Vector Fabrics Newsletter November 2011
New v1.14 releases for vfEmbedded and vfThreaded-x86
We are releasing another update to our tools. The main new features and improvements are highlighted below:

-
Support for more parallel patterns. For example, we’ve now extended our support for the generation of small tasks that produce intermediate results, which are then combined to produce the final result (the map-reduce pattern).
-
Schedule view improvement. We now intuitively show overhead associated with creating threads. Improved processor and memory timing models. Parallel program execution time prediction now takes into account contention on buses, multiple levels of cache and external memory, preventing you from introducing parallel code slow-downs.
-
Improved tool speed. We’ve greatly increased the browsing and parallelization of large programs, especially when there are lots of calls to functions that aren’t executed, which is often the case in big applications.
This month’s Very Cool Feature highlight: Schedule overhead view

Last month we highlighted our speed-up prediction, eliminating the risk of spending a lot of time on code changes that will actually slow down the code This month’s Very Cool Feature is our improved schedule view,which shows the percentage of time that is spent on creating threads, and how much is spent on properly cleaning up threads. You can now quickly determine when the overhead of introducing threading is large, for instance for short loops, or loops that don’t perform enough work in their bodies to benefit from parallelization. This is another example of how our tools save you precious coding time when you’re optimizing your code.
Video: Intel's “Parallel programming talk” features our Chief Architect

Our own Paul Stravers was a guest on the Intel’s parallel programming talk, an online talk show that focuses on programming Intel’s multicore architectures. Kathy and Clay from Intel interview Paul about our flagship vfEmbedded tool. Forward to 6:50 to see the 20-minute section that features Paul.
We’re now an Intel® Software Premier Elite Partner
Due to our excellent support for Intel products, we have been awarded the most coveted Intel Software Premier Elite Partner status. The Premier Elite Partner status is the highest possible partner level with Intel®, recognizing Vector Fabrics for our dedication to developing highly effective parallel software development tools and bringing real value to Intel’s customers. We’re proud of our status and are looking forward to continue our relationship with Intel, and helping coders out there get the most performance out of their multicore Intel architectures.
Event Calender
|
Functional Programming day |
January 6, Utrecht, the Netherlands. |
Talk by Alexey Rodriguez “How to dress up a camel in Python's clothing” |
|
Consumer Electronics Show |
January 12 – 13, Las Vegas, USA. |
Please contact Christian Graber to set up a meeting: christian@vectorfabrics.com, +1-650-224-5255 |
|
Mobile World Congress |
February 27 – March 1, Barcelona, Spain |
Please contact Marco Jacobs to set up a meeting: marco@vectorfabrics.com, +31-6504-55522 |
|
Embedded World |
February 28 – March 1, Nuremberg, Germany, Hall 4/4-607 |
Please contact Mirte Maier to set up a meeting: mirte@vectorfabrics.com, +31 40 820-0960 |
|
ESC Silicon Valley |
March 26 – 29, San Jose, USA |
Please contact Christian Graber to set up a meeting: christian@vectorfabrics.com, +1-650-224-5255 |
Please contact us at info@vectorfabrics.com to schedule a meeting with us at these shows. We are looking forward to speaking with you about your multicore development projects.
|