Viewport render crash on GPU Cuda

Use this forum for general user support and related questions.
Forum rules
Please upload a testscene that allows developers to reproduce the problem, and attach some images.
filip
Posts: 22
Joined: Sat Sep 14, 2019 9:53 am

Re: Viewport render crash on GPU Cuda

Post by filip »

As an add-on to my post above, I had a LuxCore 2.4 alpha install and tried it and the issue does not happen in that.
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Viewport render crash on GPU Cuda

Post by Dade »

filip wrote: Thu Jul 02, 2020 11:02 am So, in the scene I uploaded, if I disable Light Tracing the viewport render is split in half, with half of it looking more noisy than the other. See the attached picture.
When you enable light tracing, BlendLuxCore will use PATHOCL render engine instead of RTPATHOCL. PATHOCL is the render engine for final render (so there is no difference between view port rendering and final rendering in this case).
PATHOCL is optimize for rendering time but has long editing time. RTPATHOCL is optimized for fast scene editing but has slower overall rendering times. This is way usually RTPATHOCL is used for view port rendering (where you are likely to edit the scene often). However light tracing is not supported by RTPATHOCL so PATHOCL is used instead.

This explain the different behavior you are observing between enabling/disabling light tracing.
filip wrote: Thu Jul 02, 2020 11:02 am Even if I disable Light Tracing with the viewport render off and then start the viewport render, it will crash or have this weird half image issue.
RTPATHOCL equally splits the image across available GPUs so your 2070 renders half of the image while the 1060 renders the other half (i.e. the 1060 is slowing down your rendering with RTPATHOCL, you should use it only the 2070 for view port rendering).

@B.Y.O.B.: due the no dynamic load balancing nature of RTPATHOCL, it could be a good idea, at the moment, to always use PATHOCL for view port rendering when multiple-GPUs (with different names) are available :idea:

There is probably a bug freezing the rendering of the second GPU so this should explain the half image issue.
Support LuxCoreRender project with salts and bounties
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Viewport render crash on GPU Cuda

Post by Dade »

Dade wrote: Sat Jul 04, 2020 10:28 am There is probably a bug freezing the rendering of the second GPU so this should explain the half image issue.
I have found and fixed another problem in RTPATHOCL when using multiple GPUs. Check the next automatic build if it works fine now.
Support LuxCoreRender project with salts and bounties
Txamo
Posts: 16
Joined: Sat Jul 04, 2020 11:28 am

Re: Viewport render crash on GPU Cuda

Post by Txamo »

Good morning guys

I have already reported the bug, on github, but I record it here.

I am testing with version 2.4 and I find unexpected closings of Blender, the problem reproduces when I activate more than 2 GPUs and and I launch a render in viewport, in the final render this does not happen, I have 4x 2080ti.

The scene is nothing special, with any scene this happens, in my case I have tried with a simple box and a light and Blender always closes when using more than two GPUs.
filip
Posts: 22
Joined: Sat Sep 14, 2019 9:53 am

Re: Viewport render crash on GPU Cuda

Post by filip »

Yep, that seems to have fixed the half-image issues. Thank you very much!

I'll see if there are still crashes too. Before it crashed quite often with both cards and was much stable with just one.
Txamo
Posts: 16
Joined: Sat Jul 04, 2020 11:28 am

Re: Viewport render crash on GPU Cuda

Post by Txamo »

Hi Dade,

Sorry I just downloaded the latest version a couple of hours ago, now with the 4xGPUs Blender does not close in the render by viewport.

I will keep testing it, excellent work Dade.

If you need me to do a multi GPU test, tell me and I will help you.

Regards.
sarmath
Posts: 135
Joined: Wed Feb 07, 2018 1:22 pm

Re: Viewport render crash on GPU Cuda

Post by sarmath »

Hi.

Just tested latest build and....no more crushes. Great work. Thanks :!:
WIndows 11 Pro Ryzen 7 5700x RTX 3090 32 GB RAM
TStrolia
Posts: 40
Joined: Mon Jun 22, 2020 10:29 am

Re: Viewport render crash on GPU Cuda

Post by TStrolia »

Blender 2.83.5, 2.83.2, 2.9. i7-7700k. GTX 1080ti and 780. 32gb ram. Windows 10.

