You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
the up-front dr_memory_is_readable() check on unaddr exceptions is wasteful
for heap header exceptions. moving it to after heap header and tls
exception checks results in a 10x speedup on twolf test!
it had >100K header exceptions and spent all its time in libc vfscanf
from the /proc/maps iteration: not sure why not using cached allmem.
now, we have no pc sampling hits at all in libc.
PR 503779: heap header exceptions very slow due to unnecessary
dr_memory_is_readable() check
moved dr_memory_is_readable() check to after heap header and tls
exception checks, which results in a 10x speedup on twolf test!
it had >100K header exceptions and spent all its time in libc vfscanf
from the /proc/maps iteration: not sure why not using cached allmem.
now, we have no pc sampling hits at all in libc.
=> unknown large speedup on vortex (always killed it before), 10x speedup
on twolf and mcf, 5x on perlbmk and vpr, 40% on gcc (all test inputs)
From derek.br...@gmail.com on December 10, 2010 17:58:13
PR 503779
the up-front dr_memory_is_readable() check on unaddr exceptions is wasteful
for heap header exceptions. moving it to after heap header and tls
exception checks results in a 10x speedup on twolf test!
it had >100K header exceptions and spent all its time in libc vfscanf
from the /proc/maps iteration: not sure why not using cached allmem.
now, we have no pc sampling hits at all in libc.
PR 503779: heap header exceptions very slow due to unnecessary
dr_memory_is_readable() check
moved dr_memory_is_readable() check to after heap header and tls
exception checks, which results in a 10x speedup on twolf test!
it had >100K header exceptions and spent all its time in libc vfscanf
from the /proc/maps iteration: not sure why not using cached allmem.
now, we have no pc sampling hits at all in libc.
=> unknown large speedup on vortex (always killed it before), 10x speedup
on twolf and mcf, 5x on perlbmk and vpr, 40% on gcc (all test inputs)
Original issue: http://code.google.com/p/drmemory/issues/detail?id=171
The text was updated successfully, but these errors were encountered: