- Add device detection for Snapdragon 8 Elite (device ID 0x43052c01)
- Disable shader float controls and 64-bit integer atomics on Qualcomm
proprietary drivers to work around driver limitations
- Enable native BGR format support on S8 Elite (similar to S8 Gen 2)
- Implement VOTE_vtg instruction for vertex/tessellation/geometry shaders
with proper single-thread semantics
- Replace stub warnings with detailed comments for FCSM_TR flow test and
SR_WSCALEFACTOR special registers to document expected behavior
These changes improve compatibility with Qualcomm Snapdragon 8 Elite
devices and reduce spurious warnings by documenting shader features
that are correctly handled as conservative defaults.
Signed-off-by: Zephyron <zephyron@citron-emu.org>
Resolves compilation errors in move_register.cpp by using proper
register offset approach instead of non-existent Y/Z/W methods.
Signed-off-by: Zephyron <zephyron@citron-emu.org>
- Implement FSWZADD NDV mode support by removing stubbed warning
- Add proper SAM and RAM instruction implementations (no-op with logging)
- Implement comprehensive masked move operations for all mask patterns
- Remove all shader recompiler stubbed instruction warnings
Thanks to Dr.Stug for providing the logs that identified these missing implementations.
Signed-off-by: Zephyron <zephyron@citron-emu.org>
The SURED (Surface Reduction) instruction can operate on cube maps and 3D
texture arrays which were previously missing from the Type enum, causing
shader translation failures.
Changes:
- Add ARRAY_3D and CUBE to Type enum
- Map ARRAY_3D to Color3D and CUBE to ColorCube in GetType()
- Update MakeCoords() to handle array and cube coordinate generation
Signed-off-by: Zephyron <zephyron@citron-emu.org>
The Vulkan spec requires fragment shaders to declare an output covering
Location 0, Component 3 (alpha) when alpha-to-coverage is enabled. This change:
- Tracks alpha_to_coverage_enabled through RuntimeInfo from pipeline state
- Forces declaration of frag_color[0] with full RGBA when enabled
- Initializes alpha to 1.0 in shader epilogue if not explicitly written
Signed-off-by: Zephyron <zephyron@citron-emu.org>
This commit properly implements the ISBERD (Internal Stage Buffer Entry Read)
instruction that was previously stubbed. The implementation supports all
modes and shift types:
- Patch mode: Reads tessellation patch attributes using ir.GetPatch()
- Prim mode: Reads primitive attributes for geometry shaders
- Attr mode: Reads generic vertex attributes
- Default mode: Maintains backward compatibility
The implementation also supports:
- U16 and B32 shift modes for different data formats
- SKEW and O flags for advanced buffer addressing
- Proper data type conversions and bit casting
This fixes rendering issues in UE4 titles that rely on internal stage
buffer operations in tessellation and geometry shaders. The previous
stubbed implementation caused compatibility problems with modern
rendering pipelines.
Credit: Hayate Yoshida (吉田 疾風) <hayate@citron-emu.org> for discovering
the root cause and providing insight on the proper implementation approach.
Fixes: Internal stage buffer read operations in tessellation shaders
Resolves: UE4 title rendering issues related to ISBERD instruction
Signed-off-by: Zephyron <zephyron@citron-emu.org>
- Replace stubbed ISBERD implementation with proper internal stage buffer read
- Add support for different modes (Default, Patch, Prim, Attr)
- Implement shift operations (U16, B32) for buffer data processing
- Convert float results from GetAttributeIndexed to unsigned integers
- Handle buffer index from source register for stage buffer access
- Maintain proper error handling for unimplemented features (SKEW, O)
This completes the internal stage buffer entry read functionality for Maxwell shader translation.
Signed-off-by: Zephyron <zephyron@citron-emu.org>
Fix LTO linking issues on CachyOS with GCC 15.x
- Add LTO support to all core library targets (input_common, frontend_common,
network, shader_recompiler, web_service) that were missing LTO configuration
- Create citron_configure_lto() helper function for consistent LTO handling
- Implement CachyOS-specific detection via /etc/os-release and kernel version
- Apply conservative LTO flags (-flto=auto -ffat-lto-objects) only for
CachyOS + GCC 15+ to resolve linking errors with newer toolchains
- Other distributions continue using aggressive LTO settings for maximum performance
- Disable LTO on executable targets to prevent main function optimization issues
This resolves "undefined symbol" errors when building with -DCITRON_ENABLE_LTO=ON
on CachyOS while maintaining optimal LTO performance on other distributions.
Fixes linking errors including:
- Core::Frontend::EmuWindow symbols
- Core::System methods
- Settings::values
- Common logging functions
Tested on CachyOS with GCC 15.1.1 + LLD 20.1.8
Signed-off-by: Boss.sfc <boss.sfc@citron-emu.org>
Fix multiple compilation errors preventing successful build:
* Add const qualifier to custom fmt formatter functions across codebase
- Updated formatters in logging, shader recompiler, texture cache, and other modules
- Ensures compatibility with newer fmt library versions
* Add missing fmt/ranges.h includes for fmt::join usage
- Fixed fmt::join calls in Vulkan renderer, GDB stub, NFC service, and main window
- Resolves "no member named 'join' in namespace 'fmt'" errors
* Exclude unsupported platforms from Boost.Process usage in debugger
- Extended conditional compilation to avoid Boost.Process where unavailable
* Fix type casting issues in AOC service manager
- Resolved std::min type mismatch with explicit casting
