Each GPU board hosts eight GPU modules and six NVSwitch chips. Nvidia upped the NVLink port count on its Tesla V100 SXM3 GPU module to six ports (there is still only one PCI-Express 3.0 x16 host ...

Compute engine makers can do all they want to bring the performance of their devices on par or even reasonably close to that of Nvidia’s various GPU accelerators, but until they have something akin to ...

Compared to PCIe, NVLink offers higher bandwidth, lower latency, ... utilizing the NVSwitch chip to interconnect 16 GPUs, significantly enhancing computational efficiency.

Chipmaker Nvidia has announced a new GPU interconnect fabric, NVSwitch, which allows 16 GPUs to simultaneously communicate. The on node switch architecture has a total of 18 ports, which provide ...

In this video from the GPU Technology Conference, Marc Hamilton from NVIDIA describes the new DGX-2 supercomputer with the NVSwitch interconnect. “The rapid growth in deep learning workloads has ...

Salvator told me that the architecture of NVLink and NVSwitch is critical. “It’s already helping us today and will help us even more going forward, delivering generative AI inference to the ...

Astera Labs demos PCIe 6.0 connectivity working at NVIDIA GTC 2024 with working PCIe 6.0 retimer ready for NVIDIA's new Blackwell B100, B200, GB200 AI GPUs.

It has the same core design of the H100 PCIe, which also supports NVLink, but potentially now with more of the GPU cores enabled, and with 17.5 per cent more memory. It looks a bit pricey too.

Nvidia uses NVLink for fast communication between its own GPU and CPU chips in what it dubs the GH200 ‘Superchip.’The superchips have been used in multiple supercomputers.

Nvidia's NVLink Fusion program allows customers to use the company’s key NVLink tech for their own custom rack-scale designs with non-Nvidia CPUs or accelerators in tandem with Nvidia’s products.