[PATCH] R600: Implement getRsqrtEstimate

[Hal Finkel]

----- Original Message -----
> From: "Matt Arsenault" <Matthew.Arsenault at amd.com>
> To: "Matthew Arsenault" <Matthew.Arsenault at amd.com>
> Cc: llvm-commits at cs.uiuc.edu
> Sent: Monday, October 6, 2014 8:23:18 PM
> Subject: [PATCH] R600: Implement getRsqrtEstimate
> 
> Only do for f32 since I'm unclear on both what this is expecting
> for the refinement steps in terms of accuracy, and what
> f64 instruction actually provides.

It seems that the comment describing the convergence rate never moved out of the original PowerPC implementation. Here it is:

    // Convergence is quadratic, so we essentially double the number of digits
    // correct after every iteration. For both FRE and FRSQRTE, the minimum
    // architected relative accuracy is 2^-5. When hasRecipPrec(), this is
    // 2^-14. IEEE float has 23 digits and double has 52 digits.
    RefinementSteps = Subtarget.hasRecipPrec() ? 1 : 3;
    if (VT.getScalarType() == MVT::f64)
      ++RefinementSteps;

I hope this helps ;)

 -Hal

> 
> http://reviews.llvm.org/D5636
> 
> Files:
>   lib/Target/R600/AMDGPUISelLowering.cpp
>   lib/Target/R600/AMDGPUISelLowering.h
>   test/CodeGen/R600/fsqrt.ll
>   test/CodeGen/R600/rsq.ll
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
> 

-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory