New TR: "Static Branch Frequency and Program Profile Analysis"

Preprints of the following TR are available via anonymous ftp.


/Jim
Computer Sciences Department
University of Wisconsin
1210 West Dayton Street
Madison, WI 53706
larus@cs.wisc.edu
608-262-9519






%A Youfeng Wu
%A James R. Larus
%T Static Branch Frequency and Program Profile Analysis
%D September 1994
%R TR 1248
%I COMPUTER SCIENCES DEPARTMENT, UNIVERSITY OF WISCONSIN
%C MADISON, WI
%X Program profiles identify frequently executed portions of a program,
which are the places at which optimizations offer programmers and
compilers the greatest benefit. Compilers, however, infrequently
exploit program profiles, because profiling a program requires a
programmer to instrument and run the program. An attractive
alternative is for the compiler to statically estimate program
profiles.. This paper presents several new techniques for static
branch prediction and profiling. The first technique combines
multiple predictions of a branch's outcome into a prediction of the
probability that the branch is taken. Another technique uses these
predictions to estimate the relative execution frequency (i.e.,
profile) of basic blocks and control-flow edges within a procedure. A
third algorithm uses local frequency estimates to predict the global
frequency of calls, procedure invocations, and basic block and
control-flow edge executions. Experiments on the SPEC92 integer
benchmarks and Unix applications show that the frequently executed
blocks, edges, and functions identified by our techniques closely
match those in a dynamic profile.


To appear: 27th IEEE/ACM International Symposium on Microarchitecture
(MICRO-27), Nov. 1994.


===========================================================================
Instructions for Anonymous FTP of technical reports:


        1. Ftp to the host "ftp.cs.wisc.edu" using your ftp program.


        2. Use "anonymous" as the user name.


        3. At the password prompt, send your email address (email@host).


        4. cd /tech-reports


        5. Retrieve the file README for more instructions (typically with
              the "get" command)


        6. Reports are all under /tech-reports/reports/YY, where YY is the
              year.


              Reports are stored as compressed PostScript files, named trXXX.ps.Z,
              where XXX is the report number. If you can uncompress the files on
              your computer, use file type "binary" or "image" and fetch the report
              you want (how to do this depends on your ftp client software).


              If you cannot uncompress the report, the ftp server can do it for
              you before transmission of the file: leave off the ".Z" from the
              file name when giving the retrieve command.


        Files in /tech-reports you may want to look at:


README information about the archive
ALL_ABSTRACTS contains abstracts of all tech reports since Dec 1987.
ORDER_FORM for ordering tech reports not available via ftp.
OnLineList a list of all tech reports available for ftp
REPORT_LIST report numbers, titles, authors and publication
date for each report.


        If you have problems with the multi-line messages from the ftp server,
        connect again and try using a dash (-) as the first character of your
        password.


        If you have any problems with the ftp archive, please send electronic
        mail to "ftp@cs.wisc.edu".


Instructions for accessing technical reports via gopher:


        1. Connect to the gopher server on ftp.cs.wisc.edu. Choose the menu item
              "University of Wisconsin-Madison Computer Sciences Department"


        2. Choose the menu item "CS Technical Reports"


        3. Choose the appropriate menu items.


Instructions for accessing technical reports via afs:


        1. cd to /afs/cs.wisc.edu/public/tech-reports.


        2. look at the file README for more information.
--

Return to the comp.compilers page.
Search the comp.compilers archives again.