14 Apr
2005
14 Apr
'05
8:07 p.m.
Rene Rivera wrote:
I running the timing tests again, with some changes to see if I can get more representative numbers.
If you can get the assembly version to outperform the generic Win32 version, you mean ;-) But out of curiosity, what changes? I've been thinking to add a more representative shared_ptr and weak_ptr timing tests. The _mt versions aren't bad, though.
But attached is the end result.
// detail/sp_counted_base_gcc_x86.hpp - g++ on 486+ or AMD64
You should probably fix this line...
// Copyright (c) 2001, 2002, 2003 Peter Dimov and Multi Media Ltd. // Copyright 2004-2005 Peter Dimov
... and add your name here.