[pypy-issue] [issue766] Shadowstack root finder should be competitive with asmgcc in speed

[Dan Villiom Podlaski Christiansen]

Dan Villiom Podlaski Christiansen <danchr at gmail.com> added the comment:

I did another benchmark run a couple of days ago; a full self-hosting translation 
of PyPy. Shadowstack did it ~2% slower, but that's including the C code.

/tmp/pypy-c-jit-45336-6d1937f2d0a9-osx64-shadowstack-clang/bin/pypy
[[Timer] Timings:
[Timer] annotate                       --- 1038.1 s
[Timer] rtype_lltype                   ---  507.8 s
[Timer] pyjitpl_lltype                 ---  777.1 s
[Timer] backendopt_lltype              ---  272.0 s
[Timer] stackcheckinsertion_lltype     ---   53.2 s
[Timer] database_c                     ---  439.4 s
[Timer] source_c                       ---  488.7 s
[Timer] compile_c                      ---  925.6 s
[Timer] ===========================================
[Timer] Total:                         --- 4502.0 s
$i ./translate.py -Ojit --batch  3520,69s user 45,93s system 78% cpu 1:16:03,38 
total

/tmp/pypy-c-jit-45336-6d1937f2d0a9-osx64/bin/pypy
[Timer] Timings:
[Timer] annotate                       --- 1044.8 s
[Timer] rtype_lltype                   ---  516.2 s
[Timer] pyjitpl_lltype                 ---  748.4 s
[Timer] backendopt_lltype              ---  262.9 s
[Timer] stackcheckinsertion_lltype     ---   58.6 s
[Timer] database_c                     ---  420.9 s
[Timer] source_c                       ---  484.4 s
[Timer] compile_c                      ---  950.0 s
[Timer] ===========================================
[Timer] Total:                         --- 4486.0 s
$i ./translate.py -Ojit --batch  3457,19s user 48,06s system 76% cpu 1:16:01,20 
total

----------
status: unread -> chatting

________________________________________
PyPy bug tracker <tracker at bugs.pypy.org>
<https://bugs.pypy.org/issue766>
________________________________________