There’s been a great need for a while now for a decent, easy to understand introduction to HPC. Those of us who’ve worked in this niche understand the acronyms and weird technology, but for newcomers – even those with a good background in IT – HPC can be an intimidating arena.
Couple this with the blurring between ‘traditional’ HPC systems for research, and new high end business solutions for statistical analysis and database warehouses, and there’s a real need to de-mystify HPC for all.
Douglas Eadline, working with Sun and AMD, has written HPC for Dummies, and the blurb explains:
This special edition eBook from Sun and AMD shares details on real-world uses of HPC, explains the different types of HPC, guides you on how to choose between different suppliers, and provides benchmarks and guidelines you can use to get your system up and running.
What makes this really great is that this is a free ebook, available direct for download from Sun’s site. I can highly recommend that anyone with an interest in HPC (or just large scale systems design) grabs this and has a read through.
Head on over to Sun’s HPC for Dummies page.
As always, Sun’s HPC Newsletter is well worth a read. This month has some interesting HPC videos, details of the latest Constellation cluster upgrades, and information on the Sun HPC Developer stack, which I’ve previously mentioned and which is well worth trying out.
Excellent news has arrived from the HPC Developer OpenSolaris community, via Bruce Rothermal. Traditionally it’s been very hard to get involved in HPC – you need a lot of kit, a lot of software, and some knowledge to get it all setup.
Sun have solved all of this by making available a Virtual Machine Image for VirtualBox (or VMWare) which contains an entire HPC stack:
- Sun Grid Engine and Zones
- MPI and HPC Cluster Tools
- compilers, scripting languages, and more
The HPC Developer Stack provides a simple, easy way to start getting up to speed with the same technologies and tools that are used on monster installs like TACC’s Ranger.
Grab the download, fire up VirtualBox, and start getting involved in the world of HPC.
The Supercomputing 08 trade show is in full swing, and I’ve been blogging about some of Silicon Graphics’ offerings over at Siliconbunny – links at the end. Sun’s John Fowler, who is now in charge of the Systems Platforms group at Sun, after the recent layoffs and re-organisation, popped up at SC08 to give a preview of some high end gear coming down the line from Sun.
Sun’s Magnum Infiniband switch
Probably the most interesting preview offering is Sun’s upcoming blade server. It has two sockets and squeezes an entire server into each blade. The key thing though is that each blade provides quad data rate (QDR) Infiniband direct off the board – 40 Gb/s throughput makes 10Gb Ethernet pretty laughable, even without taking into account the advantages the Infiniband protocol has over Ethernet.
Current Sun blades use PCI-E cards to provide Infiniband connectivity, which isn’t as fast and takes up extra room on the board.
Sun are also pushing to pre-package HPC solutions. For example, on the back of their announcements of the Storage 7000 Unified Storage System, Sun will be announcing the Sun Storage Cluster. This is a rack of pre-configured storage servers and disks, running Linux and using the Lustre file system.
The Sun Compute Cluster is another bundled solution, this time with an integrated rack of servers, network and software, all pre-configured, ready to be plugged in. Options range from 1 to 8 racks of gear, and are comprised of either 32 X2250 dual-socket servers per rack, or 30 Sun Blade servers.
The point of both of these bundles is to make HPC easy to implement – and to sell. IBM and HP already do something similar, and have had a lot of success with smaller installations.
Sun also announced something which we’ll definitely be seeing more of in our datacentres – their custom water jackets that fit to the back of racks, codenamed Glacier.
If you’re reading this and thinking “What does HPC have to do with me?”, think again. Silicon Graphics said back in 1996 that they wanted their high-end graphics systems to appear in a games console in 5 years (a wish that came true). These sort of lower-end HPC systems not only represent the technologies that we’ll all soon be using in our data centre, but they also make hugely cost effective solutions for larger-scale business problems like data warehouses or ERP. High Performance Computing is no longer confined to government labs and large research centres.
Follow these links to SiliconBunny to read about the Silicon Graphics Molecule concept computer, and SGI’s haul at the HPCwire Reader’s Choice Awards.
