Yuri Kunde Schlesner
e4e962bc7c
VideoCore/Shader: Remove dynamic control flow in (Get)UniformOffset
2016-12-15 23:08:05 -08:00
Yuri Kunde Schlesner
d27cb1dedc
VideoCore/Shader: Move DebugData to a separate file
2016-12-15 23:08:05 -08:00
Yuri Kunde Schlesner
fb9e856b91
shader_jit_x64: Use LOOPCOUNT_REG as a 64-bit reg when indexing
2016-12-15 10:02:42 -08:00
Yuri Kunde Schlesner
f00ada3363
VideoCore: Eliminate an unnecessary copy in the drawcall loop
2016-12-14 21:00:29 -08:00
Yuri Kunde Schlesner
5ff3206207
shader_jit_x64: Use Reg32 for LOOP* registers, eliminating casts
2016-12-14 20:06:09 -08:00
Yuri Kunde Schlesner
f4e98ecf3f
VideoCore: Convert x64 shader JIT to use Xbyak for assembly
2016-12-14 20:06:08 -08:00
Jannik Vogel
2d8097eecc
shader_jit: Fix non-SSE4.1 path where FLR would not truncate
2016-12-04 04:26:33 +01:00
Jannik Vogel
e2cb7d7833
shader_jit: Load LOOPCOUNT_REG and LOOPINC 4 bit left-shifted
2016-12-02 04:33:15 +01:00
Yuri Kunde Schlesner
d9a904f9cb
VideoCore: Shader interpreter cleanups
2016-09-29 21:15:49 -07:00
Yuri Kunde Schlesner
26b68313b9
VideoCore: Fix out-of-bounds read in ShaderSetup::ProduceDebugInfo
...
As far as I can tell, memset was replaced by a fill without correcting
the parameter type, causing an out-of-bounds array read in the Vec4
constructor.
2016-09-29 21:11:36 -07:00
Yuri Kunde Schlesner
f120e78b56
Remove special rules for Windows.h and library includes
2016-09-21 00:16:33 -07:00
Yuri Kunde Schlesner
84fbbe2629
Use negative priorities to avoid special-casing the self-include
2016-09-21 00:15:56 -07:00
Emmanuel Gil Peyrot
ebdae19fd2
Remove empty newlines in #include blocks.
...
This makes clang-format useful on those.
Also add a bunch of forgotten transitive includes, which otherwise
prevented compilation.
2016-09-21 11:15:47 +09:00
Yuri Kunde Schlesner
396a8d91a4
Manually tweak source formatting and then re-run clang-format
2016-09-18 21:14:25 -07:00
Emmanuel Gil Peyrot
dc8479928c
Sources: Run clang-format on everything.
2016-09-18 09:38:01 +09:00
Yuri Kunde Schlesner
a3afeb4687
VideoCore: Fix dangling lambda context in shader interpreter
...
The static meant that after the first execution, these lambda context
would be pointing to a random location on the stack. Fixes a random
crash when using the interpreter.
2016-09-15 22:15:11 -07:00
Jannik Vogel
ff0fa86b17
Retrieve shader result from new OutputRegisters-type
2016-05-16 18:55:51 +02:00
Jannik Vogel
1308afe2c2
Use new shader-jit signature for interpreter
2016-05-13 09:41:55 +02:00
Jannik Vogel
4e01e9ffc5
Refactor access to state in shader-jit
2016-05-13 09:20:14 +02:00
Jannik Vogel
7e756faaba
Move program_counter and call_stack from UnitState to interpreter
2016-05-12 19:05:42 +02:00
Jannik Vogel
6c6d99ca51
Move default_attributes into Pica state
2016-05-12 19:05:41 +02:00
bunnei
f6eb62d062
Merge pull request #1690 from JayFoxRox/tex-type-3
...
Pica: Implement texture type 3 (Projection2D)
2016-05-11 21:47:08 -04:00
Jannik Vogel
ae7a82fa1c
Turn ShaderSetup into struct
2016-05-11 23:48:24 +02:00
Jannik Vogel
2f8e8e1455
Pica: Add tc0.w to OutputVertex
2016-05-11 08:07:36 +02:00
Jannik Vogel
696cb197a5
Pica: Replace logic in shader.cpp with loop
2016-05-03 01:40:47 +02:00
Emmanuel Gil Peyrot
691a42fe98
VideoCore: Run include-what-you-use and fix most includes.
2016-04-30 17:02:41 +01:00
bunnei
90243c56fb
Merge pull request #1730 from hrydgard/vertex-loader
...
