Remove a member variable which I introduced in r5907 although it never actually got used. Restores binary compatibility (at least in that regard) to pre-r5907 video plugins, but breaks any binaries after that :P
Update FIFO watermark tightness recommendations. 1000 is quite a high value I guess, but some people seem to need it.
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@6060 8ced0084-cf51-0410-be5f-012b33b47a6e
Now it only writes back dirty registers when flushed
* Fixed KillImmediate
* Renamed LoadToX64 and StoreFromX64 to BindToRegister and StoreFromRegister respectively (as suggested by ector)
* Code cleanup in calls to the reg allocator
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@6053 8ced0084-cf51-0410-be5f-012b33b47a6e
platforms as well. It would seem that ABI_PARAMx will have to be
used more consistently elsewhere to match up with a change to use
them here.
Not sure yet why I still need the final section of Jit64::stX.
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@6035 8ced0084-cf51-0410-be5f-012b33b47a6e
* Un-cleaned the Load/Store code. I think this will fix the stability issues from r6032
* Added memory exception checking to a couple more JIT FPU instructions
* Optimised the 64bit fast memory access slightly
* Optimised the MMU speed hack.
The rest of the speed optimisations from r6032 have been retained.
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@6034 8ced0084-cf51-0410-be5f-012b33b47a6e
The GCC model for extended instructions like these is that you compile
with -msse3 etc. These affect code generation for whole compilation units,
so the idea is that you have a separate .c file for each instruction set
class and then indirect to the desired one at runtime.
Without e.g. -msse4.1, the GCC built-ins used by <foointrin.h> are not
available. However, in our specific case of compiling with -msse2 and
wanting to use SSE3.1 code, enough built-ins are available that we only
need to provide a little hack for pshufb.
Upgrading this to also use SSE4.1 instructions doesn't appear feasible
without a lot of undesirable duplication of GCC built-in functions and
headers, so we'd probably have to move to the GCC model of separate
source files for that.
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@6014 8ced0084-cf51-0410-be5f-012b33b47a6e
This one needs a lot of testing, since I don't have any games that need it, except Rogue Leader (GSWP64) - and I didn't do more "testing" than watching the stormtroopers dance due to speed reasons (altho it seems twice as fast as JIT for me; it does spit out a lot more warnings too)
btw, ITS OVER 6000!!111
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@6006 8ced0084-cf51-0410-be5f-012b33b47a6e
is required to correctly link with C++ libraries like wxw.
Tell OS X that the user is doing something when using the wiimote
so that the screensaver doesn't come on when playing a Wii game.
Replace unnecessary floating point with integer math.
Remove unnecessary <sys/stat.h>'s.
Correct a few type nits.
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@5995 8ced0084-cf51-0410-be5f-012b33b47a6e
* Emulated correct behaviour of DSI and ISI exceptions
* Added memory exception checks
* Added fast TLB cache implementation (written by booto)
* Added "Enable MMU" option in the game properties
* Renamed old TLBHack to "MMU speed hack"
Thanks to booto (PowerPC) and nash (testing) who spent many weeks of their time helping me make this work. Also thanks to shuffle2 for finding and converting the map file of the original target.
There are two options for MMU emulation found under the game properties. "Enable MMU" is the accurate emulation option. "MMU speed hack" is the old TLBHack renamed. Most games will work with one or the other. Some games can work with both options ticked.
Only the JIT recompiler and Interpreter work with the MMU code. JITIL is not supported (too hard for me to add).
The speed optimised code still needs a lot more work and is disabled for now.
Fixes issue 2546
Fixes issue 2583
Fixes issue 2704
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@5994 8ced0084-cf51-0410-be5f-012b33b47a6e
* Code cleanup
* Reduce the cost of RFI instruction (heavily used on PowerPC)
* Implement missing cases in bclrx
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@5991 8ced0084-cf51-0410-be5f-012b33b47a6e
This is EXPERIMENTAL, in theory don't break any game but if so Revert this commit immediately please.
Note: Beside Eternal Darkness needs DirectX 11 or DirectX 9 with Zelda Hack to works fine.
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@5986 8ced0084-cf51-0410-be5f-012b33b47a6e
PowerPC needs 2 instructions to load a 32-bits immediate value in a register (lis + addi; lis + ori). We can merge both instructions into a single SetImmediate().
This commit shares code with r5799, so to prevent another flame war on OSX vs Windows, I added #ifdef around the "bad" lines...
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@5982 8ced0084-cf51-0410-be5f-012b33b47a6e
* Merge some loops in PPCAnalyst::Flatten()
* Put ppcState.cr and ppcState.cr_fast[] into a single C++ union. This allows quick access to the whole CR register without needing to merge the cr_fast array. The implemented solution assumes the host system is little-endian, but it seems to be already assumed in many places in the code...
* Inline the call to computeCR: it now costs a few more memory bytes per JITed instruction but it removes the CPU overhead of the CALL. This allowed to remove some unneeded MOV as well.
* Jit64::GenerateCarry() don't need a temporary register anymore
* Fix what seems to be a bug in PowerPC instruction RFI: the wrong bit was cleared in MSR
git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@5970 8ced0084-cf51-0410-be5f-012b33b47a6e