I hate drudging all this up again but I seem to be having the same issue. I have been using 2.4 that I downloaded the day of release but I realized I wasn't using cuda this whole time. Switched over and now viewport render doesn't work. I tried testing the latest daily build and no luck either. Turning off a second graphics card seems to make no difference. 1 card, 2 cards, I get hung up on compiling CUDA kernels in the viewport and blender just hangs...

This may not be a CUDA issue after all... I don't know what happened between yesterday and now but I seem to be having the problem all around now that I am switching back to OGL... hmmm, gonna run it with the console and see if I can figure out if this is me.

Console isn't telling me much. Gonna try a graphics driver update and see if that does anything but it's weird that it was working and now it's not.

update: driver update did nothing. I cannot figure out where the problem is... hmmm

Another update: I really cant tell if it's me anymore. I tried rolling back my PC. Updating my pc. sfc /scannow. uninstalling drivers. cleaning drivers. reinstalling drivers. nothing fixes the hang. I am not sure why it suddenly stopped working. :?

Ok, now we have gone one step further. Before it was just in the viewport but seemed to be rendering fine. Now it's not rendering at all and I am even more confused than when I started. This is what shows up in the console. I don't see an error but maybe B.Y.O.B or Dade will see something.

Code: Select all

[Engine/Final] Rendering layer "View Layer"
[Exporter] Creating session
[SDL][164.640] Define ImageMap: NamedObject
[SDL][164.640] Camera type: perspective
[SDL][164.640] Camera position: Point[1.70621, -5.51184, 0.688014]
[SDL][164.640] Camera target: Point[1.41258, -4.56328, 0.56961]
[SDL][164.640] Camera clipping plane disabled
WARNING: No material defined
[SDL][164.640] Material definition: __CLAY__
[SDL][164.640] Scene objects count: 1
[SDL][164.640] Light definition: __WORLD_BACKGROUND_LIGHT__
[SDL][164.640] Reading texture map: W:\TEXTURES\HDRI\517\Indoor\aft_lounge_8k.hdr
[LuxCore][166.109] Configuration:
[LuxCore][166.109]   path.pathdepth.total = 7
[LuxCore][166.109]   path.pathdepth.diffuse = 5
[LuxCore][166.109]   path.pathdepth.glossy = 5
[LuxCore][166.109]   path.pathdepth.specular = 6
[LuxCore][166.109]   path.hybridbackforward.enable = 0
[LuxCore][166.109]   path.hybridbackforward.partition = 0
[LuxCore][166.109]   path.hybridbackforward.glossinessthreshold = 0.048999998718500137
[LuxCore][166.109]   opencl.cpu.use = 0
[LuxCore][166.109]   opencl.gpu.use = 1
[LuxCore][166.109]   opencl.devices.select = "0011"
[LuxCore][166.109]   film.noiseestimation.warmup = 8
[LuxCore][166.109]   film.noiseestimation.step = 32
[LuxCore][166.109]   sampler.sobol.adaptive.strength = 0.89999997615814209
[LuxCore][166.109]   sampler.sobol.bucketsize = 16
[LuxCore][166.109]   sampler.sobol.tilesize = 16
[LuxCore][166.109]   sampler.sobol.supersampling = 1
[LuxCore][166.109]   sampler.sobol.overlapping = 1
[LuxCore][166.109]   renderengine.type = "PATHOCL"
[LuxCore][166.109]   sampler.type = "SOBOL"
[LuxCore][166.109]   film.width = 1920
[LuxCore][166.109]   film.height = 1080
[LuxCore][166.109]   film.filter.type = "BLACKMANHARRIS"
[LuxCore][166.109]   film.filter.width = 1.5