* Remove late accesses to attribute_config
* Refactor: Extract VertexLoader from command_processor.cpp.
Preparation for a similar concept to Dolphin or PPSSPP. These can be JIT-ed and cached.
* Move "&" to their proper place, add missing includes and make some properly relative.
* Don't keep base_address in the loader, it doesn't belong there (with it, the loader can't be cached).
* Optimize the vertex loader, nearly doubling its speed.
* Debugger fix
* Move and rename the MemoryAccesses class to MemoryAccessTracker.
2016-04-29 09:42:47 -04:00
Yuri Kunde Schlesner
e3a8292495
Common: Remove section measurement from profiler ( #1731 )
...
This has been entirely superseded by MicroProfile. The rest of the code
can go when a simpler frametime/FPS meter is added to the GUI.
2016-04-29 00:07:10 -07:00
Henrik Rydgard
47ff008817
Refactor: Extract VertexLoader from command_processor.cpp.
...
Preparation for a similar concept to Dolphin or PPSSPP. These can be JIT-ed and cached.
2016-04-28 19:05:55 +02:00
Sam Spilsbury
656a442433
shader: Shader size is long uint, not uint.
2016-04-25 00:40:03 +08:00
Sam Spilsbury
c6709d97bc
shader: Handle non-CALL opcodes with a break
2016-04-25 00:39:54 +08:00
Sam Spilsbury
bbffa6ad69
shader: Format string must be provided inline and not as a variable
2016-04-24 23:40:52 +08:00
bunnei
d7fe2784cc
shader_jit_x64: Rename RuntimeAssert to Compile_Assert.
2016-04-13 23:04:53 -04:00
bunnei
3f623b2561
shader_jit_x64.cpp: Rename JitCompiler to JitShader.
2016-04-13 23:04:53 -04:00
bunnei
847fb951e2
shader_jit_x64: Free memory that's no longer needed after compilation.
2016-04-13 23:04:52 -04:00
bunnei
60aa72e117
shader_jit_x64: Use a sorted vector instead of a set for keeping track of return addresses.
2016-04-13 23:04:52 -04:00
bunnei
60749f2cda
shader_jit_x64: Use CALL/RET instead of JMP for subroutines.
2016-04-13 23:04:52 -04:00
bunnei
1d45b57939
shader_jit_x64: Separate initialization and code generation for readability.
2016-04-13 23:04:50 -04:00
bunnei
6e0319eec9
shader_jit_x64: Get rid of unnecessary last_program_counter variable.
2016-04-13 23:04:49 -04:00
bunnei
f3afe24594
shader_jit_x64: Execute certain asserts at runtime.
...
- This is because we compile the full shader code space, and therefore its common to compile malformed instructions.
2016-04-13 23:04:49 -04:00
bunnei
ffcf7ecee9
shader: Remove unused 'state' argument from 'Setup' function.
2016-04-13 23:04:48 -04:00
bunnei
a5a74eb121
shader_jit_x64: Specify shader main offset at runtime.
2016-04-13 23:04:47 -04:00
bunnei
c9d10de644
shader_jit_x64: Allocate each program independently and persist for emu session.
2016-04-13 23:04:47 -04:00
bunnei
4632791a40
shader_jit_x64: Rewrite flow control to support arbitrary CALL and JMP instructions.
2016-04-13 23:04:44 -04:00
bunnei
135aec7bea
shader_jit_x64: Fix strict memory aliasing issues.
2016-04-13 23:04:43 -04:00
Mathew Maidment
aa6380e5bc
Merge pull request #1643 from MerryMage/make_unique
...
Common: Remove Common::make_unique, use std::make_unique
2016-04-05 20:10:11 -04:00
MerryMage
a06dcfeb61
Common: Remove Common::make_unique, use std::make_unique
2016-04-05 13:31:17 +01:00
bunnei
ebbba0d381
Merge pull request #1508 from JayFoxRox/vs-output-map
...
Respect vs output map
2016-03-22 11:59:12 -04:00
bunnei
784c5539ea
Merge pull request #1538 from lioncash/dot
...
shader_interpreter: use std::inner_product for the dot product
2016-03-20 00:35:06 -04:00
Lioncash
63e956cc7a
video_core: Don't cast away const
2016-03-17 02:01:38 -04:00
Lioncash
4d89df8df2
shader_interpreter: use std::inner_product for the dot product
...
Same thing, less code.
2016-03-17 01:00:30 -04:00
bunnei
96cafbe4cc
Merge pull request #1503 from bunnei/clear-jit-cache
...
