Commit graph

423 commits

Author SHA1 Message Date
Kelebek1 9d8f793969 Review 1 2021-02-15 05:26:28 +00:00
Kelebek1 fb54c38631 Implement texture offset support for TexelFetch and TextureGather and add offsets for Tlds
Formatting
2021-02-15 00:36:37 +00:00
ReinUsesLisp 82c2601555 video_core: Reimplement the buffer cache
Reimplement the buffer cache using cached bindings and page level
granularity for modification tracking. This also drops the usage of
shared pointers and virtual functions from the cache.

- Bindings are cached, allowing to skip work when the game changes few
  bits between draws.
- OpenGL Assembly shaders no longer copy when a region has been modified
  from the GPU to emulate constant buffers, instead GL_EXT_memory_object
  is used to alias sub-buffers within the same allocation.
- OpenGL Assembly shaders stream constant buffer data using
  glProgramBufferParametersIuivNV, from NV_parameter_buffer_object. In
  theory this should save one hash table resolve inside the driver
  compared to glBufferSubData.
- A new OpenGL stream buffer is implemented based on fences for drivers
  that are not Nvidia's proprietary, due to their low performance on
  partial glBufferSubData calls synchronized with 3D rendering (that
  some games use a lot).
- Most optimizations are shared between APIs now, allowing Vulkan to
  cache more bindings than before, skipping unnecesarry work.

This commit adds the necessary infrastructure to use Vulkan object from
OpenGL. Overall, it improves performance and fixes some bugs present on
the old cache. There are still some edge cases hit by some games that
harm performance on some vendors, this are planned to be fixed in later
commits.
2021-02-13 02:17:22 -03:00
Lioncash bcafef4b94 half_set: Resolve -Wmaybe-uninitialized warnings 2020-12-30 17:59:42 -05:00
ReinUsesLisp 9764c13d6d video_core: Rewrite the texture cache
The current texture cache has several points that hurt maintainability
and performance. It's easy to break unrelated parts of the cache
when doing minor changes. The cache can easily forget valuable
information about the cached textures by CPU writes or simply by its
normal usage.The current texture cache has several points that hurt
maintainability and performance. It's easy to break unrelated parts
of the cache when doing minor changes. The cache can easily forget
valuable information about the cached textures by CPU writes or simply
by its normal usage.

This commit aims to address those issues.
2020-12-30 03:38:50 -03:00
Lioncash 4c5f5c9bf3 video_core: Remove unnecessary enum class casting in logging messages
fmt now automatically prints the numeric value of an enum class member
by default, so we don't need to use casts any more.

Reduces the line noise a bit.
2020-12-07 00:41:50 -05:00
Lioncash f95602f152 video_core: Resolve more variable shadowing scenarios pt.3
Cleans out the rest of the occurrences of variable shadowing and makes
any further occurrences of shadowing compiler errors.
2020-12-05 16:02:23 -05:00
Lioncash 414a87a4f4 video_core: Resolve more variable shadowing scenarios pt.2
Migrates the video core code closer to enabling variable shadowing
warnings as errors.

This primarily sorts out shadowing occurrences within the Vulkan code.
2020-12-05 06:39:35 -05:00
Rodrigo Locatti fbda5e9ec9
Merge pull request #3681 from lioncash/component
decoder/image: Fix incorrect G24R8 component sizes in GetComponentSize()
2020-11-24 04:38:03 -03:00
bunnei a111a9ae2c
Merge pull request #4854 from ReinUsesLisp/cube-array-shadow
shader: Partially implement texture cube array shadow
2020-11-05 16:25:00 -08:00
ReinUsesLisp 657771bdcb shader: Partially implement texture cube array shadow
This implements texture cube arrays with shadow comparisons but doesn't
fix the asserts related to it.

Fixes out of bounds reads on swizzle constructors and makes them use
bounds checked ::at instead of the unsafe operator[].
2020-10-28 17:12:40 -03:00
ReinUsesLisp 44b552be71 shader/arithmetic: Implement FCMP immediate + register variant
Trivially add the encoding for this.
2020-10-28 17:05:41 -03:00
ReinUsesLisp dffaffaac1 shader/texture: Implement CUBE texture type for TMML and fix arrays
TMML takes an array argument that has no known meaning, this one appears
as the first component in gpr8 followed by s, t and r. Skip this
component when arrays are being used. Also implement CUBE texture types.