[LuxCore][166.109]   lightstrategy.type = "LOG_POWER"
[LuxCore][166.109]   scene.epsilon.min = 9.9999997473787516e-06
[LuxCore][166.109]   scene.epsilon.max = 0.10000000149011612
[LuxCore][166.109]   film.opencl.enable = 1
[LuxCore][166.109]   film.opencl.device = 2
[LuxCore][166.109]   path.forceblackbackground.enable = 0
[LuxCore][166.109]   context.cuda.optix.enable = 1
[LuxCore][166.109]   renderengine.seed = 1
[LuxCore][166.109]   film.outputs.0.type = "RGB_IMAGEPIPELINE"
[LuxCore][166.109]   film.outputs.0.index = 0
[LuxCore][166.109]   film.outputs.0.filename = "RGB_IMAGEPIPELINE_0.png"
[LuxCore][166.109]   film.outputs.1.type = "ALBEDO"
[LuxCore][166.109]   film.outputs.1.filename = "ALBEDO.exr"
[LuxCore][166.109]   film.outputs.2.type = "AVG_SHADING_NORMAL"
[LuxCore][166.109]   film.outputs.2.filename = "AVG_SHADING_NORMAL.exr"
[LuxCore][166.125]   film.outputs.3.type = "RGB_IMAGEPIPELINE"
[LuxCore][166.125]   film.outputs.3.index = 1
[LuxCore][166.125]   film.outputs.3.filename = "RGB_IMAGEPIPELINE_1.png"
[LuxCore][166.125]   film.outputs.4.type = "RGB_IMAGEPIPELINE"
[LuxCore][166.125]   film.outputs.4.index = 2
[LuxCore][166.125]   film.outputs.4.filename = "RGB_IMAGEPIPELINE_2.png"
[LuxCore][166.125]   film.imagepipelines.001.0.type = "INTEL_OIDN"
[LuxCore][166.125]   film.imagepipelines.001.0.oidnmemory = 6000
[LuxCore][166.125]   film.imagepipelines.001.0.sharpness = 0
[LuxCore][166.125]   film.imagepipelines.001.1.type = "NOP"
[LuxCore][166.125]   film.imagepipelines.001.2.type = "TONEMAP_LINEAR"
[LuxCore][166.125]   film.imagepipelines.001.2.scale = 1
[LuxCore][166.125]   film.imagepipelines.001.radiancescales.0.enabled = 1
[LuxCore][166.125]   film.imagepipelines.001.radiancescales.0.globalscale = 1
[LuxCore][166.125]   film.imagepipelines.001.radiancescales.0.rgbscale = 1 1 1
[LuxCore][166.125]   film.imagepipelines.002.0.type = "NOP"
[LuxCore][166.125]   film.imagepipelines.002.1.type = "TONEMAP_LINEAR"
[LuxCore][166.125]   film.imagepipelines.002.1.scale = 1
[LuxCore][166.125]   film.imagepipelines.002.radiancescales.0.enabled = 1
[LuxCore][166.125]   film.imagepipelines.002.radiancescales.0.globalscale = 1
[LuxCore][166.125]   film.imagepipelines.002.radiancescales.0.rgbscale = 1 1 1
[LuxCore][166.125]   film.imagepipelines.002.2.type = "GAMMA_CORRECTION"
[LuxCore][166.125]   film.imagepipelines.002.2.value = 2.2000000000000002
[LuxCore][166.125]   film.noiseestimation.index = 2
[LuxCore][166.125]   film.imagepipelines.000.0.type = "NOP"
[LuxCore][166.125]   film.imagepipelines.000.1.type = "TONEMAP_LINEAR"
[LuxCore][166.125]   film.imagepipelines.000.1.scale = 1
[LuxCore][166.125]   film.imagepipelines.000.radiancescales.0.enabled = 1
[LuxCore][166.125]   film.imagepipelines.000.radiancescales.0.globalscale = 1
[LuxCore][166.125]   film.imagepipelines.000.radiancescales.0.rgbscale = 1 1 1
[LuxCore][166.125]   batch.haltspp = 1,000 0
[LuxCore][166.125]   batch.halttime = 0
[LuxCore][166.125] File Name Resolver Configuration:
Export took 1.5 s
====================================================================
Step: 0/2
Creating kernel cache entry with configuration properties:
renderengine.type = "PATHOCL"
sampler.type = "SOBOL"
scene.epsilon.min = 9.9999997473787516e-06
scene.epsilon.max = 0.10000000149011612
opencl.devices.select = "0011"

