3e38a63ebb
D3D12_Shader_Debug_Zoo test for helper lane UAV writes are ignored
2025-10-04 07:16:49 +01:00
a0db9d5a1f
Comment about thread safety requirements in Spirv debugger ids container
2025-10-03 15:20:57 +01:00
d7f7be63a9
Spirv debugger small tidy up for consistency
...
Rename:
simulationFinished -> atomic_simulationFinished
Vulkan_Hack_ShaderDebugUsesJobSystemQueue -> Vulkan_Hack_ShaderDebugUsesJobSystemJobs
2025-10-03 06:24:49 +01:00
36f8d654dc
Add NULL aliasing barrier around buffer discard patterns
2025-10-02 14:17:20 +01:00
2018b5d790
Use the correct state for creating whole memory buffers on D3D12
2025-09-30 11:26:01 +01:00
a92e0d2669
Ensure capabilities are properly cached when only prepared once
2025-09-30 11:26:01 +01:00
30f5a4023d
Add support for VK_EXT_vertex_attribute_robustness
2025-09-30 19:10:06 +09:00
9711e0eb57
Add support for maintenance9 vertex attribute robustness
2025-09-30 19:10:06 +09:00
c03b2e2fb0
Remove unused methods from GpuMathOperation & GpuSampleGatherOperation
...
Clear() methods are not used
2025-09-29 17:09:04 +01:00
8864a1d81e
Improve time listing available demos
...
* Remove redundant repeated calls when preparing multiple tests
2025-09-29 16:40:15 +01:00
dadf4a14d5
Serialise D3D12 buffer initial contents less memory-intensivesly
2025-09-29 16:37:45 +01:00
41e9f3c426
Add a demo project with extremely high overlap/sparseness buffers
...
* For stressing memory overhead in buffer initial states
2025-09-29 16:37:45 +01:00
9952483f2b
Make D3D12_Groupshared test deterministic
2025-09-28 07:26:30 +01:00
1ccdf7f4f4
Op::ControlBarrier requires full lockstep when being simulated
...
Lockstep before executing the instruction
Lockstep after executing the instruction
2025-09-28 07:26:30 +01:00
357102fd3f
Make VK_Groupshared test deterministic
2025-09-28 07:26:30 +01:00
d6a49c4c18
Make D3D12_Workgroup_Zoo test determistic
2025-09-28 07:26:30 +01:00
9e63b3fb8c
Enable VK_Groupshared test
2025-09-28 07:26:30 +01:00
11a4104bde
DXIL Debugger support for thread local global memory
2025-09-28 07:26:30 +01:00
61cc31062e
Change D3D12_Shader_Debug_Zoo test to exercise thread local global mem
2025-09-28 07:26:15 +01:00
a5c3ee4b70
Tweak to DXIL debugger cbuffer handling
...
Initialise the ShaderVariable for the handle to 0xCC
Only memory copy the bytes available (do not assume the source buffer is a multiple of 16 bytes in size)
Fixes non-determinism problem (exposed by root constant buffer of size four bytes)
2025-09-25 17:35:35 +01:00
fa40e51e7d
Tweak d3d12_workgroup_zoo : set both root constants for both dispatches
2025-09-25 17:35:35 +01:00
8ecc23e78d
Return 0,0,0,0 when sampling NULL resources in Spirv Shader Debugger
2025-09-25 17:35:35 +01:00
4df53b41d3
Delete parents after base objects
...
* This fixes a potential crash deleting an image view after an image, by
ensuring a non-owning pointer to the image's resinfo remains valid until after
the image view is destroyed.
2025-09-25 16:17:46 +01:00
39c7dc1027
Fix python message handling reading out of bounds read
2025-09-24 15:47:14 +01:00
8203032ec3
Fix out of bounds writes during spirv debugging
...
Previously it would allow writes to width/height/depth
2025-09-24 15:46:06 +01:00
da03329b12
Improve handling of NULL resources in Spirv Shader Debugger
2025-09-24 15:35:01 +01:00
8898fa2e3e
Use liveID to find the descriptor buffer resource in shader debug
2025-09-24 15:34:51 +01:00
668fa75660
Change some asserts to be development only in Spirv Debugger
...
Asserts used during development of multi-threading and related to the consistency of expected state
2025-09-24 15:31:49 +01:00
56415992ac
Added CHECK_DEBUGGER_THREAD() to Spirv_Debug.cpp
...
