This documents the release of TotalView V3X.9.0-1-LLNL installed into totalview on blue, s, k, and y on May 17, 1999 and on compass and forest May 19,1999.
Below are the features added to create TotalView V3X.9.0-1-LLNL.
|§ Etnus Release Notes for TotalView Version 3X.9.0-2|
|§ TotalView Quick Reference Page|
|§ Displaying C Array [Vector][Vector]...|
|§ Setting the Display Precision Globally|
|§ Format -> Precision|
|§ Find -> Threshold for ==NaN and != NaN|
|§ Performance Monitoring for IBM Temporarily Removed|
|§ Previous LLNL Enhancements to TotalView|
|§ TotalView V3.9.0-0-LLNL Release Features|
For more information, send e-mail to
Bor Chan, Karen Warren, Rich Zwakenberg (LLNL)
or Laurie McGavran (LANL).
TotalView Quick Reference Page
A TotalView Quick Reference Page is available. Click here.
Performance Monitoring for IBM Temporarily Removed
We have disabled TotalView's performance monitoring capability because it was based on an IBM unsupported Performance Monitoring Application Programming Interface (PMAPI). IBM will be delivering a supported PMAPI to us soon and we will be incorporating that new PMAPI into TotalView.
Displaying C Array [Vector][Vector]...
Previously one could only display multi-dimensional Fortran arrays, such as Array(Vector,Vector,...). TotalView now allows displaying a multi-dimensional C array where one, or any, of the scripts is a vector.
Setting the Display Precision Globally
You can now select how you want your floating point variables displayed and with what precision. From the root window menu commands, the command:
Define Global Variable Display Type -> Scientific Notation Floating Point
lets you choose what form to use to display floating point values; scientific notation or floating point. A dialog box will pop up which allows you to define the precision you want used.
Changing the global variable display type will affect display of all floating point variables currently being displayed and all floating point variables subsequently displayed.
The default is to display floating point variables as floating point with 8 digits of precision.
You can also set this value in your X resources file as well.
totalview*displayFormat: sci(entific), nn or totalview*displayFormat: float, nnThe precision (specified by nn) is optional. Remember to run xrdb -merge .Xdefaults.
This feature does not interact with the Format -> Precision setting. However, if you revert your variable display back to Format -> Native, your global variable display type setting will again be in effect.
Format -> Precision
We have added the Precision subcommand to the variable window Format command.
The Format -> Precision command allows you to set the number of significant digits displayed when the Format -> Scientific Notation format has been or will be requested for the format of a displayed floating point variable.
Once the number of digits has been defined via the Format -> Precision command, the precision will be carried through to all newly displayed floating point variable windows after you select Format -> Scientific Notation for them. If Format -> Scientific Notation has not been selected, than the display uses the Global Variable Display Type setting. Existing floating point variable windows displayed will not be updated until you again select Format -> Scientific Notation for them.
The Format -> Precision command allows you to overide the the global scientific notation precision setting you may have selected via the Global Variable Display Type setting.
The new default
is 10 digits, rather than the 6 that was used previously.
Find -> Threshold for ==NaN and != NaN
We have added to the Find -> Threshold command the checking for equal to or not equal to "not a number", specifically ==NaN and !=NaN, where NaN is not context sensitive, that is, you can check for NaN, Nan, nan, etc.
Last revised May 19, 1999