Distributed Symmetric Multi-Processing

Our Duets and Trios are Symmetric Multi-Processing (SMP) large shared-memory supercomputers. To programmers, both our Duet and Trio Departmental Supercomputers look just like a single huge-memory Linux box. Programmers can use standard threading packages to get access to all 128 (with the Duet) or 192 (with the Trio) cores and up to 1.5 TeraBytes of memory. With our supercomputers, programmers do not need to worry about message passing interface (MPI) programming, which is what supercomputing clusters and other limited memory systems demand. There's also no need to build complex file-access program components; programmers can just read a big dataset into memory and access it as an array.

We create large shared-memory (i.e., Symmetric Multi-Processing) high performance computers by marrying state-of-the-art, off-the-shelf (OTS) server hardware with our patent-pending Distributed Symmetric Multi-Processing (DSMP) extension to the Linux kernel. Other SMP supercomputer vendors either rely on expensive proprietary hardware or depend on slower, much less efficient hypervisor implementations. Our implementation of Symmetric Multi-Processing is at the Operating System-level which is inherently faster than hypervisor implementations.

Hypervisors represent the first generation of software-based SMP systems using homogeneous cluster nodes. We represent the next generation of software-based SMP on homogeneous cluster nodes. Our DSMP is a Linux kernel extension

Employing a Linux kernel extension with state-of-the-art, OTS server hardware, Symmetric Computing's DSMP provides large shared memory, multiple-core SMP computing with both economy and performance. The table below compares Message Programming Interface (MPI) Clusters, SMP supercomputers and our DSMP-enabled clusters.

	MPI Cluster	SMP Supercomputer	DSMP Cluster
Off-the-Shelf Hardware	Yes	No	Yes
Affordability	$	$$$$$$	$
Shared Memory	No	Yes	Yes
Single Software Image	No	Yes	Yes
Intelligent Platform Management Interface	Yes	Yes	Yes
Virtualization Support	No	Yes	No
Scalable	Yes	No	Yes

Click here to download a PDF of a White Paper describing our DSMP Linux Kernel Extension.