14 DAY TRIAL // Though not exactly the oldest and best known of computing solutions, GPUs have been steadily making a name for themselves on the HPC (high performance computing) market because they can provide a serious amount of FLOPS. IBM obviously knows this, which is why its newly created iDataPlex dx360 M3 server, designed for use in data-center environments, doesn't shy away from including a pair of the latest GPU computing modules on the market
Like any self-respecting HPC system, IBM's offer is available with the latest CPUs, namely the Intel Xeon 5600 Series of quad-core or six-core units. Two such processors are joined with up to 128GB of DDR3 memory, clocked at 1333MHz.
Additionally, each system boasts up to 10TB of storage and, of course, PCI Express x16 slots (two to be exact) meant to accommodate GPU computing modules. Said GPU modules can be either two NVIDIA Tesla M1060 or a pair of M2050. It should be noted that the latter is based on the Fermi architecture, which is why it can deliver 1Tflop of single precision power.
“Nvidia provides an innovative solution for customers who push the envelope in high-performance computing,” Dave Turek, vice president, Deep Computing, IBM, said. “GPU acceleration provides performance boosts for many applications in energy exploration, science and financial services. It is among the significant emerging supercomputer technologies to watch in the years ahead.”
Other benefits that IBM's newest server is said to have include a half-depth form factor (for greater power and cooling efficiency), a flexible design, simplified system management, a dual-zone, high-efficiency power supply (that further reduces power requirements) and front I/O access, for a higher level of serviceability. Finally, a complete iDataPlex rack can combine 84 servers, switches and PDUs in a standard 42U enterprise rack footprint, thanks to the implementation of the innovative shallow-depth design.