- Used by Pikmin 3: Deluxe Demo.
2020-10-07 23:17:46 -03:00
Lioncash e3a615a616 arithmetic_integer_immediate: Make use of std::move where applicable
Same behavior, minus any redundant atomic reference count increments and
decrements.
2020-09-24 13:28:45 -04:00
Rodrigo Locatti 31461589c5
Merge pull request #4672 from lioncash/narrowing
decoder/texture: Eliminate narrowing conversion in GetTldCode()
2020-09-17 21:17:54 +00:00
Lioncash 4944d48ee8 decode/image: Eliminate switch fallthrough in DecodeImage()
Fortunately this didn't result in any issues, given the block that code
was falling through to would immediately break.
2020-09-17 15:12:18 -04:00
Lioncash ffc66f089d decoder/texture: Eliminate narrowing conversion in GetTldCode()
The assignment was previously truncating a u64 value to a bool.
2020-09-17 15:04:17 -04:00
ReinUsesLisp eb914b6c50 video_core: Enforce -Werror=switch
This forces us to fix all -Wswitch warnings in video_core.
2020-09-16 17:48:01 -03:00
Lioncash dcc5562cd5 shader/memory: Amend UNIMPLEMENTED_IF_MSG without a message
We need to provide a message for this variant of the macro, so we can
simply log out the type being used.
2020-08-14 08:38:37 -04:00
Lioncash b724a4d90c General: Tidy up clang-format warnings part 2 2020-08-13 14:19:08 -04:00
bunnei f650cf8a9a
Merge pull request #4391 from lioncash/nrvo
video_core: Allow copy elision to take place where applicable
2020-07-24 06:33:09 -07:00
Rodrigo Locatti 9ea9a60e17
Merge pull request #4361 from ReinUsesLisp/lane-id
decode/other: Implement S2R.LaneId
2020-07-21 04:50:45 -03:00
Lioncash 6adc824d9d video_core: Allow copy elision to take place where applicable
Removes const from some variables that are returned from functions, as
this allows the move assignment/constructors to execute for them.
2020-07-21 00:36:13 -04:00
ReinUsesLisp 210cc0204d decode/other: Implement S2R.LaneId
This maps to host's thread id.

- Fixes graphical issues on Paper Mario.
2020-07-16 16:09:39 -03:00
ReinUsesLisp fbc232426d video_core: Rearrange pixel format names
Normalizes pixel format names to match Vulkan names. Previous to this
commit pixel formats had no convention, leading to confusion and
potential bugs.
2020-07-13 01:44:23 -03:00
bunnei efef7b1517
Merge pull request #4147 from ReinUsesLisp/hset2-imm
shader/half_set: Implement HSET2_IMM
2020-06-26 23:14:56 -04:00
ReinUsesLisp 39ab33ee1c shader/half_set: Implement HSET2_IMM
Add HSET2_IMM. Due to the complexity of the encoding avoid using
BitField unions and read the relevant bits from the code itself.
This is less error prone.
2020-06-22 20:51:18 -03:00
Morph 480e1fa987 decode/image: Implement B10G11R11F
- Used by Kirby Star Allies
2020-06-20 00:28:30 -04:00
ReinUsesLisp 5b2b6d594c shader/texture: Join separate image and sampler pairs offline
Games using D3D idioms can join images and samplers when a shader
executes, instead of baking them into a combined sampler image. This is
also possible on Vulkan.

One approach to this solution would be to use separate samplers on
Vulkan and leave this unimplemented on OpenGL, but we can't do this
because there's no consistent way of determining which constant buffer
holds a sampler and which one an image. We could in theory find the
first bit and if it's in the TIC area, it's an image; but this falls
apart when an image or sampler handle use an index of zero.

The used approach is to track for a LOP.OR operation (this is done at an
IR level, not at an ISA level), track again the constant buffers used as
source and store this pair. Then, outside of shader execution, join
the sample and image pair with a bitwise or operation.

This approach won't work on games that truly use separate samplers in a
meaningful way. For example, pooling textures in a 2D array and
determining at runtime what sampler to use.

This invalidates OpenGL's disk shader cache :)

- Used mostly by D3D ports to Switch
2020-06-05 00:24:51 -03:00
LC 9a0c1456e3
Merge pull request #4016 from ReinUsesLisp/invocation-info
shader/other: Fix hardcoded value in S2R INVOCATION_INFO
2020-06-02 09:47:53 -04:00
ReinUsesLisp f2d1aa97ad shader/other: Fix hardcoded value in S2R INVOCATION_INFO
Geometry shaders built from Nvidia's compiler check for bits[16:23] to
be less than or equal to 0 with VSETP to default to a "safe" value of
0x8000'0000 (safe from hardware's perspective). To avoid hitting this
path in the shader, return 0x00ff'0000 from S2R INVOCATION_INFO.

