get started today