Nvidia Develops NVLink Interconnect for Next-Gen “Pascal” GPUs

Today at GTC 2014, Nvidia announced new technologies that will ship with their next-generation “Pascal” GPUs. First up is a new interconnect called NVLink, a point-to-point interconnect that offers the following advantages over PCI-E:

  • Very high speed. NVLink provides a high-bandwidth path between the GPU and the CPU at data rates 5 to 12 times that of the current PCIe Gen3. NVLink will provide between 80 and 200 GB/s of bandwidth, allowing the GPU full-bandwidth access to the CPU’s memory system.
  • Much higher energy efficiency than PCI-E at that speed. Developed by Bill Dally’s team at Nvidia Labs, NVLink leverages a “breakthrough” in low-energy data movement. This technology may have been funded by DARPA, but sources could not confirm that as of this writing
  • Nvidia can drive it across motherboards.

PascalRoadmap_575px

NVLink enables fast data exchange between CPU and GPU, thereby improving data throughput through the computing system and overcoming a key bottleneck for accelerated computing today,” said Bradley McCredie, vice president and IBM Fellow at IBM. “NVLink makes it easier for developers to modify high-performance and data analytics applications to take advantage of accelerated CPU-GPU systems. We think this technology represents another significant contribution to our OpenPOWER ecosystem.”

The basic building block for NVLink is a high-speed, 8-lane, differential, dual simplex bidirectional link. Our Pascal GPUs will support a number of these links, providing configuration flexibility. The links can be ganged together to form a single GPU↔CPU connection or used individually to create a network of GPU↔CPU and GPU↔GPU connections allowing for fast, efficient data sharing between the compute elements.

According to Nvidia, NVLink will initially be available on IBM Power systems. Read the Full Story.