Subscribe to the Corespan newsletter

We'll email you once to confirm. We never share your address.

Three panels: software-defined networking, today's server with eight GPUs fixed to one CPU at the factory, and Corespan's Software Defined Machine, where Composer carves 16 to 32 GPUs per CPU from photonic GPU and SSD pools.

Airline Economics, and the Part Financing Can't Fix

This issue of the Corespan newsletter is about the airline analogy that went around investor circles last month, the part of it financing cannot fix, and what we are bringing to SC26 in two weeks.

Enter your email to keep reading

8× AMD Radeon AI PRO R9700 run as one fabric: ~111 GB/s bidirectional on every GPU-to-GPU pair

In the Lab: 8× AMD Radeon AI PRO R9700, One Fabric

This issue of the Corespan newsletter covers one result from our lab, using AMD Radeon AI PRO R9700 GPUs. What we saw is roughly twice the effective GPU-to-GPU bandwidth provisioned in NVIDIA's 8-GPU RTX PRO 6000 reference architecture.

Enter your email to keep reading