Clear JIT cache
2016-03-16 13:18:51 -04:00
Jannik Vogel
9aad2f29bb
PICA: Fix MAD/MADI encoding
2016-03-15 20:01:25 +01:00
Jannik Vogel
f746a00964
Respect vs output map
2016-03-14 13:03:34 +01:00
bunnei
6efb710b28
shader_jit_x64: Clear cache after code space fills up.
2016-03-12 12:15:49 -05:00
bunnei
c103759cdc
shader_jit_x64: Make assert outputs more useful & cleanup formatting.
2016-03-12 12:06:28 -05:00
bunnei
46f78b7f19
shader: Update log message to use proper log class.
2016-03-12 12:03:32 -05:00
Lioncash
88d604383e
Common: Get rid of alignment macros
...
The gl rasterizer already uses alignas,
so we may as well move everything over.
2016-03-09 01:31:14 -05:00
Dwayne Slater
6b775034dd
Add immediate mode vertex submission
2016-03-02 22:16:38 -05:00
bunnei
b003075570
pica: Implement decoding of basic fragment lighting components.
...
- Diffuse
- Distance attenuation
- float16/float20 types
- Vertex Shader 'view' output
2016-02-05 17:17:28 -05:00
bunnei
a43f8d2fb7
Merge pull request #1367 from yuriks/jit-jmp
...
Shader JIT: Fix off-by-one error when compiling JMPs
2016-01-27 09:19:28 -05:00
Yuri Kunde Schlesner
083d2d89a5
Shader: Implement "invert condition" feature of IFU instruction
...
If the bit 0 of the JMPU instruction is set, then the jump condition
will be inverted. That is, a jump will happen when the boolean is false
instead of when it is true.
2016-01-24 20:29:06 -08:00
Yuri Kunde Schlesner
c1071c1ff7
Shader JIT: Fix off-by-one error when compiling JMPs
...
There was a mistake in the JMP code which meant that one instruction at
the destination would be skipped when the jump was taken. This commit
also changes the meaning of the culprit parameter to make it less
confusing and avoid similar mistakes in the future.
2016-01-24 02:15:56 -08:00
Lioncash
aec28ed91e
video_core: Reorganize headers
2015-09-11 07:31:15 -04:00
Lioncash
1fa772393b
video_core: Remove unnecessary includes from headers
2015-09-11 00:10:03 -04:00
Lioncash
526eb33d1e
video_core: Remove unused variables
2015-09-10 10:26:21 -04:00
aroulin
1484a23530
Shader JIT: Use SCALE constant from emitter
2015-09-07 16:50:28 +02:00
aroulin
87e3b9ffc0
Shader: Fix size_t to int casts of register offsets
2015-09-07 16:50:28 +02:00
bunnei
918ca40c68
Merge pull request #1088 from aroulin/x64-emitter-abi-call
...
x64: Proper stack alignment in shader JIT function calls
2015-09-02 08:46:58 -04:00
aroulin
ba998b85a1
video_core: Fix format specifiers warnings
2015-09-02 08:20:00 +02:00
aroulin
179ad35c2e
x64: Proper stack alignment in shader JIT function calls
...
Import Dolphin stack handling and register saving routines
Also removes the x86 parts from abi files
2015-09-01 23:39:52 +02:00
aroulin
84959be150
Shader JIT: Fix SGE/SGEI NaN behavior
...
SGE was incorrectly emulated w.r.t. NaN behavior as the CMPSS SSE
instruction was used with NLT
2015-08-31 08:16:15 +02:00
Yuri Kunde Schlesner
c5a4025b65
Merge pull request #1065 from yuriks/shader-fp
...
Shader FP compliance fixes
2015-08-27 16:34:13 -07:00
aroulin
f52d8c1a9b
Shader JIT: Fix float to integer rounding in MOVA
...
MOVA converts new address register values from floats to integers using truncation
2015-08-27 15:26:41 +02:00
archshift
dd0e1061ef
Shader JIT: ifdef out reference to ifdef'd out shader_map
...
shader_map was only defined on x86 architectures, but was cleared on shutdown
with no ifdef protection. Ifdef this out so non-x86 architectures can be built.
2015-08-26 22:28:19 +00:00
Yuri Kunde Schlesner
0fcabd2b11
Integrate the MicroProfile profiling library
...
This brings goodies such as a configurable user interface and
multi-threaded timeline view.
