diff options
author | Michael Gottesman <mgottesman@apple.com> | 2013-11-17 03:25:24 +0000 |
---|---|---|
committer | Michael Gottesman <mgottesman@apple.com> | 2013-11-17 03:25:24 +0000 |
commit | e7a1e3ee8279f12d0f2b49fb198d577949795c88 (patch) | |
tree | 14e5e6af966229b0636a5b96eb1a736b33c18df8 /include | |
parent | 8417e857818e972f318bd549caabc54bdf537c26 (diff) | |
download | external_llvm-e7a1e3ee8279f12d0f2b49fb198d577949795c88.zip external_llvm-e7a1e3ee8279f12d0f2b49fb198d577949795c88.tar.gz external_llvm-e7a1e3ee8279f12d0f2b49fb198d577949795c88.tar.bz2 |
[block-freq] Add BlockFrequency::scale that returns a remainder from the division and make the private scale in BlockFrequency more performant.
This change is the first in a series of changes improving LLVM's Block
Frequency propogation implementation to not lose probability mass in
branchy code when propogating block frequency information from a basic
block to its successors. This patch is a simple infrastructure
improvement that does not actually modify the block frequency
algorithm. The specific changes are:
1. Changes the division algorithm used when scaling block frequencies by
branch probabilities to a short division algorithm. This gives us the
remainder for free as well as provides a nice speed boost. When I
benched the old routine and the new routine on a Sandy Bridge iMac with
disabled turbo mode performing 8192 iterations on an array of length
32768, I saw ~600% increase in speed in mean/median performance.
2. Exposes a scale method that returns a remainder. This is important so
we can ensure that when we scale a block frequency by some branch
probability BP = N/D, the remainder from the division by D can be
retrieved and propagated to other children to ensure no probability mass
is lost (more to come on this).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194950 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'include')
-rw-r--r-- | include/llvm/Support/BlockFrequency.h | 10 |
1 files changed, 8 insertions, 2 deletions
diff --git a/include/llvm/Support/BlockFrequency.h b/include/llvm/Support/BlockFrequency.h index 666b3c8..21879e7 100644 --- a/include/llvm/Support/BlockFrequency.h +++ b/include/llvm/Support/BlockFrequency.h @@ -27,8 +27,10 @@ class BlockFrequency { uint64_t Frequency; static const int64_t ENTRY_FREQ = 1 << 14; - // Scale frequency by N/D, saturating on overflow. - void scale(uint32_t N, uint32_t D); + /// \brief Scale the given BlockFrequency by N/D. Return the remainder from + /// the division by D. Upon overflow, the routine will saturate and + /// additionally will return the remainder set to D. + uint32_t scale(uint32_t N, uint32_t D); public: BlockFrequency(uint64_t Freq = 0) : Frequency(Freq) { } @@ -57,6 +59,10 @@ public: BlockFrequency &operator+=(const BlockFrequency &Freq); const BlockFrequency operator+(const BlockFrequency &Freq) const; + /// \brief Scale the given BlockFrequency by N/D. Return the remainder from + /// the division by D. Upon overflow, the routine will saturate. + uint32_t scale(const BranchProbability &Prob); + bool operator<(const BlockFrequency &RHS) const { return Frequency < RHS.Frequency; } |