Presentation is loading. Please wait.

Presentation is loading. Please wait.

15-745 Spring 20051 Path Profile Estimation and Superblock Formation Jeff Pang Jimeng Sun.

Similar presentations


Presentation on theme: "15-745 Spring 20051 Path Profile Estimation and Superblock Formation Jeff Pang Jimeng Sun."— Presentation transcript:

1 15-745 Spring 20051 Path Profile Estimation and Superblock Formation Jeff Pang Jimeng Sun

2 15-745 Spring 20052 Motivation Continuous Optimization Dynamic Optimization Realistic Profiles OptimizeCompileRun Profile

3 15-745 Spring 20053 Challenges Automate optimization Low overhead profiling Accurate profiling OptimizeCompileRun ProfilePath Profile Sample Estimate

4 15-745 Spring 20054 Project Goals Simulate performance monitoring unit –Like in Pentium 4, Itanium, PPC 970, etc. –Allows sampling of last couple branches Estimate full path profile using samples –Leverage data mining techniques similar to PageRank Validate by doing Superblock formation –Powerful optimization to improve scheduling (especially on VLIW processors) Superblock Formation Run with Simulated PMU Path ProfileSample Data Mining based Path Estimation

5 15-745 Spring 20055 Superblock Formation

6 15-745 Spring 20056 Project Outline Implement PMU simulator and Superblock optimization as SUIF passes Implement Estimator offline using sampled branch profiles and SUIF CFG sourcelabel instrument c2dil instrumented program superblock optimized program Offline estimator path profile

7 15-745 Spring 20057 Completed PMU Simulation: modified HALT profiling lib Initial offline estimator (?) sourcelabel instrument c2dil instrumented program superblock optimized program Offline estimator path profile

8 15-745 Spring 20058 Initial Results

9 15-745 Spring 20059 Todo Superblock SUIF pass –Have c2dil build hyperblocks using superblocks Refine path estimation heuristics sourcelabel instrument c2dil instrumented program superblock optimized program Offline estimator path profile

10 15-745 Spring 200510 Questions or Comments? Anyone have a good scheduler from Assignment 2? –Better scheduler = better comparison of superblock scheduling performance The HALT lib maybe useful –If you need to uniquely label branches, basic blocks, loads, stores, etc. –Insert instrumentation at those points

11 15-745 Spring 200511 Extra Slides

12 15-745 Spring 200512 Related Work Shye, et al. 2005 –Used heuristics to rebuild path profiles using PMU partial path samples –Achieved 80-99% path accuracy with <5% overhead –Did not evaluate impact on optimizations Chen, et al. 2003 –Used PMU partial path sampling to dynamically form hot traces –Used to adapt to dynamic phase transitions


Download ppt "15-745 Spring 20051 Path Profile Estimation and Superblock Formation Jeff Pang Jimeng Sun."

Similar presentations


Ads by Google