4078dcd917
Support save/restore of context on list run/finish.
2013-09-20 09:51:44 -07:00
bd70d04930
Include GE_CMD_PATCHFACING just in case.
2013-09-20 00:45:48 -07:00
50e9e45d65
Check version in each DoState() func.
...
They bail on PointerWrap error or bad version.
2013-09-14 20:23:03 -07:00
8ae6694e1d
Use a waiting thread list in sceGe as well.
2013-09-08 11:57:44 -07:00
08b41df6a2
Change a few more log types.
2013-09-07 13:31:14 -07:00
8998a1b303
Don't trash a list the GE interrupt handler needs.
...
Even after it's marked COMPLETED, the CPU needs it to start and finish the
interrupt. Fixes #2956 , where an existing completed list was not having
its interrupt run correctly.
2013-09-01 12:39:45 -07:00
dc9b5156d9
Cleanup comment (broke savestates anyway.)
2013-09-01 00:52:25 -07:00
15a0f39fa1
Return yet more errors while inside interrupts.
2013-08-28 23:15:13 -07:00
211c2bd955
Refuse a few more waits w/ dispatch/intr disabled.
2013-08-28 22:35:05 -07:00
4d9eb24f29
Fix hang in Project Diva 2nd in single-threaded mode. Doesn't seem to affect other games, but I don't have every game in existence, so there could be one or two out there that might break from this.
2013-08-24 21:13:34 -04:00
8dbcf6572e
Add some additional reporting.
2013-08-24 11:34:38 -07:00
520453a5bd
Fix gpu cycle call level counting, of course halving the numbers I mentioned in the previous commit comment.
2013-08-23 11:42:15 +02:00
dfb91d4532
Track number of GPU commands executed per call level.
...
This is to get an idea if it would be beneficial to compile and cache display lists. Looks like it would be for some games, for example in GTA it's often 20k/80k/20k commands per frame - that is 100k executed commands per frame that could potentially just be executed instead of interpreted. Likewise in Wipeout. Of course, this means yet another cache with invalidation issues etc..
2013-08-23 11:29:55 +02:00
1ed8edb0d3
Avoid some dangerous hex constant widths.
2013-08-22 23:23:48 -07:00
e68ddf671c
Temporary fix issue #3167 in non-multithread mode #2
2013-08-15 03:44:20 +08:00
49ab98c7fd
Use atomic operations on Android + warning fixes.
2013-08-12 08:51:49 -07:00
cf7c718706
Use a timer to keep gpu/cpu in sync periodically.
...
Fixes Diva Extend demo, at least, losing FPS.
2013-08-11 13:41:42 -07:00
e0f699fd5f
Use a possibly harmless hack to prevent hangs.
...
Not sure where the problem is, but this fixes it pretty consistently for
me... seems like lists are somehow not being processed after becoming
processable?
2013-08-10 20:03:43 -07:00
26c072df51
Don't wait directly from GPUCommon, do it in sceGe.
...
Makes debugging a bit easier.
2013-08-10 18:24:40 -07:00
55c07bdc73
Avoid a possible thread sync issue.
...
Could be that it's about to wake listsync, does, and then we wait.
2013-08-10 17:18:31 -07:00
abc396cbe1
Sync the GPU thread on list/draw sync.
...
Otherwise, the CPU gets ahead (inside the frame) and games get confused.
I don't think there's any good workaround to avoid this.
2013-08-10 15:16:49 -07:00
450e4c7fea
Factor out event queue logic to a template mixin.
2013-08-10 03:33:09 -07:00
a80a7690c8
Fix SyncThread() not syncing during last event.
2013-08-09 23:18:37 -07:00
e82bae3708
Simplify easy_guard to avoid coding errors.
2013-08-09 22:57:53 -07:00
1226c3dbd9
Try to prevent a deadlock if CPU doesn't NEXTFRAME.
2013-08-09 22:57:53 -07:00
e71cf2ec7e
Don't queue Flush() etc. while on GPU thread.
...
Oops, that makes it not actually flush properly of course.
Also made a simpler guard wrapper to avoid extra braces in too many
places. Reduced contention areas slightly.
2013-08-09 22:57:52 -07:00
68fb78e203
Be more careful with the stall address.
...
In case it's moved while running the list.
2013-08-09 22:57:52 -07:00
8c69020fa8
Use separate condition vars/locks for GPU sync.
2013-08-09 22:57:51 -07:00
b0e6967697
Rename UseCPUThread to SeparateCPUThread.
2013-08-09 22:41:33 -07:00
a56cac1eb3
Bail when CPU thread disabled to ensure good perf.
2013-08-09 22:41:32 -07:00
3819886e2e
Add a way to synchronize with the GPU thread.
2013-08-09 22:41:32 -07:00
02e301e5fe
Add a simple message event system to the GE.
...
With this, the CPU thread feature now mostly works.
2013-08-09 22:41:31 -07:00
52ca93aab2
Add mutex locks around list management.
...
This way we can safely execute lists on a separate thread.
2013-08-09 22:41:31 -07:00
00fc5ec40e
Enable interrupts as a property of the list.
...
This way there's no behavior change when enqueue is async.
2013-08-07 00:32:30 -07:00
3aa2db9a8e
Simplify ProcessDLQueue() slightly.
2013-08-04 16:31:54 -07:00
f7a39d1b12
Move ReapplyGfxState() to GPUCommon.
...
This way it can be overridden, and simplifies some other things.
2013-08-04 15:15:50 -07:00
a95daf4647
Reduce logspam for some GE signals. doesn't seem to be a problem (although that might change if we paralellize).
2013-06-27 17:49:29 +02:00
9c85bd92e4
Show statistics for GPU cycles executed per frame.
2013-05-31 10:40:16 -07:00
46e5b84242
Small opt. for common immediately stalled list.
2013-04-28 14:56:38 -07:00
597cdbca3f
Simplify, don't let InterpretList do downcount.
2013-04-28 14:30:28 -07:00
1edbfa3212
Use a downcount and fast run loop in the GE.
2013-04-28 14:23:30 -07:00
5a03888b56
Centralize state and pc changes in the GE.
...
To make later things easier.
2013-04-28 13:34:29 -07:00
9208f6389d
Avoid calling time_update in GE w/o debug stats.
...
Just wastes time, especially if the game hammers sceGeListUpdateStallAddr,
which many games do.
2013-04-22 22:04:27 -07:00
628c223234
GPUCommon: Get rid of prev variable.
2013-04-21 19:32:56 +02:00
b54eb41041
Don't signal again if a PAUSE handler unstalls.
...
Basically, if the handler calls sceGeListUpdateStallAddr(), and it hits
a finish (which it might), before we were triggering another PAUSE.
THat's wrong, it should just be a FINISH as usual.
2013-04-20 20:15:47 -07:00
98daf18d56
Add skeleton code for Windows multitouch, to be filled in by someone who can test :P
2013-04-16 21:14:46 +02:00
f86e01707f
Use a struct for the GE stack entries, cleaner.
2013-04-15 23:06:56 -07:00
7b184602df
Remember the GE offset in the stack on call/ret.
2013-04-15 22:20:49 -07:00
ea11c36091
Fix a vshader id bit collision, avoid color test when set to 'ALWAYS', zeroinitialize dls.
2013-04-15 22:57:54 +02:00
cbbc4e2c9e
Reset gstate_c.offsetAddr on RET. Fixes some "CALL to illegal address"
...
(as seen in Flatout for example).
2013-04-15 22:24:30 +02:00
Unknown W. Brackets