Does nothing in RELEASE, replaces calls to
RDCASSERT(debugger.IsDeviceThread());
2025-09-24 15:29:01 +01:00
5222954723
Add a lock when simulating Atomic Memory instructions
2025-09-24 15:21:41 +01:00
234a78ecf2
Remove Vulkan_Hack_EnableGroupCaps
2025-09-24 15:18:14 +01:00
b536bbaf95
Spirv Debugger run jobs in a hot spin loop in a single JobQueue job
...
Each JobQueue Job runs SimulationJobHelper which is a while loop looking for threads to simulate
2025-09-24 15:14:55 +01:00
e9e081a55f
Spirv Debugger run multiple simulation steps per job
2025-09-24 15:08:43 +01:00
d78edcbbea
Added Debugger::GetBufferLength()
2025-09-24 15:02:32 +01:00
b9945d563e
Active thread runs scope changes on the device thread
...
Run EnterFunction(), EnterEntryPoint(), ProcessScopeChange() on device thread
2025-09-24 14:59:28 +01:00
7bdf8a8d88
Make pointer read/write methods return DeviceOpResult
...
ThreadState
DeviceOpResult ReadPointerValue(Id pointer, ShaderVariable &ret);
Debugger
DeviceOpResult ReadFromPointer(const ShaderVariable &ptr, ShaderVariable &ret) const;
DeviceOpResult GetPointerValue(const ShaderVariable &v, ShaderVariable &ret) const;
2025-09-24 14:52:42 +01:00
781984703b
Change WritePointerValue, WriteThroughPointer to return DeviceOpResult
...
Mark any pointer writes that require a callback to run on the device thread
2025-09-24 14:50:46 +01:00
d93be6f2b2
Multi-threaded support for SPIRV wrapper APIs ReadTexel, WriteTexel
2025-09-24 14:45:14 +01:00
8d9bb3f168
Add Spirv Debugger APIs to wrap API Wrapper APIs
...
The Debugger APIs run on the device thread if required,
FillInputValue(), ReadTexel(), WriteTexel()
2025-09-24 14:38:33 +01:00
35fdf44c20
Spirv debugger support for queue and run of Simulation on Device Thread
2025-09-24 14:33:01 +01:00
6e28ff1ca8
Convert Spirv Debugger to run Thread steps via the JobSystem
...
Enabled by default : config option "Vulkan_Debug_EnableShaderDebugMT"
2025-09-24 14:29:59 +01:00
94d7f4365d
Batch Spirv debugger MathOp and SampleGatherOp GPU operations
...
Thread Simulation queues MathOp, SampleGatherOp requests and then the Thread is paused.
The Debugger loop processes the MathOp queue, batches MathOps together into a single command list, Submit, Sync.
The Debugger loop processes the SampleGatherOp queue, batches GPU SampleGatherOps together into a single command list, Submit, Sync. Some SampleGatherOps are run on the CPU and do not use the GPU.
When the queued operations are completed, the results are written back into the Thread Simulation and the paused Threads are rerun.
2025-09-24 14:22:04 +01:00
b9d86315ff
Change Spirv Shader Debugger steps chunk size to 1,000,000
...
Helps to reduce the number of replays during shader debugging which helps when captures have non-deterministic replay
2025-09-24 14:18:43 +01:00
f12aa91ba2
Python helpers to generate and check a full trace from shader debug
2025-09-24 14:13:59 +01:00
e6f75e1530
Update Spirv debugger methods to be const methods
...
Add read/write locking for containers "usedNames" and "dynamicNames"
2025-09-24 14:09:43 +01:00
64b9e517e0
VK_Shader_Debug_Zoo test of overlapped GPU math and sampling operations
2025-09-24 14:01:47 +01:00
fc2c080a77
Remove unused cycles variable in VK_ShaderDebugZoo
2025-09-24 14:01:30 +01:00
d0d385ceef
Use const reference for ShaderBindIndex argument passing
2025-09-24 13:53:56 +01:00
3b74e038d2
Assert for SPIRV Subgroup Ops that the subgroup is converged
2025-09-24 13:49:51 +01:00
e5121962f3
Spirv Debugger control flow helper variable tweaks
...
Removed unused mergeBlock
Made control flow variables private with helper methods to get them
2025-09-24 13:46:00 +01:00
Jake Turner