Cycles GPU Compute baking often fails #59286

New Issue

Dan Sinni · 2018-12-13T04:45:14+01:00

Dan Sinni commented

2018-12-13 04:45:14 +01:00

System Information

Operating system: Windows 10 Home 64
Graphics card: NVIDIA GeForce GTX 860M

Blender Version

Broken: 2.80 Beta (all)

Worked: 2.79b

Short description of error

When using CUDA and Cycles GPU Compute to bake textures, it usually fails halfway through the process resulting in half-baked textures with horizontal lines through them, like the one displayed below.

In each subsequent attempt, the bake fails immediately resulting in errors like below. It won't bake again unless I use CPU or restart Blender, but I can't recall it ever baking fully with GPU Compute even after restarting.

Exact steps for others to reproduce the error

Select CUDA from Edit > Preferences > System > General > Cycles Compute Device
Check any or all devices

{F5937164}

Use following settings:

{F5937188}
{F5937195}

Bake

**System Information** Operating system: Windows 10 Home 64 Graphics card: NVIDIA GeForce GTX 860M **Blender Version** Broken: 2.80 Beta (all) ![image.png](https://archive.blender.org/developer/F5937069/image.png) Worked: 2.79b **Short description of error** When using `CUDA` and `Cycles` `GPU Compute` to bake textures, it usually fails halfway through the process resulting in half-baked textures with horizontal lines through them, like the one displayed below. ![image.png](https://archive.blender.org/developer/F5937235/image.png) In each subsequent attempt, the bake fails immediately resulting in errors like below. It won't bake again unless I use `CPU` or restart Blender, but I can't recall it ever baking fully with `GPU Compute` even after restarting. ![image.png](https://archive.blender.org/developer/F5937116/image.png) **Exact steps for others to reproduce the error** 1. Select `CUDA` from `Edit > Preferences > System > General > Cycles Compute Device` 2. Check any or all devices ``` {F5937164} ``` 3. Use following settings: ``` {F5937188} {F5937195} ``` 4. Bake

Dan Sinni commented

2018-12-13 04:45:14 +01:00

Added subscriber: @DanSinni

blender-admin commented

2018-12-13 04:45:14 +01:00

#62233 was marked as duplicate of this issue

blender-admin commented

2018-12-13 04:45:14 +01:00

#61882 was marked as duplicate of this issue

blender-admin commented

2018-12-13 04:45:14 +01:00

#60630 was marked as duplicate of this issue

blender-admin commented

2018-12-13 04:45:14 +01:00

#61508 was marked as duplicate of this issue

blender-admin commented

2018-12-13 04:45:14 +01:00

#60121 was marked as duplicate of this issue

blender-admin commented

2018-12-13 04:45:14 +01:00

#59238 was marked as duplicate of this issue

ronan ducluzeau commented

2018-12-14 16:24:21 +01:00

Added subscriber: @zeauro

ronan ducluzeau commented

2018-12-14 16:24:21 +01:00

I can't reproduce with 21523b5 under ubuntu 16.04 with a titan black. Can you upload a .blend file with the issue ?

I can't reproduce with 21523b5 under ubuntu 16.04 with a titan black. Can you upload a .blend file with the issue ?

Dan Sinni commented

2018-12-16 06:50:55 +01:00

Sure. Try this one. I also packed the resulting texture. If I try to bake it again, it fails and throws an error.

gpu_bake_bug.blend

Sure. Try this one. I also packed the resulting texture. If I try to bake it again, it fails and throws an error. [gpu_bake_bug.blend](https://archive.blender.org/developer/F5968656/gpu_bake_bug.blend)

ronan ducluzeau commented

2018-12-16 13:14:44 +01:00

No error. Correct baking, here.
It reminds to me problems at first attempts of hybrid rendering.

Does it work if you only use Graphic Card as CUDA device ?

No error. Correct baking, here. It reminds to me problems at first attempts of hybrid rendering. Does it work if you only use Graphic Card as CUDA device ?

Dan Sinni commented

2018-12-16 19:56:35 +01:00

Just gave that a try. No luck. Actually, it failed immediately.