And scene properties:

[SDL][166.140] Define ImageMap: NamedObject
[SDL][166.140] Define ImageMap: image.png
[SDL][166.140] Camera type: perspective
[SDL][166.140] Camera position: Point[1, 6, 3]
[SDL][166.140] Camera target: Point[0, 0, 0.5]
[SDL][166.140] Camera clipping plane disabled
[SDL][166.140] Light definition: infinite_light
[SDL][166.140] Material definition: matte_mat
[SDL][166.140] Scene objects count: 1
[SDL][166.140] Texture definition: constfloat3_tex
[SDL][166.140] Material definition: constfloat3_tmat
[SDL][166.140] Scene objects count: 1
[LuxCore][166.140] Configuration:
[LuxCore][166.140]   renderengine.type = "PATHOCL"
[LuxCore][166.140]   sampler.type = "SOBOL"
[LuxCore][166.140]   scene.epsilon.min = 9.9999997473787516e-06
[LuxCore][166.140]   scene.epsilon.max = 0.10000000149011612
[LuxCore][166.140]   opencl.devices.select = "0011"
[LuxCore][166.140]   film.outputs.1.type = "RGB_IMAGEPIPELINE"
[LuxCore][166.140]   film.outputs.1.filename = "image.png"
[LuxCore][166.140] File Name Resolver Configuration:
[LuxCore][166.140] Film resolution: 640x640
[SDL][166.140] Film output definition: RGB_IMAGEPIPELINE [image.png]
[SDL][166.140] Film output definition: RGB_IMAGEPIPELINE [image.png]
[LuxRays][166.140] OpenCL support: enabled
[LuxRays][166.140] OpenCL Platform 0: NVIDIA CUDA
[LuxRays][166.140] CUDA support: enabled
[LuxRays][166.140] CUDA support: available
[LuxRays][166.140] CUDA driver version: 11.0
[LuxRays][166.140] CUDA device count: 2
[LuxRays][166.140] Optix support: available
[LuxRays][166.140] Device 0 name: Native
[LuxRays][166.140] Device 0 type: NATIVE_THREAD
[LuxRays][166.140] Device 0 compute units: 1
[LuxRays][166.140] Device 0 preferred float vector width: 4
[LuxRays][166.140] Device 0 max allocable memory: 17592186044415MBytes
[LuxRays][166.140] Device 0 max allocable memory block size: 17592186044415MBytes
[LuxRays][166.140] Device 0 has out of core memory support: 0
[LuxRays][166.140] Device 1 name: GeForce GTX 1080 Ti
[LuxRays][166.140] Device 1 type: OPENCL_GPU
[LuxRays][166.140] Device 1 compute units: 28
[LuxRays][166.140] Device 1 preferred float vector width: 1
[LuxRays][166.140] Device 1 max allocable memory: 11264MBytes
[LuxRays][166.140] Device 1 max allocable memory block size: 2816MBytes
[LuxRays][166.140] Device 1 has out of core memory support: 0
[LuxRays][166.140] Device 2 name: GeForce GTX 780
[LuxRays][166.140] Device 2 type: OPENCL_GPU
[LuxRays][166.140] Device 2 compute units: 12
[LuxRays][166.140] Device 2 preferred float vector width: 1
[LuxRays][166.156] Device 2 max allocable memory: 3072MBytes
[LuxRays][166.156] Device 2 max allocable memory block size: 768MBytes
[LuxRays][166.156] Device 2 has out of core memory support: 0
[LuxRays][166.156] Device 3 name: GeForce GTX 1080 Ti
[LuxRays][166.156] Device 3 type: CUDA_GPU
[LuxRays][166.156] Device 3 compute units: 128
[LuxRays][166.156] Device 3 preferred float vector width: 1
[LuxRays][166.156] Device 3 max allocable memory: 11264MBytes
[LuxRays][166.156] Device 3 max allocable memory block size: 17592186044415MBytes
[LuxRays][166.156] Device 3 has out of core memory support: 1
[LuxRays][166.156] Device 3 CUDA compute capability: 6.1
[LuxRays][166.156] Device 4 name: GeForce GTX 780
[LuxRays][166.156] Device 4 type: CUDA_GPU
[LuxRays][166.156] Device 4 compute units: 192
[LuxRays][166.156] Device 4 preferred float vector width: 1
[LuxRays][166.156] Device 4 max allocable memory: 3072MBytes
[LuxRays][166.156] Device 4 max allocable memory block size: 17592186044415MBytes
[LuxRays][166.156] Device 4 has out of core memory support: 1
[LuxRays][166.156] Device 4 CUDA compute capability: 3.5
[LuxRays][166.156] Creating 10 intersection device(s)
[LuxRays][166.156] Allocating intersection device 0: GeForce GTX 1080 Ti (Type = CUDA_GPU)
[LuxRays][166.218] Allocating intersection device 1: GeForce GTX 780 (Type = CUDA_GPU)
[LuxRays][166.250] Allocating intersection device 2: Native (Type = NATIVE_THREAD)
[LuxRays][166.250] Allocating intersection device 3: Native (Type = NATIVE_THREAD)
[LuxRays][166.265] Allocating intersection device 4: Native (Type = NATIVE_THREAD)
[LuxRays][166.265] Allocating intersection device 5: Native (Type = NATIVE_THREAD)
[LuxRays][166.265] Allocating intersection device 6: Native (Type = NATIVE_THREAD)
[LuxRays][166.265] Allocating intersection device 7: Native (Type = NATIVE_THREAD)
[LuxRays][166.265] Allocating intersection device 8: Native (Type = NATIVE_THREAD)
[LuxRays][166.265] Allocating intersection device 9: Native (Type = NATIVE_THREAD)
[LuxCore][166.265] CUDA devices used:
[LuxCore][166.265] [GeForce GTX 1080 Ti CUDAIntersect (Optix enabled: 0)]
[LuxCore][166.265] [GeForce GTX 780 CUDAIntersect (Optix enabled: 0)]
[LuxCore][166.265] OpenCL devices used:
[LuxCore][166.265] Native devices used: 8
[LuxCore][166.265] Configuring 2 OpenCL render threads
[LuxCore][166.265] Configuring 8 native render threads
[LuxRays][166.265] Preprocessing DataSet
[LuxRays][166.265] Total vertex count: 48
[LuxRays][166.265] Total triangle count: 24
[LuxRays][166.265] Preprocessing DataSet done
[LuxRays][166.265] Adding DataSet accelerator: BVH
[LuxRays][166.265] Total vertex count: 48
[LuxRays][166.265] Total triangle count: 24
[LuxRays][166.265] BVH Dataset preprocessing time: 0ms
[LuxRays][166.265] BVH builder: EMBREE_BINNED_SAH
[LuxRays][166.265] BVH build hierarchy time: 0ms
[LuxRays][166.265] BVH total build time: 0ms
[LuxRays][166.265] Total BVH memory usage: 1Kbytes
[LuxRays][166.265] Adding DataSet accelerator: EMBREE
[LuxRays][166.265] Total vertex count: 48
[LuxRays][166.265] Total triangle count: 24
[LuxRays][166.265] EmbreeAccel build time: 0ms
[LuxRays][166.265] [Device GeForce GTX 1080 Ti CUDAIntersect] BVH mesh vertices buffer size: 576bytes
[LuxRays][166.265] [Device GeForce GTX 1080 Ti CUDAIntersect] BVH nodes buffer size: 1248bytes
[LuxRays][166.265] [BVHKernel] Compiler options: -D LUXRAYS_OPENCL_KERNEL -D PARAM_RAY_EPSILON_MIN=1e-05f -D PARAM_RAY_EPSILON_MAX=0.1f -D LUXRAYS_CUDA_DEVICE -D LUXRAYS_OS_WINDOWS --use_fast_math
[LuxRays][166.265] [BVHKernel] Compiling kernels
[LuxRays][166.265] [BVHKernel] Program cached
[LuxRays][166.265] [Device GeForce GTX 780 CUDAIntersect] BVH mesh vertices buffer size: 576bytes
[LuxRays][166.265] [Device GeForce GTX 780 CUDAIntersect] BVH nodes buffer size: 1248bytes
[LuxRays][166.265] [BVHKernel] Compiler options: -D LUXRAYS_OPENCL_KERNEL -D PARAM_RAY_EPSILON_MIN=1e-05f -D PARAM_RAY_EPSILON_MAX=0.1f -D LUXRAYS_CUDA_DEVICE -D LUXRAYS_OS_WINDOWS --use_fast_math
[LuxRays][166.265] [BVHKernel] Compiling kernels
[LuxRays][166.265] [BVHKernel] Program cached
[LuxCore][166.281] [PathOCLRenderEngine] OpenCL task count: 131072
[LuxCore][166.281] [PathOCLBaseRenderEngine] OpenCL max. page memory size: 18014398509481983Kbytes
[LuxCore][166.281] Compile Geometry
[LuxCore][166.281] Scene geometry compilation time: 0ms
[LuxCore][166.281] Compile 2 Textures
[LuxCore][166.281] Texture evaluation ops count: 6
[LuxCore][166.281] Texture evaluation max. stack size: 3
[LuxCore][166.281] Textures compilation time: 0ms
[LuxCore][166.281] Compile 2 Materials
[LuxCore][166.281] Material evaluation ops count: 14
[LuxCore][166.281] Material evaluation max. stack size: 8
[LuxCore][166.281] Material compilation time: 0ms
[LuxCore][166.281] Compile Lights
[LuxCore][166.281] Lights compilation time: 0ms
[LuxCore][166.281] Compile ImageMaps
[LuxCore][166.281] Image maps page(s) count: 1
[LuxCore][166.281]  RGB channel page 0 size: 3264Kbytes
[LuxCore][166.281] Image maps compilation time: 0ms
[LuxCore][166.281] Always enabled OpenCL code:
[LuxCore][166.281] Compile Geometry
[LuxCore][166.281] Scene geometry compilation time: 0ms
[LuxCore][166.281] Compile 2 Textures
[LuxCore][166.281] Texture evaluation ops count: 6
[LuxCore][166.281] Texture evaluation max. stack size: 3
[LuxCore][166.281] Textures compilation time: 0ms
[LuxCore][166.281] Compile 2 Materials
[LuxCore][166.281] Material evaluation ops count: 14
[LuxCore][166.281] Material evaluation max. stack size: 8
[LuxCore][166.281] Material compilation time: 0ms
[LuxCore][166.281] Compile Lights
[LuxCore][166.297] Lights compilation time: 0ms
[LuxCore][166.297] Compile ImageMaps
[LuxCore][166.297] Image maps page(s) count: 1
[LuxCore][166.297]  RGB channel page 0 size: 3264Kbytes
[LuxCore][166.297] Image maps compilation time: 0ms
