Bug 512865. qcms: Improve SSE2 performance, add SSE support. r=jrmuizel
This patch greatly improves the performance of QCMS transformations on x86 &
x86_64 systems. Some notes:
0. On 32-bit x86 systems it does runtime selection between non-SIMD, SSE, and
SSE2 code paths.
1. On x86_64 systems the SSE2 code path is always taken. The non-SIMD and SSE
code paths are left intact, but contemporary versions of the GCC and MSVC
compilers will see that they cannot be reached and optimize them away.
2. The execution of the SSE2 code path is reduced by 67%, relative to the
original Intel/Microsoft formatted ASM code. The relative performance is seen
on a Pentium4 (Northwood) 2.4GHz CPU with DDR1 RAM.
3. The SSE code path provides a 80% reduction in execution time, relative to
the non-SIMD code path. The relative performance is seen on a Pentium3
(Coppermine) 1.26GHz CPU with SDRAM.
4. The code has been split out into separate files so that it can be built
with different cflags (-msse, and -msse2) when using gcc.
5. Try to land again, this time with __attribute__((__force_align_arg_pointer__))
to avoid crashes on linux.
History:
--------
v0.0.10a - 20081004 (AndersK)
*Added SEE_MASK_NOZONECHECKS flag (experimental)
v0.0.10 - 20080812 (AndersK)
*Added ugly hook hack to the shells run-as dialog on xp, defaults to other user
v0.0.9 - 20080721 (AndersK)
*Fixed UAC_RealWorldFullyLoadedDualModeExample.nsi related bug (Thanks Case)
v0.0.8 - 20080310 (AndersK)
+HTML Readme
+Added UAC::GetOuterHwnd (used by UAC_RealWorldFullyLoadedDualModeExample.nsi)
*Fixed UAC_RealWorldFullyLoadedDualModeExample.nsi
*Major code cleanup in UAC.cpp
-Removed UAC::RunElevatedAndProcessMessages (UAC::RunElevated now supports non NULL $HWNDParent)
-Removed several useless sample scripts
v0.0.7e - 20080229 (AndersK)
*Added ugly hack for hackwnd to find correct title and give us a proper taskbar so the elevation dialog does not get lost (2000,XP (This also fixed Alt-Tab icon on Vista))
*Should compile with MSVC2005 now (Thanks Case)
*More unicode fixes, this time even tested with NSIS Unicode (Only RunElevated and Exec tested)
v0.0.7d - 20080226 (AndersK)
*Fixed a couple of unicode version bugs (Unicode version still untested)
*Fixed weird XP string length bug (Thanks kfank)
v0.0.7c - 20080218 (AndersK)
*Fixed SyncVars string length bug
v0.0.7b - 20080205 (AndersK)
*Fixed DelayLoadDlls() problem on NT4
v0.0.7 - 20080120 (AndersK)
+Added UAC::StackPush (For use with ExecCodeSegment)
v0.0.6d - 20071108 (AndersK)
+Now syncs basic registers/variables before calling UAC::*Exec* and UAC::ExecCodeSegment (r0-r9,R0-R9,$CMDLINE,$INSTDIR,$OUTDIR,$EXEDIR,$LANGUAGE)
+Added UAC::RunElevatedAndProcessMessages, this can be called after .onInit (Very experimental, DO NOT USE)
+New include file with helper macros: UAC.nsh
*Replazed Clammerz hack with a better version
v0.0.6c - 20071014 (AndersK)
+Check for and split up "domain\user" style input in RunAs.cpp for CreateProcessWithLogonW
*Added a ugly hack to trick messagebox'es in .OnInit to appear correctly on Vista (Thanks Clammerz)
v0.0.6b - 20070523 (AndersK)
*Fixed showwindow flag (Thanks for the help kichik)
v0.0.6 - 20070512 (AndersK)
+Added basic language support for MyRunAs dialog.
v0.0.5e - 20070509 (AndersK)
*Fixed detection of UAC mode?
+IPC window is visible (but offscreen) during elevation to help with SetForegroundWindow/Focus problems
v0.0.5d - 20070324 (AndersK)
*Fixed stupid IsAdmin bug
v0.0.5c - 20070304 (AndersK)
*_IsAdmin now uses CheckTokenMembership if it exists ( MSKB:Q118626 / http://blogs.msdn.com/larryosterman/archive/2007/03/14/why-does-kb-118626-use-accesscheck-to-check-if-you-re-a-member-of-the-administrators-group.aspx )
v0.0.5b - 20070301 (AndersK)
*Fixed ExecCodeSegment (Thread now calls CoInitialize)
v0.0.5 - 20070228 (AndersK)
+Added ExecCodeSegment (You can now call ANY code in the context of the original user)
v0.0.4b - 20070226 (AndersK)
*Fixed (My)RunAs font (http://blogs.msdn.com/oldnewthing/archive/2005/02/04/366987.aspx)
v0.0.4 - 20070225 (AndersK)
+Added (My)RunAs dialog, used on Vista when running as LUA with UAC off
+Always uses /NCRC for elevated instance
*Now compiles as UNICODE (Untested, no UnicodeNSIS to test on)
v0.0.3 - 20070224 (AndersK)
+Added Exec/ExecWait
+Added Verb & ShowWindow support for ShellExec[Wait]
v0.0.2 - 20070219 (AndersK)
+Added ShellExecWait
*IPC srv wnd now has its own thread and msg loop
*Removed CRT dependency
*Hopefully loads on Win95 now
v0.0.1 - 20070215 (AndersK)
*Initial release