2015-08-24 22:16:28 -03:00
Yuri Kunde Schlesner
d8ef20c856
Shader JIT: Tiny micro-optimization in DPH
2015-08-24 01:48:37 -03:00
Yuri Kunde Schlesner
630a850d4d
Shaders: Fix multiplications between 0.0 and inf
...
The PICA200 semantics for multiplication are so that when multiplying
inf by exactly 0.0, the result is 0.0, instead of NaN, as defined by
IEEE. This is relied upon by games.
Fixes #1024 (missing OoT interface items)
2015-08-24 01:48:15 -03:00
Yuri Kunde Schlesner
082b74fa24
Shaders: Explicitly conform to PICA semantics in MAX/MIN
2015-08-24 01:46:58 -03:00
Yuri Kunde Schlesner
76247170df
Shader JIT: Add name to second scratch register (XMM4)
2015-08-24 01:46:10 -03:00
Lioncash
fa5076eb9b
shader_jit: Replace two MDisp usages with MatR
2015-08-24 00:39:50 -04:00
Yuri Kunde Schlesner
455147ee95
Shader JIT: Fix CMP NaN behavior to match hardware
2015-08-24 01:29:40 -03:00
bunnei
83c214f6d8
Merge pull request #1062 from aroulin/shader-rcp-rsq
...
Shader: RCP and RSQ computes only the 1st component
2015-08-23 17:56:35 -04:00
aroulin
03c5cfead4
Shader: Use std::sqrt for float instead of sqrt
2015-08-23 22:03:07 +02:00
aroulin
fa552f11ef
Shader: RCP and RSQ computes only the 1st component
2015-08-23 22:01:17 +02:00
aroulin
2f1514b904
Shader: implement DPH/DPHI in JIT
2015-08-22 11:09:53 +02:00
aroulin
2e7cf2f6cf
Shader: implement DPH/DPHI in interpreter
...
Tests revealed that the component with w=1 is
SRC1 and not SRC2, it is now fixed on 3dbrew.
2015-08-22 11:09:53 +02:00
aroulin
f3e8f42718
Shader: implement SGE, SGEI and SLT in JIT
2015-08-19 14:29:39 +02:00
aroulin
863730f6a7
Shader: implement SGE, SGEI in interpreter
2015-08-19 14:29:39 +02:00
aroulin
2f9eb98f03
Shader: Save caller-saved registers in JIT before a CALL
2015-08-19 03:40:07 +02:00
aroulin
7d3a6016d6
Shader: implement EX2 and LG2 in JIT
2015-08-17 01:12:34 +02:00
aroulin
638e47c04d
Shader: implement EX2 and LG2 in interpreter
2015-08-16 15:54:30 +02:00
Tony Wasserka
96820ae42a
Build fix for Debug configurations.
2015-08-16 15:14:54 +02:00
Tony Wasserka
33ba604fd9
Introduce a shader tracer to allow inspection of input/output values for each processed instruction.
2015-08-16 14:12:11 +02:00
Tony Wasserka
4cb302c8ae
citra-qt: Improve shader debugger.
...
Now supports dumping the current shader and recognizes a larger number of output semantics.
2015-08-16 13:22:00 +02:00
bunnei
db97090cad
Shader: Use a POD struct for registers.
2015-08-15 18:03:27 -04:00
bunnei
b39c053785
Rename ARCHITECTURE_X64 definition to ARCHITECTURE_x86_64.
2015-08-15 18:03:27 -04:00
bunnei
0ee00861f6
Common: Cleanup CPU capability detection code.
2015-08-15 18:03:26 -04:00
bunnei
a1942238f5
Common: Move cpu_detect to x64 directory.
2015-08-15 18:03:26 -04:00
bunnei
bd7e691f78
x64: Refactor to remove fake interfaces and general cleanups.
2015-08-15 18:03:25 -04:00
bunnei
cfb354f11f
JIT: Support negative address offsets.
2015-08-15 18:01:22 -04:00
bunnei
094ae6fadb
Shader: Initial implementation of x86_x64 JIT compiler for Pica vertex shaders.
...
- Config: Add an option for selecting to use shader JIT or interpreter.
- Qt: Add a menu option for enabling/disabling the shader JIT.
2015-08-15 18:01:07 -04:00
bunnei
d67e2f78b7
Common: Added MurmurHash3 hash function for general-purpose use.
2015-08-15 17:33:46 -04:00
bunnei
3f69c2039d
Shader: Define a common interface for running vertex shader programs.
2015-08-15 17:33:44 -04:00
bunnei
18527b9e21
Shader: Move shader code to its own subdirectory, "shader".
2015-08-15 17:33:42 -04:00