This appears in the taskbar:

Just gave that a try. No luck. Actually, it failed immediately. This appears in the taskbar: ![image.png](https://archive.blender.org/developer/F5973596/image.png)

Brecht Van Lommel commented

2019-01-03 16:50:43 +01:00

Added subscribers: @JamesOvnik, @alaslipknot

Brecht Van Lommel commented

2019-01-03 16:50:48 +01:00

Added subscriber: @ostapblender

Brecht Van Lommel self-assigned this 2019-01-03 16:51:21 +01:00

Brent Lio commented

2019-01-10 19:38:10 +01:00

Added subscriber: @remotecrab131

Brent Lio commented

2019-01-10 19:38:10 +01:00

Win 10 Pro, gtx 1080

First bake give me an image like the one OP demonstrated, with scan-line like effect, bad bake.

Second try give me nothing. And my viewport cycle straight out vanishes, give me only the GL layer where I can see the light objects and cameras, selection outline, etc.

It also throws an error: "CUDA error at cuCtxCreate: Illegal address", but then reports "Baking map saved to internal image, save it..."

Win 10 Pro, gtx 1080 First bake give me an image like the one OP demonstrated, with scan-line like effect, bad bake. Second try give me nothing. And my viewport cycle straight out vanishes, give me only the GL layer where I can see the light objects and cameras, selection outline, etc. It also throws an error: "CUDA error at cuCtxCreate: Illegal address", but then reports "Baking map saved to internal image, save it..."

dev369 commented

2019-01-21 13:09:20 +01:00

Added subscriber: @dev369

dev369 commented

2019-01-21 13:09:20 +01:00

Hell on not sure if this its the same issue but looks similar - baking doesnt work gtx 1070

here is the entire error log

https://pastebin.com/JLepKyff

Hell on not sure if this its the same issue but looks similar - baking doesnt work gtx 1070 here is the entire error log https://pastebin.com/JLepKyff

Win commented

2019-01-26 00:19:25 +01:00

Added subscriber: @Win7yes

artem ivanov commented

2019-01-26 13:36:43 +01:00

Added subscriber: @ixd

Roy van de Water commented

2019-02-05 14:19:52 +01:00

Added subscriber: @royvandewater

Roy van de Water commented

2019-02-05 14:24:59 +01:00

Reproduced on a GTX 1070 in blender-2.80.0-git.445433a6913f-windows64 using the default cube scene and attempting to do a combined bake onto a 1024x1024 image with all default render settings.

https://pastebin.com/DgAp8hxg

Reproduced on a GTX 1070 in blender-2.80.0-git.445433a6913f-windows64 using the default cube scene and attempting to do a combined bake onto a 1024x1024 image with all default render settings. https://pastebin.com/DgAp8hxg

John commented

2019-02-06 01:43:59 +01:00

Added subscriber: @John514

RK commented

2019-02-07 11:29:46 +01:00

Added subscriber: @Minimalistiker

Adrian commented

2019-02-12 14:46:39 +01:00

Added subscriber: @Fjury-4

Adrian commented

2019-02-12 14:46:39 +01:00

Reproduced this issue in every comitted Blender file, latest blender-2.80.0-git.d3870471edd7-windows64, using my own Shaderstatue for Shader testing. All bakings fail before 10% got actually baked.

Message Bake stored internally please safe and trying to bake again results in the orange CUDA message " CUDA error: illegal adress".

GPU: 1080 GTX latest graphics driver
CPU: i5 4460
Windows 10 latest build{F6590069}

Reproduced this issue in every comitted Blender file, latest blender-2.80.0-git.d3870471edd7-windows64, using my own Shaderstatue for Shader testing. All bakings fail before 10% got actually baked. Message Bake stored internally please safe and trying to bake again results in the orange CUDA message " CUDA error: illegal adress". GPU: 1080 GTX latest graphics driver CPU: i5 4460 Windows 10 latest build{[F6590069](https://archive.blender.org/developer/F6590069/Shader_Mesh5.blend)}

dev369 commented

2019-02-12 15:09:54 +01:00

From my research this issue has been present for more then a year now. Its a pretty big issue, id say. So far devss kept complete radio silence on this issue, or i havent been able to find anywhere where it was anything said about it. Could devs say anything about it like, is it getting fixed at all and when? Why has it been neglected for so long, is this that hard to fix? Considering it has been around for such a long time.

Brecht Van Lommel commented

2019-02-12 15:11:46 +01:00

So far we've not found a way to reproduce this yet, tested on multiple computers with different graphics cards.

It will get fixed, but it's one of 1000 other bugs in the tracker that we are working through.

So far we've not found a way to reproduce this yet, tested on multiple computers with different graphics cards. It will get fixed, but it's one of 1000 other bugs in the tracker that we are working through.

Adrian commented

2019-02-12 17:57:52 +01:00

Are there any ways to log those things in blender so u can get further details from users?

Brecht Van Lommel commented

2019-02-13 13:58:41 +01:00

Added subscriber: @Rawalanche

Brecht Van Lommel commented

2019-02-13 18:31:23 +01:00

Added subscribers: @rileybrown704, @WilliamReynish

Artem commented

2019-02-28 07:47:16 +01:00

Added subscriber: @bulbfish

Ricardo Diaz commented

2019-03-05 06:08:10 +01:00

Added subscriber: @poyodiaz

Ricardo Diaz commented

2019-03-05 06:08:10 +01:00

hi everyone , this bug has not been fixed right? is there a possibility in a close future? seems to work ok in ubuntu...is the only reason I need to reboot my pc... to bake a texture,(not trying to offend anyone I do really appreciate your hard work guys) thanks for all the great effort ! :)

Jorge commented

2019-03-05 12:57:50 +01:00

Added subscriber: @JAlcaraz

Brecht Van Lommel commented

2019-03-08 16:53:14 +01:00

Still can't confirm myself, but marking as high priority.

Brecht Van Lommel commented

2019-03-08 16:53:29 +01:00

Added subscribers: @szymi, @brecht, @JacquesLucke

Bohdan Lvov commented

2019-03-08 17:11:05 +01:00

I tred it on 0ba143a1d6 - same thing, and it's also impossible to start baking again due to CUDA error, as I said in #60121 and it works only after restart of a Blender
BUT! New thing appeared - it looks different with different tile size. Looks like it can render only top half of each tile

Target resolution: 1024x1024

Tile 64:

Tile 256:

Tile 512:

Tile 1024 and upwards:

CubeBake.blend1

I tred it on 0ba143a1d675 - same thing, and it's also impossible to start baking again due to CUDA error, as I said in #60121 and it works only after restart of a Blender BUT! New thing appeared - it looks different with different tile size. Looks like it can render only top half of each tile Target resolution: 1024x1024 Tile 64: ![image.png](https://archive.blender.org/developer/F6785247/image.png) Tile 256: ![image.png](https://archive.blender.org/developer/F6785263/image.png) Tile 512: ![image.png](https://archive.blender.org/developer/F6785244/image.png) Tile 1024 and upwards: ![image.png](https://archive.blender.org/developer/F6785258/image.png) [CubeBake.blend1](https://archive.blender.org/developer/F6785266/CubeBake.blend1)

Brecht Van Lommel commented

2019-03-14 13:32:19 +01:00

Added subscribers: @Ureh, @ZedDB

Brecht Van Lommel commented

2019-03-15 13:11:33 +01:00

I was able to redo this on a GTX 960 now. The exact cause is not clear to me, but it seems that CUDA 10.1 works correctly. Possibly it's the same compiler bug(s) we've reported to NVIDIA before and that they fixed in 10.0 and 10.1.

blender-admin commented

2019-03-15 14:18:22 +01:00

This issue was referenced by blender/cycles@bbd8cb7582

This issue was referenced by blender/cycles@bbd8cb7582e028c220c690278df6f3e13f606b12

blender-admin commented

2019-03-15 14:18:22 +01:00

This issue was referenced by 65d95879f7

This issue was referenced by 65d95879f73951861a90efe1c3ac5d1d03530fd6

Brecht Van Lommel commented

2019-03-15 16:54:56 +01:00

Changed status from 'Open' to: 'Resolved'

Brecht Van Lommel closed this issue

2019-03-15 16:54:56 +01:00

artem ivanov commented

2019-03-15 21:24:49 +01:00

@brecht A bit too late, but I did some research on this (just about 2 hours before it got fixed lol) trying to compile bake kernel with various CUDA versions (9.1, 10.0, 10.1) and optimizations. Here are the results:

OS: Win 10, GPU: GTX 1070

Legend: OK = bakes well, ERROR = cuda errors while baking

Compiler	Assembler	Optimization	Format	MSVC Version	Result
NVCC 9.1	-	O3	cubin	14.11	OK
NVCC 9.1	-	None	cubin	14.11	OK
NVCC 9.1	-	O3	ptx + JIT	14.11	OK
NVCC 9.1	-	None	ptx + JIT	14.11	OK
NVCC 10.0	-	O3	cubin	14.11	OK
NVCC 10.0	-	None	cubin	14.11	OK
NVCC 10.0	-	O3	ptx + JIT	14.11	OK
NVCC 10.0	-	None	ptx + JIT	14.11	OK
NVCC 10.0	-	O3	cubin	14.16	OK
NVCC 10.0	-	None	cubin	14.16	OK
NVCC 10.0	-	O3	ptx + JIT	14.16	OK
NVCC 10.0	-	None	ptx + JIT	14.16	OK
NVCC 10.1	-	O3	cubin	14.11	OK
NVCC 10.1	-	None	cubin	14.11	OK
NVCC 10.1	-	O3	ptx + JIT	14.11	OK
NVCC 10.1	-	None	ptx + JIT	14.11	OK
NVCC 10.1	-	O3	cubin	14.16	OK
NVCC 10.1	-	None	cubin	14.16	OK
NVCC 10.1	-	O3	ptx + JIT	14.16	OK
NVCC 10.1	-	None	ptx + JIT	14.16	OK
NVCC 9.1	ptxas 9.1	O3	cubin	14.11	OK
NVCC 9.1	ptxas 9.1	None	cubin	14.11	OK
NVCC 9.1	ptxas 10.0	O3	cubin	14.11	OK
NVCC 9.1	ptxas 10.0	None	cubin	14.11	OK
NVCC 9.1	ptxas 10.1	O3	cubin	14.11	OK
NVCC 9.1	ptxas 10.1	None	cubin	14.11	OK
NVCC 10.0	ptxas 10.0	O3	cubin	14.11	OK
NVCC 10.0	ptxas 10.0	None	cubin	14.11	OK
NVCC 10.0	ptxas 10.0	O3	cubin	14.16	OK
NVCC 10.0	ptxas 10.0	None	cubin	14.16	OK
NVCC 10.0	ptxas 10.1	O3	cubin	14.11	OK
NVCC 10.0	ptxas 10.1	None	cubin	14.11	OK
NVCC 10.1	ptxas 10.1	O3	cubin	14.11	OK
NVCC 10.1	ptxas 10.1	None	cubin	14.11	OK
NVCC 10.1	ptxas 10.1	O3	cubin	14.16	OK
NVCC 10.1	ptxas 10.1	None	cubin	14.16	OK
NVRTC 9.1	ptxas 9.1	O3	cubin	14.11	ERROR
NVRTC 9.1	ptxas 9.1	None	cubin	14.11	OK
NVRTC 9.1	-	O3	ptx + JIT	14.11	OK
NVRTC 9.1	ptxas 10.0	O3	cubin	14.11	Weird pixels, but not crashed
NVRTC 9.1	ptxas 10.0	None	cubin	14.11	OK
NVRTC 10.0	ptxas 10.0	O3	cubin	14.11	Weird pixels, but not crashed
NVRTC 10.0	ptxas 10.0	None	cubin	14.11	OK
NVRTC 10.0	-	O3	ptx + JIT	14.11	OK
NVRTC 10.0	-	None	ptx + JIT	14.11	OK
NVRTC 9.1	ptxas 10.1	O3	cubin	14.11	OK
NVRTC 9.1	ptxas 10.1	None	cubin	14.11	OK
NVRTC 10.0	ptxas 10.1	O3	cubin	14.11	OK
NVRTC 10.0	ptxas 10.1	None	cubin	14.11	OK
NVRTC 10.1	ptxas 10.1	O3	cubin	14.11	OK
NVRTC 10.1	ptxas 10.1	None	cubin	14.11	OK
ptx + JIT means loading ptx assembly into blender (instead of .cubin) and letting nvidia driver compile it. (Requires no change to blender source code as driver can handle ptx as well as cubin)

"Weird dots" image from NVRTC 9.1 & ptxas 10.0:

And cycles_cubin_cc (which was used to compile cuda kernels shipped with buildbot) did NVRTC 9.1 & ptxas 9.1 with optimization. So it seems to me that nvrtc 9.1 and ptxas 9.1 and dont play nice together on some gpu's (but nvcc 9.1 & ptxas 9.1 works good). Latest NVIDIA driver jit compiles nvrtc 9.1 ptx kernel just fine.

Running cuda_memcheck blender.exe shows:

========= Invalid __global__ read of size 4
=========     at 0x000bab78 in <>/blender/intern/cycles/kernel/../kernel/kernel_path_state.h:250:kernel_cuda_bake
=========     by thread (10,0,0) in block (14,0,0)
=========     Address 0x3e80000000000022 is misaligned
=========     Device Frame:<>/blender/intern/cycles/kernel/../kernel/kernel_bake.h:121:kernel_cuda_bake (kernel_cuda_bake : 0x2cbd8)
=========     Saved host backtrace up to driver entry point at kernel launch time
=========     Host Frame:C:\WINDOWS\SYSTEM32\nvcuda.dll (cuGetExportTable + 0x1b9c47) [0x1c82c5]
=========     Host Frame:<>\build_windows_Release_x64_vc15_Debug\bin\Debug\blender.exe (ccl::CUDADevice::shader + 0x8a3) [0x29a2c83]
=========     Host Frame:<>\build_windows_Release_x64_vc15_Debug\bin\Debug\blender.exe (ccl::CUDADevice::thread_run + 0x356) [0x29a5796]
...

And after further investigation it shows something like a stack corruption on a random thread is happening in

kernel_bake_evaluate ->
    compute_light_pass (create variable "PathState state;") ->
        kernel_path_indirect (passing "state" by pointer) <corruption here> -> ... -> path_state_ao_bounce (kernel_path_state.h:250) ->
                     "access to pointer to stack allocated (and not shared) var "state", pointer points to garbage"

And all subsequent calls to CUDA api return an error. (cuCtxSynchronize, etc)
It seems to me it's an optimization codegen bug with nvrtc 9.1 (and possible in 10.0).

CUDA 10.1 resolves these issues (and we can use the latest msvc with it!).

