OK, so you read my blog post yesterday about the new HP ProLiant SL390 G7 4U server and went to the HP web site and purchased one, complete with eight Nvidia M2070 GPUs, on-board 10G/IB networking, and twice the internal PCI I/O capability of most competitors x86 servers so that you can actually make use of all that GPU power. What should you do now while your purpose-built HPC server from HP is built and shipped to you?
Start by registering for the Nvidia Registered Developer Program which will give you free download access to the latest CUDA 4.0 developer toolkit.
Next, plan on attending this Friday’s CUDA 4.0 webinar.
Before Friday’s webinar, be sure to read what HPCWire had to say about the latest CUDA 4.0 toolkit. The 4.0 toolkit brings with it three new technologies: GPUDirect 2.0 for peer-to-peer communication among graphics chips on the same compute node; Unified Virtual Addressing (UVA) to provide a single address space that melds CPU and GPU memories; and Thrust, a set of C++ library of data structures and algorithms aimed at parallel programming.
Then sit back and get ready to experience one of the world’s fastest and most energy efficient GPU server platforms, your new SL390 G7 4U. When Nvidia’s Project Denver, the codename for Nvidia’s future CPU-GPU architecture for workstations, servers, and supercomputers, comes to fruition, you are sure to see even more powerful GPU platforms, but for now, the SL390 G7 4U sets a new bar for GPU computing power and efficiency.