Xtreme part 2, Appro's sequel

Print Friendly, PDF & Email

Early November last year Appro announced its new Xtreme-X1 line of quad Xeon-based cluster HPCs. The X1, is designed to be the lego building block of cluster solutions by allowing customers to build systems up in 128-node blocks. Earlier this year they announced the sale of a 95 TFLOPS Xtreme to the University of Tsukuba in Japan, and the company has had recent wins of note with 438 TFLOPS in 8 clusters across DOE’s NNSA and 33 TFLOPS in another system at LLNL.

The Tsukuba system is AMD-based, not Intel, and today Appro let the other X shoe drop. Today the company announced its new Xtreme-X2 line of AMD-based clusters. The machines feature dual-socket quad core Opterons. The company expects to announce another large Xtreme customer in early March.

I got to speak with John Lee, VP of Advanced Technology Solutions at Appro, ahead of the announcement. According to John the Xtreme series is targeted toward the mid-sized scientific HPC buyer in an organization that is probably already familiar with HPC, and features reliability features that make it very attractive to users without a lot of resources to put into running the machine.

For example, Appro builds the Xtreme systems with cheaper 24-port IB switches rather than the more expensive 288-port switches. This can reduce the cost sufficiently that redundant networks can be put in place to deal with link failures. You can configure your X1 or X2 with the standard fat tree interconnect, or pick a mesh or torus if that’s the way you roll.

Appro is also working on software in its Cluster Engine that will support application checkpointing without user intervention (I did ask some questions about this and I’m working on a follow up interview to tell you more), and a variety of other management and reliability features.

Appro is building an in-house organization to service its North American customers, but it has its eyes on the rest of the world, too. The Tsukuba announcement included details that Cray, Japan would be providing support services for the machine in a partnership.

Lee indicated that this kind of support arrangement deal is negotiated “where it makes sense.” I suppose it makes business sense not to build out an entire global support organization until you are sure there is enough business to sustain it, and Cray is certainly a partner with a lot of HPC cred. But I can’t help thinking that they would want to have Appro-jacketed engineers servicing their systems as soon as possible, though. No sense in giving your current customers any ideas about where they might buy their next machine.