[LuxCore][166.297] Starting 2 OpenCL render threads
[LuxRays][166.297] [Device GeForce GTX 1080 Ti CUDAIntersect] RADIANCE_PER_PIXEL_NORMALIZEDs[0] buffer size: 6400Kbytes
[LuxRays][166.297] [Device GeForce GTX 1080 Ti CUDAIntersect] NOISE buffer size: 1600Kbytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] RADIANCE_PER_PIXEL_NORMALIZEDs[0] buffer size: 6400Kbytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Camera buffer size: 5488bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Triangle normals buffer size: 288bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Vertices buffer size: 576bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Triangles buffer size: 288bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Mesh description buffer size: 624bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] ImageMap descriptions buffer size: 64bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] ImageMaps buffer size: 3264Kbytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Textures buffer size: 592bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Texture evaluation ops buffer size: 48bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Texture evaluation stacks buffer size: 1536Kbytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Materials buffer size: 456bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Material evaluation ops buffer size: 168bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Material evaluation stacks buffer size: 4096Kbytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Scene objects buffer size: 48bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Lights buffer size: 332bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Env. light indices buffer size: 4bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Light offsets (Part I) buffer size: 8bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Env. light distributions buffer size: 516Kbytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] LightsDistribution buffer size: 16bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] InfiniteLightSourcesDistribution buffer size: 16bytes
[LuxRays][166.312] [Device GeForce GTX 1080 Ti CUDAIntersect] Ray buffer size: 6144Kbytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] RayHit buffer size: 2560Kbytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] GPUTaskConfiguration buffer size: 328bytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] GPUTask buffer size: 84992Kbytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] GPUTaskDirectLight buffer size: 7680Kbytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] GPUTaskState buffer size: 50176Kbytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] GPUTask Stats buffer size: 512Kbytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] SamplerSharedData buffer size: 1608Kbytes
[LuxCore][166.328] [PathOCLBaseRenderThread::0] Size of a Sample: 40bytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] Sample buffer size: 5120Kbytes
[LuxCore][166.328] [PathOCLBaseRenderThread::0] Size of a SampleData: 8bytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] SampleData buffer size: 1024Kbytes
[LuxCore][166.328] [PathOCLBaseRenderThread::0] Size of a SampleResult: 428bytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] SampleResult buffer size: 54784Kbytes
[LuxRays][166.328] [Device GeForce GTX 1080 Ti CUDAIntersect] PathInfo buffer size: 13824Kbytes
[LuxRays][166.343] [Device GeForce GTX 1080 Ti CUDAIntersect] DirectLightVolumeInfo buffer size: 5632Kbytes
[LuxRays][166.343] [Device GeForce GTX 1080 Ti CUDAIntersect] Pixel Filter Distribution buffer size: 33Kbytes
[LuxCore][166.343] [PathOCLBaseRenderThread::0] Compiling kernels
[LuxRays][166.343] [PathOCL kernel] Compiler options: -D LUXRAYS_OPENCL_KERNEL -D SLG_OPENCL_KERNEL -D RENDER_ENGINE_PATHOCL -D PARAM_RAY_EPSILON_MIN=1e-05f -D PARAM_RAY_EPSILON_MAX=0.1f -D LUXRAYS_CUDA_DEVICE -D LUXRAYS_OS_WINDOWS --use_fast_math
[LuxRays][166.343] [PathOCL kernel] Compiling kernels
Then just hangs on compiling kernels
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Viewport render crash on GPU Cuda

