* shader_recompiler: Replace buffer pulling with attribute divisor for instance step rates
* flatten_extended_userdata: Remove special step rate buffer handling
* Review comments
* spirv_emit_context: Name all instance rate attribs properly
* spirv: Merge ReadConstBuffer again
template function only has 1 user now
* attribute: Add missing attributes
* translate: Reimplement step rate instance id
* Resolve validation warnings
* shader_recompiler: Separate vertex inputs from LS stage, cleanup tess
* buffer_cache: Handle inline data to flexible memory
* control_flow: Fix single instruction scopes edge case
Fixes the following pattern
v_cmpx_gt_u32 cond
buffer_store_dword value
.LABEL:
Before
buffer[index] = value;
After
if (cond)
{
buffer[index] = value;
}
* vector_memory: Handle soffset when offen is false
When offen is not used we can substitute the offset argument with soffset and have it handled correctly
* scalar_alu: Handle sharp moves with S_MOV_B64
This fixes unable to track sharp errors when this pattern is used in a shader
* emulator: Add log
* video_core: Bump binary info search range and buffer num
* vector_alu: Improve handling of mbcnt append/consume patterns
The existing implementation was written to handle a single pattern of mbcnt before the DS_APPEND instruction
v_mbcnt_hi_u32_b32 vX, exec_hi, 0
v_mbcnt_lo_u32_b32 vX, exec_lo, vX
ds_append vY offset:4 gds
v_add_i32 vX, vcc, vY, vX
In this case however the DS_APPEND is before the mbcnt pattern
ds_append vX gds
v_mbcnt_hi_u32_b32 vY, exec_hi, vX
v_mbcnt_lo_u32_b32 vZ, exec_lo, vY
The mbcnt instructions are always in pairs of hi/lo and in general are quite flexible. But they assume the subgroup size is 64 so they are not recompiled literally. Together with DS_APPEND they are used to derive a unique per thread index in a buffer (different from using thread_id as order could be random). DS_APPEND instruction works on per subgroup level, by adding number of active threads of subgroup to the GDS counter, essentially giving a multiple-of-64 base index to all threads. Then each thread executes the mbcnt pair which returns the number of active threads with id less than the itself and adds it with the base.
The recompiler translates DS_APPEND into an atomic increment of a storage buffer counter, which already gives the desired unique index, so this pattern is a no-op. On main it was set to zero as per the first pattern to avoid altering the DS_APPEND result. The new handling passes through the initial value of the pattern instead, which has the same effect but works on either case.
* vk_rasterizer: Always sync DMA buffers
* Added V_CMP_EQ_U64 shader opcode support and added 64-bit relational operators (<,>,<=,>=)
* Fixed clang-format crying because I typed xargs clang-format instead of xargs clang-format-19
* Replaced V_CMP_EQ_U64 code to match V_CMP_U32 to test
* Updated V_CMP_U64 for future addons
* shader_recompiler: Simplify dma types
Only U32 is needed for S_LOAD_DWORD
* shader_recompiler: Perform address shift on IR level
Buffer instructions now expect address in the data unit they work on. Doing the shift on IR level will allow us to optimize some operations away on common case
* shader_recompiler: Optimize common buffer access pattern
* emit_spirv: Use 32-bit integer ops for fault buffer
Not many GPUs have 8-bit bitwise or operations so that would probably require some overhead to emulate from the driver
* resource_tracking_pass: Fix texel buffer shift
* Import memory
* 64K pages and fix memory mapping
* Queue coverage
* Buffer syncing, faulted readback adn BDA in Buffer
* Base DMA implementation
* Preparations for implementing SPV DMA access
* Base impl (pending 16K pages and getbuffersize)
* 16K pages and stack overflow fix
* clang-format
* clang-format but for real this time
* Try to fix macOS build
* Correct decltype
* Add testing log
* Fix stride and patch phi node blocks
* No need to check if it is a deleted buffer
* Clang format once more
* Offset in bytes
* Removed host buffers (may do it in another PR)
Also some random barrier fixes
* Add IR dumping from my read-const branch
* clang-format
* Correct size insteed of end
* Fix incorrect assert
* Possible fix for NieR deadlock
* Copy to avoid deadlock
* Use 2 mutexes insteed of copy
* Attempt to range sync error
* Revert "Attempt to range sync error"
This reverts commit dd287b48682b50f215680bb0956e39c2809bf3fe.