This seems to be the maximum number of vertices a geometry shader can
emit in a primitive.
2020-05-30 01:49:14 -03:00
ReinUsesLisp 32e6727dae shader/other: Implement MEMBAR.CTS
This silences an assertion we were hitting and uses workgroup memory
barriers when the game requests it.
2020-05-27 00:19:45 -03:00
bunnei 508242c267
Merge pull request #3981 from ReinUsesLisp/bar
shader/other: Implement BAR.SYNC 0x0
2020-05-26 14:40:13 -04:00
bunnei 623d9c47a2
Merge pull request #3980 from ReinUsesLisp/red-op
shader/memory: Implement non-addition operations in RED
2020-05-26 12:50:41 -04:00
ReinUsesLisp 5d0986a53b shader/other: Implement BAR.SYNC 0x0
Trivially implement this particular case of BAR. Unless games use OpenCL
or CUDA barriers, we shouldn't hit any other case here.
2020-05-21 23:20:43 -03:00
ReinUsesLisp 103809a0ca shader/memory: Implement non-addition operations in RED
Trivially implement these instructions. They are used in Astral Chain.
2020-05-21 23:19:46 -03:00
ReinUsesLisp e2b67a868b shader/other: Implement thread comparisons (NV_shader_thread_group)
Hardware S2R special registers match gl_Thread*MaskNV. We can trivially
implement these using Nvidia's extension on OpenGL or naively stubbing
them with the ARB instructions to match. This might cause issues if the
host device warp size doesn't match Nvidia's. That said, this is
unlikely on proper shaders.

Refer to the attached url for more documentation about these flags.
https://www.khronos.org/registry/OpenGL/extensions/NV/NV_shader_thread_group.txt
2020-05-21 23:18:37 -03:00
ReinUsesLisp 4e57f9d5cf shader_ir: Separate float-point comparisons in ordered and unordered
This allows us to use native SPIR-V instructions without having to
manually check for NAN.
2020-05-09 04:55:15 -03:00
bunnei e6b4311178
Merge pull request #3693 from ReinUsesLisp/clean-samplers
shader/texture: Support multiple unknown sampler properties
2020-05-02 00:45:41 -04:00
ReinUsesLisp 871aadbe36 shader/arithmetic_integer: Fix tracking issue in temporary
This temporary is not needed as we mark Rd.CC + IADD.X as unimplemented.
It caused issues when tracking global buffers.
2020-04-28 17:14:53 -03:00
ReinUsesLisp e895a4e2d7 shader/arithmetic_integer: Fix edge case and mark IADD.X Rd.CC as unimplemented
IADD.X Rd.CC requires some extra logic that is not currently
implemented. Abort when this is hit.
2020-04-25 22:58:33 -03:00
ReinUsesLisp 2a96bea6a7 shader/arithmetic_integer: Change IAdd to UAdd to avoid signed overflow
Signed integer addition overflow might be undefined behavior. It's free
to change operations to UAdd and use unsigned integers to avoid
potential bugs.
2020-04-25 22:57:54 -03:00
ReinUsesLisp c788f9c0bd shader/arithmetic_integer: Implement IADD.X
IADD.X takes the carry flag and adds it to the result. This is generally
used to emulate 64-bit operations with 32-bit registers.
2020-04-25 22:56:11 -03:00
ReinUsesLisp 255197e643 shader/arithmetic_integer: Implement CC for IADD 2020-04-25 22:55:26 -03:00
ReinUsesLisp ffc5ec6fa8 decode/register_set_predicate: Implement CC
P2R CC takes the state of condition codes and puts them into a register.
We already have this implemented for PR (predicates). This commit
implements CC over that.
2020-04-25 22:54:42 -03:00
ReinUsesLisp d523734266 decode/register_set_predicate: Use move for shared pointers
Avoid atomic counters used by shared pointers.
2020-04-25 22:54:14 -03:00
bunnei 4e37825dab
Merge pull request #3734 from ReinUsesLisp/half-float-mods
decode/arithmetic_half: Fix HADD2 and HMUL2 absolute and negation bits
2020-04-25 00:41:43 -04:00
bunnei 7c8acb0025
Merge pull request #3749 from ReinUsesLisp/lea-imm
shader/arithmetic_integer: Fix LEA_IMM encoding
2020-04-24 14:30:13 -04:00
ReinUsesLisp dbaebd8582 decode/arithmetic_half: Fix HADD2 and HMUL2 absolute and negation bits
The encoding for negation and absolute value was wrong.
Extracting is now done manually. Similar instructions having different
encodings is the rule, not the exception. To keep sanity and readability
I preferred to extract the desired bit manually.

This is implemented against nxas:
8dbc389957/table.h (L68)

That is itself tested against nvdisasm (Nvidia's official disassembler).
2020-04-23 18:29:38 -03:00
ReinUsesLisp 4fb921ff6b shader/texture: Support multiple unknown sampler properties
This allows deducing some properties from the texture instruction before
asking the runtime. By doing this we can handle type mismatches in some
instructions from the renderer instead of the shader decoder.

Fixes texelFetch issues with games using 2D texture instructions on a 1D
sampler.
2020-04-23 18:04:13 -03:00