May 2004 - Posts

• FP Whitepaper Addition: Setting the floating-point semantic mode on a function-by-function basis in VC8.0

  The command-line switches -fp:precise, -fp:strict etc. are simply shorthand for setting the four different floating-point options (these are explained in the whitepaper). To explicitly select a particular floating-point semantic mode on a function-by-function Read More... Posted Tuesday, May 18, 2004 5:35 PM by ericflee | 2 Comments

• FP Whitepaper Errata: The float_control Pragma in VC8.0

  The /fp:precise, /fp:fast, /fp:strict and /fp:except switches control floating-point semantics on a file-by-file basis. The float_control pragma combined with the fenv_access and fp_contract pragmas allow such control on a function-by-function basis. Read More... Posted Tuesday, May 18, 2004 5:33 PM by ericflee | 0 Comments

• Changes to the floating-point whitepaper

  There are several changes to the floating-point whitepaper that won’t make it onto msdn.com until later this month. The copyists and editors and MSDN are working hard, but their pipeline is quite full. In the interim I’ll post revisions here Read More... Posted Tuesday, May 18, 2004 5:32 PM by ericflee | 0 Comments

• Microsoft Visual C++ Floating-Point Optimization

  MSDN recently published one of my papers : Microsoft Visual C++ Floating-Point Optimization Compiler optimizers have long been the bane of numerical algorithms. Either they generate fast code at the expense of floating-point correctness or they generate Read More... Posted Monday, May 10, 2004 5:57 PM by ericflee | 0 Comments

• Introductions...

  I work for the Visual C++ group at Microsoft where I am responsible for determining the quality of floating-point code generation. My interests include validated numerical computing and generative programming. My most recent research involves validating Read More... Posted Monday, May 10, 2004 5:54 PM by ericflee | 1 Comments