Google Cloud launches the Nvidia Blackwell racks to enhance the platform AI | Nvidia hardware | Gpu vs cpu vs graphics card | Hardware computer | Turtles AI

Google Cloud recently presented its first NVIDIA Blackwell GB200 NVL based rack server, a significant step to optimize the performance of its cloud platform. This implementation reflects the company’s commitment to enhance the AI ​​and data processing services.

Key points:

• Google has started the distribution of personalized NVIDIA GB200 NVL rack for its cloud platform.
• Each GB200 chip combines a Grace CPU with a B200 GPU, reaching 90 TFLOPS of FP64 performance.
• GB200 NVL72 officials can offer performance up to 3240 TFLOPS thanks to the configuration of 36 CPUs and 72 GPUs.
• Several companies, including Microsoft and Foxconn, have already adopted this technology to improve their systems.

Google has announced the distribution of Nvidia Blackwell GB200 NVL racks, a strategic step to enhance the skills of its cloud platform. These racks are equipped with liquid cooling technology, optimizing the thermal efficiency of high -performance components. The configuration provides for the integration of GB200 chips, which include a Grace CPU and a Blackwell B200 GPU, with performances that reach 90 TFLOPS for double precision mobile permanent operations (FP64). Although Google has specified that the racks are "personalized", it is not clear if these reflect exactly the configuration of the GB200 NVL72 model, which provides 36 CPUs and 72 interconnected GPUs via NVLINK, capable of generating a bandwidth up to 130 TB/s. In the official configuration, each rack is able to reach total performance up to 3240 TFLOPS and to manage up to 13.5 TB of HBM3E memory. It is interesting to note that Google is not the only company to exploit this technology; Microsoft recently presented a similar server, while Foxconn has started the distribution of the GB200 NVL72 racks to develop a cutting -edge supercomputer in Taiwan. The adoption of GB200 chips by numerous actors in the sector highlights the importance of this technology for a variety of applications, ranging from AI to scientific research and data archive. With the Nvidia Blackwell in full production, a further increase is expected in their implementation in different contexts.

The expansion of Google cloud systems is an element to keep an eye on in the current technological panorama.