* Fix size truncated when syncing range
And memory barrier
* Some fixes (and async testing (doesn't work))
* Use compute to parse fault buffer
* Process faults on submit
* Only sync in the first time we see a readconst
Thsi is partialy wrong. We need to save the state into the submission context itself, not the rasterizer since we can yield and process another sumission (if im not understanding wrong).
* Use spec const and 32 bit atomic
* 32 bit counter
* Fix store_index
* Better sync (WIP, breaks PR now)
* Fixes for better sync
* Better sync
* Remove memory coveragte logic
* Point sirit to upstream
* Less waiting and barriers
* Correctly checkout moltenvk
* Bring back applying pending operations in wait
* Sync the whole buffer insteed of only the range
* Implement recursive shared/scoped locks
* Iterators
* Faster syncing with ranges
* Some alignment fixes
* fixed clang format
* Fix clang-format again
* Port page_manager from readbacks-poc
* clang-format
* Defer memory protect
* Remove RENDERER_TRACE
* Experiment: only sync on first readconst
* Added profiling (will be removed)
* Don't sync entire buffers
* Added logging for testing
* Updated temporary workaround to use 4k pages
* clang.-format
* Cleanup part 1
* Make ReadConst a SPIR-V function
---------
Co-authored-by: georgemoralis <giorgosmrls@gmail.com>
* shader_recompiler: Add lowering pass for when 64-bit float is unsupported.
* shader_recompiler: Fix PackDouble2x32/UnpackDouble2x32 type.
* shader_recompiler: Remove extra bit cast implementations.
* Implementing DS_SUB_U32, DS_INC_U32, DS_DEC_U32, DS_WRITE_SRC2_B32, DS_WRITE_SRC2_B64.
* Added ir instructions for new opcodes.
Removing Write implementations.
Maping operation S_BFE_I32 as it was added in translate but wasnt pointing to anything.
* Suggestions
* shader_recompiler: Implement AMD buffer bounds checking behavior.
* shader_recompiler: Use SRT flatbuf for bounds check size.
* shader_recompiler: Fix buffer atomic bounds check.
* buffer_cache: Prevent false image-to-buffer sync.
Lowering vertex fetch to formatted buffer surfaced an issue where a CPU modified range may be overwritten with stale GPU modified image data.
* Address review comments.
* shader_recompiler: Move shared mem lowering into emitter
* IR can be quite verbose during first stages of translation, before ssa and constant prop passes have run that drastically simplify it. This lowering can also be done during emission so why not do it then to save some compilation time
* runtime_info: Pack PsColorBuffer into 8 bytes
* Drops the size of the total structure by half from 396 to 204 bytes. Also should make comparison of the array a bit faster, since its a hot path done every draw
* emit_spirv_context: Add infrastructure for buffer aliases
* Splits out the buffer creation function so it can be reused when defining multiple type aliases
* shader_recompiler: Merge srt_flatbuf into buffers list
* Its no longer a special case, yay
* shader_recompiler: Complete buffer aliasing support
* Add a bunch more types into buffers, such as F32 for float reads/writes and 8/16 bit integer types for formatted buffers
* shader_recompiler: Remove existing shared memory emulation
* The current impl relies on backend side implementaton and hooking into every shared memory access. It also doesnt handle atomics. Will be replaced by an IR pass that solves these issues
* shader_recompiler: Reintroduce shared memory on ssbo emulation
* Now it is performed with an IR pass, and combined with the previous commit cleanup, is fully transparent from the backend, other than requiring workgroup_index be provided as an attribute (computing this on every shared memory access is gonna be too verbose
* clang format
* buffer_cache: Reduce buffer sizes
* vk_rasterizer: Cleanup resource binding code
* Reduce noise in the functions, also remove some arguments which are class members
* Fix gcc
