Search Issue Tracker

Postponed means that the issue was either a feature request or something that requires major refactoring on our side. Since that makes the issue not actionable in the close future we choose to close it as Postponed and add it on our internal roadmaps and technical debt pages instead.

Postponed

Votes

0

Found in

Issue ID

1296454

Regression

No

[GPU PLM] 100% repro crash when baking EnterTheRoom with 4K maps when using system allocator

Global Illumination

-

How to reproduce:
1) Pull https://ono.unity3d.com/test-data/gi-tests/EnterTheRoom revision a9e374d93cbc
2) Launch Unity 2021.1.0a9 (built from trunk rev 18321e85013a) using the -systemallocator command line option
2) Open the scene `4 years ago`
3) Use GPU lightmapper and auto denoiser.
4) Increase lightmap size to 4K and set lightmap resolution to 300.

Observed result:
Editor crashes with the following callstack every time, indicating an allocator problem:

========== OUTPUTTING STACK TRACE ==================
0x00007FF8D0E81323 (ntdll) RtlGetCurrentServiceSessionId
0x00007FF8D0E806E1 (ntdll) RtlFreeHeap
0x00007FF75B390344 (Unity) [minkernel\crts\ucrt\src\appcrt\heap\free_base.cpp:105] _free_base
0x00007FF75B376F5E (Unity) [minkernel\crts\ucrt\src\appcrt\heap\align.cpp:545] _aligned_free
0x00007FF756F138DC (Unity) [S:\code\2021.1\Runtime\Allocator\MemoryManager.cpp:1964] MemoryManager::Deallocate
0x00007FF756F135E7 (Unity) [S:\code\2021.1\Runtime\Allocator\MemoryManager.cpp:1896] MemoryManager::Deallocate
0x00007FF756F1336D (Unity) [S:\code\2021.1\Runtime\Allocator\MemoryManager.cpp:1870] MemoryManager::Deallocate
0x00007FF756F1336D (Unity) [S:\code\2021.1\Runtime\Allocator\MemoryManager.cpp:1870] MemoryManager::Deallocate
0x00007FF756F2553A (Unity) [S:\code\2021.1\Runtime\Allocator\MemoryManager.cpp:391] free_alloc_internal
0x00007FF7564FA911 (Unity) [S:\code\2021.1\Runtime\Utilities\dynamic_array.h:185] dynamic_array<unsigned char,0>::~dynamic_array<unsigned char,0>
Using memoryadresses from more than 16GB of memory
Using memoryadresses from more than 16GB of memory
Received signal SIGSEGV
0x00007FF75A047AF7 (Unity) [S:\code\2021.1\Editor\Src\GI\Progressive\RadeonRays\Jobs\RRJobCompositeLightmap.cpp:592] RadeonRaysJobCompositeLightmap::RunDenoiser
0x00007FF75A03F33E (Unity) [S:\code\2021.1\Editor\Src\GI\Progressive\RadeonRays\Jobs\RRJobCompositeLightmap.cpp:845] RadeonRaysJobCompositeLightmap::CompositeLightmap
0x00007FF759FFCBF5 (Unity) [S:\code\2021.1\Editor\Src\GI\Progressive\PVRJobs\PVRJobCompositeLightmap.cpp:2099] BakeJobCompositeLightmap::Execute
0x00007FF7589FCD94 (Unity) [S:\code\2021.1\Editor\Src\JobManager\JobManager.cpp:597] JobManager::UpdateFunction
0x00007FF757614740 (Unity) [S:\code\2021.1\Runtime\Jobs\Internal\JobQueue.cpp:625] JobQueue::Exec
0x00007FF75761C621 (Unity) [S:\code\2021.1\Runtime\Jobs\Internal\JobQueue.cpp:734] JobQueue::Steal
0x00007FF7576150F0 (Unity) [S:\code\2021.1\Runtime\Jobs\Internal\JobQueue.cpp:855] JobQueue::ExecuteJobFromQueue
0x00007FF757616D9B (Unity) [S:\code\2021.1\Runtime\Jobs\Internal\JobQueue.cpp:949] JobQueue::ProcessJobs
0x00007FF75761D97F (Unity) [S:\code\2021.1\Runtime\Jobs\Internal\JobQueue.cpp:1033] JobQueue::WorkLoop
0x00007FF757C5C385 (Unity) [S:\code\2021.1\Runtime\Threads\Thread.cpp:81] Thread::RunThreadWrapper
0x00007FF8CFCD7C24 (KERNEL32) BaseThreadInitThunk
0x00007FF8D0EAD4D1 (ntdll) RtlUserThreadStart
========== END OF STACKTRACE ===========

Please note that the GPU used is an Nvidia GeForce 2080Ti, AMD will not use optix and therefore the bug might not reproduce.

  1. Response avatar

    Resolution Note:

    As the crash is triggered when we are low on GPU memory, a potential fix is to denoise the 4K map in smaller tiles. The tiled denoising task is tracked here https://jira.unity3d.com/browse/GFXGI-411

Add comment

Log in to post comment

All about bugs

View bugs we have successfully reproduced, and vote for the bugs you want to see fixed most urgently.