It is my pleasure to announce the Beta for TotalView 8.13.

The TotalView 8.13 beta updates our Intel Xeon Phi support with two major features. First, we can now debug Xeon Phi applications running in symmetric mode on both the Xeon host processor and the Xeon Phi coprocessor. Second, MemoryScape memory debugging is now available for native and symmetric mode applications running on the Intel Xeon Phi coprocessor. It also features updated CUDA and Mac OS X support.

TotalView users who work with large-scale systems will now be able to take advantage of our new scalable early access version. This capability is based on MRNet overlay trees and allows us to debug larger sessions as well as provide higher performance when you are debugging at scale. This will be offered as an early access feature in TotalView 8.13 because it is still undergoing significant optimization. All beta users who are testing on Linux, Blue Gene or Cray systems are welcome to try it and provide feedback.

This version of TotalView also includes an updated FLEXlm license server, which resolves a possible security issue.

The list of new features and enhancements in this release is:

CUDA 5.0 and 5.5 support. There are some limitations with respect to dynamic parallelism.
Memory debugging of Xeon Phi native and symmetric mode applications with MemoryScape.
Support for the debugging of Xeon Phi symmetric mode applications.
Support for the new Macintosh OS X Mavericks.
Improved performance and scalability for those using our tree based scalability early access versions.
Scalable data aggregation for the dwhere and dstatus CLI commands.
Improved performance when setting breakpoints.
Improved security through FLEXlm 11.11 license manager upgrade.

The beta program will feature at least two interactive webinars, one introducing the new features and capabilities of TotalView 8.13 and a second to answer questions and gather any feedback.

To participate, please email with the following information:

Will you be able to test TotalView with the Intel Xeon Phi?
Will you be able to test TotalView with CUDA?
Will you be able to test TotalView on Mac OS X Mavericks?
What are the platforms you plan to test TotalView on?
How long have you used TotalView?
Might you be willing to test the new FLEXLM vendor daemon?
Would you be interested in testing TotalView at scale?