Xeon Phi 5110p and Free Intel Parallel Studio Cluster Edition

Another amazing deal on Xeon Phi from Intel! This time you can get a 90% discount on a Phi 5110p and get the Intel Parallel Studio Cluster edition with a 1 year license for free. After that 1 year you can get a perpetual license for an 80% discount if you participate in their “Code Modernization workshops” or “webinars”. This is a really great deal!

Note: You can get a Windows version of Intel Parallel Studio Cluster Edition under this offer too!

Puget Systems will be offering this Phi + Parallel Studio bundle together with a new fully setup workstation. If you are mostly interested in the great deal for the Intel compilers and tools then I believe there are other vendors that are offering the deal without the purchase of a new system.

Important: If you are not familiar with the Xeon Phi then you should read Top 5 Xeon Phi Misconceptions I wrote this 2 years ago but it is still relevant.

Xeon Phi 5110p

The Xeon Phi 5110p is a passive cooled coprocessor card. It was the first “retail” version of the Phi introduced in Q4 2012. See the Intel ARK entry for card specs. Please keep in mind what “passive cooled” means. … It has no on-card fan! The card is intended to be run in chassis with push-pull high pressure fans to move air through the card. Puget Systems has a push-pull fan setup that is relatively quiet that we can use in our Peak Tower systems. Otherwise you need a rack chassis with the proper cooling arrangement. Performance of the 5110p is very good. It will provide well over 700 GFLOP/s double precision on the Intel tuned Linpack benchmark.

Intel Parallel Studio Cluster Edition

The bundle includes the full Cluster Edition of Intel Parallel Studio. That includes Intel’s optimized MPI stack which save you the extra work of having to build an MPI implementation to link into your Intel compiled code. The Intel development tools are very good. Parallelization guidance, debugging and profiling are provided by Advisor XE, Inspector XE and VTune Amplifier.

The license is a 1 year “named license”. I’m not sure exactly what that means since I haven’t seen the license agreement yet. I assume it is a single user node locked license. I don’t know how it is “expired” after 1 year. Intel’s normal licensing scheme is 1 year support and updates followed by a renewable yearly subscription to maintain eligibility for updates and support. After I see the EULA I’ll post something in the comments for clarification (assuming the EULA provides clarification 🙂

In any case this is a great deal even just considering the compiler and tools license value alone!

“Accelerate your code with a special offer on the Intel Xeon Phi coprocessor, plus a free 12-month named-user license of Intel Parallel Studio XE Cluster Edition.”

Code Modernization workshops and webinars

The bundle announcement mentions that a perpetual license is available after the first year for user who participate in a Code Modernization workshop or webinar. The link to the workshops doesn’t list any events in the USA and just 11 worldwide so this may be of limited access.

“Attend our Code Modernization workshops or webinars, and you’ll also qualify for 80% off the purchase of a perpetual license for Intel Parallel Studio XE Cluster Edition.”

The “webinars” in announcement are the workshops that have been conducted by Colfax International. I attended one of these last year presented by Andrey Vladimirov
and found it enjoyable and informative. Andrey is a great guy and really knows how to code for the Phi! I recommend attending one if you can.

Should you get the Phi?

First, I think this deal is a “no-brainer” for anyone who is working on Intel hardware and doesn’t already have a Parallel Studio license. However, the question –should you be developing for the Phi now given that the next version, Knights Landing, will become available around the end of the year? — is a little more open. If you are serious about developing high quality parallel code targeting Intel hardware then the answer is probably yes. The effort you put into optimizing thread count and vectorization on the Phi will translate into improved code performance on Xeon CPU’s too. Also, the work you put into developing parallel applications with high thread count in mind is going to directly translate to new compute hardware for the foreseeable future. The Xeon Phi Knights Landing will be very different from the current Knights Corner in many ways but the principles for code development and algorithm design will remain the same.

Happy computing –dbk