RVM

Image bloat (code +25% & data +5%)

Details

  • Type: Bug Bug
  • Status: Closed Closed
  • Priority: Major Major
  • Resolution: Fixed
  • Affects Version/s: None
  • Fix Version/s: 2.9.3
  • Component/s: Infrastructure: Build
  • Labels:
    None
  • Environment:
    x86 lnx
  • Number of attachments :
    0

Description

Unfortunately while rvmx86lnx32b was off the air, we blew out the code image size by 25%, the data image by 5% and the gross by 10% [1].

The revision window is r13822-13836. It looks like the only changes to the trunk in this window were a number of compiler changes, so presumably that's where the problem is (Dave & Daniel's changes were just to branches).

Ian, can you please take a look?

Thanks,

--Steve

[1] http://jikesrvm.anu.edu.au/cattrack/results/rvmx86lnx32b.anu.edu.au/sanity/1672/statistics_report

Issue Links

is related to

Bug - A problem which impairs or prevents the functions of the product. RVM-809 Image bloat

  • Major - Major loss of function.
  • Open - The issue is open and ready for the assignee to start work on it.
relates to

Bug - A problem which impairs or prevents the functions of the product. RVM-394 Tune inlining of arraycopy

  • Major - Major loss of function.
  • Closed - The issue is considered finished, the resolution is correct. Issues which are not closed can be reopened.

Task - A task that needs to be done. RVM-222 Reduce bootimage bloat

  • Minor - Minor loss of function, or other problem where easy workaround is present.
  • Open - The issue is open and ready for the assignee to start work on it.

Activity

Ascending order - Click to sort in descending order
Hide
Permalink
Ian Rogers added a comment -

Yep, unfortunately the typing and register improvements introduced an enum associated overhead that I did a fix of by forcing inlining all over the assembler. I'm working on a few things that should improve this situation. In the mean-time the assembler improvements are hopefully helping work on improved register allocators. Hopefully also the vectorization work when the relevant people get time.

Show
Ian Rogers added a comment - Yep, unfortunately the typing and register improvements introduced an enum associated overhead that I did a fix of by forcing inlining all over the assembler. I'm working on a few things that should improve this situation. In the mean-time the assembler improvements are hopefully helping work on improved register allocators. Hopefully also the vectorization work when the relevant people get time.
Hide
Permalink
Ian Rogers added a comment -

Patch committed in r14020.

Show
Ian Rogers added a comment - Patch committed in r14020.
Hide
Permalink
Ian Rogers added a comment -

Space is down about 28% for code and about 6% for data.

Show
Ian Rogers added a comment - Space is down about 28% for code and about 6% for data.

People

  • Assignee:
    Ian Rogers
    Reporter:
    Steve Blackburn
Vote (0)
Watch (0)

Dates

  • Created:
    Updated:
    Resolved: