Jason Zander - About the CLR team

The Discussion

• 

  There was a reference in there to a document in MSDN. Perhaps someone can put up a link to that.
  

• 

  I believe he is referring to the document located at:
  
  https://msdn.microsoft.com/perf
  
  
  Rico Mariani mentioned it in CLR Team Part II.  He also has a lot of good perf info in his blog:
  
  https://blogs.msdn.com/ricom/
  
  
  But I could be wrong, that might not be what he was talking about at all.  Good info anyway.

• 

  There's a lot of backup material on my blog.  There is a link to the Headtrax report and more discussion there.  There is also conceptual information on Ngen here as well.
  
  
  Two pieces of trivia:
  
  1. The white cup on my desk says:
     Cup<T>
  a clever idea that Anthony Moore, BCL dev lead, came up with for our Generics release.
  
  2.  On the left speaker is the "Checkin Cartman".  During V1 we didn't have great checkin automation (the system rocks now).  After a long tree freeze, we'd bring the tree back up for checkins by serializing per dev lead.  Only the dev lead with the Cartman could check in.

• 

  JasonZ wrote:

  1. The white cup on my desk says:
     Cup<T>
  a clever idea that Anthony Moore, BCL dev lead, came up with for our Generics release.

  
  That is bloody brilliant. You should market that.
  
  It almost works better in VB syntax: Cup(Of T)
  But because of the familiarity of the angle-bracket generics syntax (thanks to C++), I like the Cup<T> version best.

• 

  Perhaps there should be a smaller performance document that summarizes all a bit... The 1500 pages thing is huuuuge! I read a bit of it...
  
  Christian

• 

  Comment removed at user's request.

• 

  Jason spoke about slow startup times due to JIT compilation and how this is a big target for optimization..  I thought this compilation was done only the first time a new assembly was loaded...?  It seems to me to be a negligible issue.
  

• 

  dudenumber4 wrote:

  Jason spoke about slow startup times due to JIT compilation and how this is a big target for optimization..  I thought this compilation was done only the first time a new assembly was loaded...?  It seems to me to be a negligible issue.
  

  You are correct compliation is done only on first load, and then only for the methods you actually call.  Even so, the time it takes is demonstrably longer than simply paging the existing code in from disk.  Ideally, pages are also hot and sharable withother processes.  We get our best startup perf through ngen with tuned scenarios (locality of methods).  I have a lot of background material on JIT vs ngen and paging on my blog.