Eliminating Memory References

1
Eliminating Memory References

• Joshua Dunfield
• Alina Oprea

2
Problem Definition Register Promotion

• Memory-reuse analysis
• find loads and stores that access the same
  address and the execution path along which the
  reuse exists
• Program transformation
• promote values from memory to registers
• replace redundant loads and stores with register
  references

3
Different Approaches

• Memory-reuse analysis
• Using the SSAPRE algorithm
• Modeled as a data flow problem
• Program transformation
• Formulated as a PRE problem

• path 1

• path 2

• path 3

• load a

• load a

• load b

• a d on path 1
• b d on path 2
• c ! d

• store c

• load d

4
Different Approaches

• Memory-reuse analysis
• Using the SSAPRE algorithm
• Modeled as a data flow problem
• Program transformation
• Formulated as a PRE problem

• path 1

• path 2

• path 3

• load a

• load a

• load d

• load b

• a d on path 1
• b d on path 2
• c ! d

• store c

5
Register promotion using the SSA representation

• Paper Register Promotion by Sparse Partial
  Redundancy Elimination of Loads and Store Lo,
  Chow, Kennedy, Liu, Tu
• Register promotion 2 problems
• PRE of loads
• PRE of stores

6
Duality between loads and stores

• Loads as ordinary expressions with respect to
  redundancy (have to delete the latter
  occurrences)
• Stores reverse (have to delete the earlier
  occurrences)

• load a

• load a

• store a

• store a

• store a

• load a

7
PRE of loads

• Replace each store x expr by r expr
• 
  x r
• where r is a pseudo-register
• Apply the SSAPRE algorithm, but take into account
  the occurrences of stores
• Effect of stores on loads

x r

• load x

• load x

8
Improving Code Motion in SSAPRE by Speculation

• Speculation inserting computations during
  SSAPRE at ?s where the computation is not
  down-safe (anticipated)
• Is not permitted by the original SSAPRE
• 2 strategies
• conservative speculation (when profile data is
  not available)
• profile-driven speculation

9
Speculation

• Conservative Speculation
• Move loop-invariant computations out of
  single-entry loops (can perform worse if the body
  of the loop not executed)
• Profile-driven Speculation
• Pb of determining the optimum code placement is
  undecidable (solution between no-speculate and
  fully-speculate)
• Heuristics do speculation at the granularity of
  the connected components of the SSA graph (for
  each connected component either no speculate or
  fully speculate)

10
PRE of Stores SSU form

• Loads

• Stores

• Single Assignment
• Redundant Available, later
• Fully redundant Dominated
• (by earlier
  load)
• Where to factor Merge points
• Factoring op
• Insertion points iterated DF
• Movement Backward

• Use
• Anticipated, earlier
• Post-dominated
• (by later store)
• Split points
• iterated post-DF
• Forward

11
Performance

• PRE of loads reduces the number of loads by 25
• PRE of stores reduces the number of stores by 1
• Reasons have already applied a dead store
  elimination algorithm
• Speculation results
• conservative same performance, even worse on
  some cases
• profile-driven 2 reduction in the of loads
  and 0.5 in the of stores

12
Load-Reuse Analysis

• Paper Load-Reuse Analysis Design and
  Evaluation Bodik, Gupta, Soffa
• Modeled as a data-flow problem
• 3-fold contributions
• load-reuse analysis supporting indirect memory
  accesses
• simulation of the dynamic amount of load reuse
• profile-based estimators using edge profile
  information to assign a dynamic weight to the
  static load-reuse analysis

13
Framework
14
Load-Reuse Analysis

• Uses Value Name Graph (VNG) representation
  keeps track of address expressions that compute
  the same value
• Traditionally values identified by lexical name
• VNG supports symbolic equivalences
• Enhance VNG
• handle indirect addressing
• develop a sparse version (more space efficient)

15
VNG Example

• 
  name 2v 10
• u store (v-2)
• 
  name 2 (u) 12
• x load (u)
• 
  name 2x12
• y 2x8
• 
  name y4
• z load (y4)
• name address of last load

16
Load-Reuse Analysis (cont)

• In computing symbolic names, do substitutions for
  w iterations for each loop
• Set the max no of indirection levels (0 or 1)
• Find congruence classes (names that refer to the
  same memory address)
• Extract a VNG sparse representation that contains
  only loads and stores
• Solve the data flow problem on the sparse VNG
  representation