Post by Dade »

TStrolia wrote: Sat Sep 05, 2020 1:22 am Then just hangs on compiling kernels
Compiling a kernel takes about 90-100 seconds on a AMD Ryzen 3900x (so ~200 seconds for 2 GPUs). How long have you waited ?
Support LuxCoreRender project with salts and bounties
TStrolia
Posts: 40
Joined: Mon Jun 22, 2020 10:29 am

Re: Viewport render crash on GPU Cuda

Post by TStrolia »

Dade wrote: Sat Sep 05, 2020 11:32 am
TStrolia wrote: Sat Sep 05, 2020 1:22 am Then just hangs on compiling kernels
Compiling a kernel takes about 90-100 seconds on a AMD Ryzen 3900x (so ~200 seconds for 2 GPUs). How long have you waited ?
around 30 minutes. Not specifically in what I copied from the console above. That time I only ran a render so I could copy what was happening for this post and I only waited a few minutes after copying, but I tried well over 30 multiple times. When it was working it was taking around a minute or so on my 1080ti on a large scene so I know it takes a minute but it really LOCKS UP locks up now. Before I could see the progress bar moving in blender, now it just stays at 0. Compiling kernel would usually take a few moments and then it would add parenthesis and tell me what it was compiling, nothing now. It just sits there for half an hour laughing at me lol :lol: but in all seriousness after using luxcore for the last couple months I have not had a problem like this that completely locksup blender over a cube and an IBL even if I walk away and watch a show and come back.
Post Reply