Tested on this file (default cube + texture):
bake_bug_2.blend

@brecht A bit too late, but I did some research on this (just about 2 hours before it got fixed lol) trying to compile bake kernel with various CUDA versions (9.1, 10.0, 10.1) and optimizations. Here are the results: OS: Win 10, GPU: GTX 1070 *Legend: OK = bakes well, ERROR = cuda errors while baking* | Compiler | Assembler | Optimization | Format | MSVC Version | Result | |---|---|---|---|---|---| |NVCC 9.1| - | O3 |cubin|14.11|OK| |NVCC 9.1| - | None |cubin|14.11|OK| |NVCC 9.1| - | O3 |ptx + JIT|14.11|OK| |NVCC 9.1| - | None |ptx + JIT|14.11|OK| |NVCC 10.0|-|O3|cubin|14.11|OK| |NVCC 10.0|-|None|cubin|14.11|OK| |NVCC 10.0|-|O3|ptx + JIT|14.11|OK| |NVCC 10.0|-|None|ptx + JIT|14.11|OK| |NVCC 10.0|-|O3|cubin|14.16|OK| |NVCC 10.0|-|None|cubin|14.16|OK| |NVCC 10.0|-|O3|ptx + JIT|14.16|OK| |NVCC 10.0|-|None|ptx + JIT|14.16|OK| |NVCC 10.1|-|O3|cubin|14.11|OK| |NVCC 10.1|-|None|cubin|14.11|OK| |NVCC 10.1|-|O3|ptx + JIT|14.11|OK| |NVCC 10.1|-|None|ptx + JIT|14.11|OK| |NVCC 10.1|-|O3|cubin|14.16|OK| |NVCC 10.1|-|None|cubin|14.16|OK| |NVCC 10.1|-|O3|ptx + JIT|14.16|OK| |NVCC 10.1|-|None|ptx + JIT|14.16|OK| |NVCC 9.1| ptxas 9.1|O3|cubin|14.11|OK| |NVCC 9.1| ptxas 9.1|None|cubin|14.11|OK| |NVCC 9.1| ptxas 10.0|O3|cubin|14.11|OK| |NVCC 9.1| ptxas 10.0|None|cubin|14.11|OK| |NVCC 9.1| ptxas 10.1|O3|cubin|14.11|OK| |NVCC 9.1| ptxas 10.1|None|cubin|14.11|OK| |NVCC 10.0| ptxas 10.0|O3|cubin|14.11|OK| |NVCC 10.0| ptxas 10.0|None|cubin|14.11|OK| |NVCC 10.0| ptxas 10.0|O3|cubin|14.16|OK| |NVCC 10.0| ptxas 10.0|None|cubin|14.16|OK| |NVCC 10.0| ptxas 10.1|O3|cubin|14.11|OK| |NVCC 10.0| ptxas 10.1|None|cubin|14.11|OK| |NVCC 10.1| ptxas 10.1|O3|cubin|14.11|OK| |NVCC 10.1| ptxas 10.1|None|cubin|14.11|OK| |NVCC 10.1| ptxas 10.1|O3|cubin|14.16|OK| |NVCC 10.1| ptxas 10.1|None|cubin|14.16|OK| |NVRTC 9.1| ptxas 9.1|O3|cubin|14.11|**ERROR**| |NVRTC 9.1| ptxas 9.1|None|cubin|14.11|OK| |NVRTC 9.1| -|O3|ptx + JIT|14.11|OK| |NVRTC 9.1| ptxas 10.0|O3|cubin|14.11|**Weird pixels, but not crashed**| |NVRTC 9.1| ptxas 10.0|None|cubin|14.11|OK| |NVRTC 10.0| ptxas 10.0|O3|cubin|14.11|**Weird pixels, but not crashed**| |NVRTC 10.0| ptxas 10.0|None|cubin|14.11|OK| |NVRTC 10.0|-|O3|ptx + JIT|14.11|OK| |NVRTC 10.0|-|None|ptx + JIT|14.11|OK| |NVRTC 9.1| ptxas 10.1|O3|cubin|14.11|OK| |NVRTC 9.1| ptxas 10.1|None|cubin|14.11|OK| |NVRTC 10.0| ptxas 10.1|O3|cubin|14.11|OK| |NVRTC 10.0| ptxas 10.1|None|cubin|14.11|OK| |NVRTC 10.1| ptxas 10.1|O3|cubin|14.11|OK| |NVRTC 10.1| ptxas 10.1|None|cubin|14.11|OK| ***ptx + JIT** means loading ptx assembly into blender (instead of .cubin) and letting nvidia driver compile it. (Requires no change to blender source code as driver can handle ptx as well as cubin)* "Weird dots" image from NVRTC 9.1 & ptxas 10.0: ![combined.png](https://archive.blender.org/developer/F6831122/combined.png) And **cycles_cubin_cc** (which was used to compile cuda kernels shipped with buildbot) did `NVRTC 9.1 & ptxas 9.1 with optimization`. So it seems to me that `nvrtc 9.1` and `ptxas 9.1` and dont play nice together on some gpu's (but `nvcc 9.1 & ptxas 9.1` works good). Latest NVIDIA driver jit compiles `nvrtc 9.1 ptx` kernel just fine. Running `cuda_memcheck blender.exe` shows: ``` ========= Invalid __global__ read of size 4 ========= at 0x000bab78 in <>/blender/intern/cycles/kernel/../kernel/kernel_path_state.h:250:kernel_cuda_bake ========= by thread (10,0,0) in block (14,0,0) ========= Address 0x3e80000000000022 is misaligned ========= Device Frame:<>/blender/intern/cycles/kernel/../kernel/kernel_bake.h:121:kernel_cuda_bake (kernel_cuda_bake : 0x2cbd8) ========= Saved host backtrace up to driver entry point at kernel launch time ========= Host Frame:C:\WINDOWS\SYSTEM32\nvcuda.dll (cuGetExportTable + 0x1b9c47) [0x1c82c5] ========= Host Frame:<>\build_windows_Release_x64_vc15_Debug\bin\Debug\blender.exe (ccl::CUDADevice::shader + 0x8a3) [0x29a2c83] ========= Host Frame:<>\build_windows_Release_x64_vc15_Debug\bin\Debug\blender.exe (ccl::CUDADevice::thread_run + 0x356) [0x29a5796] ... ``` And after further investigation it shows something like a stack corruption on a random thread is happening in ``` kernel_bake_evaluate -> compute_light_pass (create variable "PathState state;") -> kernel_path_indirect (passing "state" by pointer) <corruption here> -> ... -> path_state_ao_bounce (kernel_path_state.h:250) -> "access to pointer to stack allocated (and not shared) var "state", pointer points to garbage" ``` And all subsequent calls to CUDA api return an error. (cuCtxSynchronize, etc) It seems to me it's an optimization codegen bug with nvrtc 9.1 (and possible in 10.0). CUDA 10.1 resolves these issues (and we can use the latest msvc with it!). Tested on this file (default cube + texture): [bake_bug_2.blend](https://archive.blender.org/developer/F6831214/bake_bug_2.blend)

Brecht Van Lommel commented

2019-03-15 21:57:19 +01:00

@ixd, interesting tests. There is indeed an optimization codegen bug that we reported and that was fixed in 10.1.

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Cycles GPU Compute baking often fails #59286