Re: Optix/RTX support
Posted: Fri Jul 31, 2020 4:02 pm
I have the game drivers, no Optix or CUDA SDK, and Optix is working fine here on Windows.
Show your work, get help, participate in development
https://forums.luxcorerender.org/
You can run CUDA 10 PTX on CUDA 11 but not CUDA 11 PTX on CUDA 10 so using an older compiler (NVRTC) is not a problem but a newer requires the very latest drivers.acasta69 wrote: ↑Sat Aug 01, 2020 11:57 am Since our Optix/RTX implementation needs Cuda 11, should we update the nvrtc and nvrtc-builtins libraries to v11?
At the moment we are distributing v10.1, at least in Windows packages.
I was testing if the GTX970 + driver v451.67 is working with Optix and had some strange behaviour (i.e. LuxCoreUI suddenly exiting without error messages). I was wondering if that could be the cause.
Code: Select all
F:\Ale\Devel\LuxCoreRender\LuxCoreTestScenes>C:\Users\Alessandro\Desktop\luxcorerender-latest-win64\luxcoreui.exe scenes\LuxMark-Hotel\render.cfg
LuxCoreUI v2.5alpha0 (LuxCore demo: http://www.luxcorerender.org)
[LuxCore][0.000] Configuration:
[LuxCore][0.000] opencl.platform.index = "-1"
[LuxCore][0.000] opencl.cpu.use = "0"
[LuxCore][0.000] opencl.gpu.use = "1"
[LuxCore][0.000] opencl.devices.select = "01"
[LuxCore][0.000] opencl.native.threads.count = "0"
[LuxCore][0.000] scene.epsilon.min = "9.99999972e-10"
[LuxCore][0.000] scene.epsilon.max = "0.100000001"
[LuxCore][0.000] renderengine.type = "PATHOCL"
[LuxCore][0.000] accelerator.instances.enable = "1"
[LuxCore][0.000] path.maxdepth = "6"
[LuxCore][0.000] path.clamping.radiance.maxvalue = "32000.0"
[LuxCore][0.000] biaspath.clamping.radiance.maxvalue = "32000.0"
[LuxCore][0.000] lightstrategy.type = "UNIFORM"
[LuxCore][0.000] film.width = "1024"
[LuxCore][0.000] film.height = "576"
[LuxCore][0.000] sampler.type = "SOBOL"
[LuxCore][0.000] film.opencl.enable = "0"
[LuxCore][0.000] film.filter.type = "GAUSSIAN"
[LuxCore][0.000] film.filter.xwidth = "2"
[LuxCore][0.000] film.filter.ywidth = "2"
[LuxCore][0.000] film.imagepipeline.0.type = "TONEMAP_LINEAR"
[LuxCore][0.000] film.imagepipeline.0.scale = "0.000776737288"
[LuxCore][0.000] film.imagepipeline.1.type = "GAMMA_CORRECTION"
[LuxCore][0.000] film.imagepipeline.1.value = "2.20000005"
[LuxCore][0.000] scene.file = "scene.scn"
[LuxCore][0.000] periodicsave.film.outputs.period = "600"
[LuxCore][0.000] batch.haltspp = "400"
[LuxCore][0.000] File Name Resolver Configuration:
[LuxCore][0.000] .
[LuxCore][0.000] scenes/LuxMark-Hotel
[SDL][0.000] Reading scene: scenes/LuxMark-Hotel/scene.scn
[SDL][0.015] Texture definition: texture-0x7f9fb27f3290
[SDL][0.015] Texture definition: texture-0x7f9fa03399b0
[SDL][0.015] Texture definition: texture-0x7f9f9f815310
[SDL][0.015] Texture definition: texture-0x7f9fb29d0010
[SDL][0.015] Reading texture map: scenes/LuxMark-Hotel/imagemap-00000.exr
...
[SDL][0.312] Material definition: material-0x7f9fa03396a0
...
[SDL][0.328] Camera type: perspective
[SDL][0.328] Camera position: Point[4.03518, -5.64324, 1.16999]
[SDL][0.328] Camera target: Point[3.46807, -4.81972, 1.15577]
[SDL][0.328] Camera clipping plane disabled
[SDL][0.812] Scene objects count: 496
[SDL][0.812] Light definition: light-0x7f9fa0339bd0
[SDL][0.812] Light definition: light-0x7f9fa033a770
[SDL][0.984] Camera type: perspective
[SDL][0.984] Camera position: Point[4.03518, -5.64324, 1.16999]
[SDL][0.984] Camera target: Point[3.46807, -4.81972, 1.15577]
[SDL][0.984] Camera clipping plane disabled
Film size adjusted: 1024x576 (Frame buffer size: 1024x576)
RenderConfig has cached kernels: True
[LuxCore][1.000] Film resolution: 1024x576
[SDL][1.000] Film output definition: RGB_IMAGEPIPELINE [image.png]
[SDL][1.000] Image pipeline: film.imagepipeline
[SDL][1.000] Image pipeline step 0: TONEMAP_LINEAR
[SDL][1.000] Image pipeline step 1: GAMMA_CORRECTION
[LuxRays][1.000] OpenCL support: enabled
[LuxRays][1.015] OpenCL Platform 0: NVIDIA CUDA
[LuxRays][1.015] CUDA support: enabled
[LuxRays][1.015] CUDA support: available
[LuxRays][1.015] CUDA driver version: 11.0
[LuxRays][1.015] CUDA device count: 1
[LuxRays][1.015] Optix support: available
[LuxRays][1.015] Device 0 name: Native
[LuxRays][1.015] Device 0 type: NATIVE_THREAD
[LuxRays][1.015] Device 0 compute units: 1
[LuxRays][1.015] Device 0 preferred float vector width: 4
[LuxRays][1.015] Device 0 max allocable memory: 17592186044415MBytes
[LuxRays][1.015] Device 0 max allocable memory block size: 17592186044415MBytes
[LuxRays][1.015] Device 0 has out of core memory support: 0
[LuxRays][1.015] Device 1 name: GeForce GTX 970
[LuxRays][1.015] Device 1 type: OPENCL_GPU
[LuxRays][1.015] Device 1 compute units: 13
[LuxRays][1.015] Device 1 preferred float vector width: 1
[LuxRays][1.015] Device 1 max allocable memory: 4096MBytes
[LuxRays][1.015] Device 1 max allocable memory block size: 1024MBytes
[LuxRays][1.015] Device 1 has out of core memory support: 0
[LuxRays][1.015] Device 2 name: GeForce GTX 970
[LuxRays][1.015] Device 2 type: CUDA_GPU
[LuxRays][1.015] Device 2 compute units: 128
[LuxRays][1.015] Device 2 preferred float vector width: 1
[LuxRays][1.015] Device 2 max allocable memory: 4096MBytes
[LuxRays][1.015] Device 2 max allocable memory block size: 17592186044415MBytes
[LuxRays][1.015] Device 2 has out of core memory support: 1
[LuxRays][1.015] Creating 1 intersection device(s)
[LuxRays][1.015] Allocating intersection device 0: GeForce GTX 970 (Type = CUDA_GPU)
[LuxRays][1.078] [Optix][4][KNOBS] All knobs on default.
[LuxRays][1.125] [Optix][4][DISK CACHE] Opened database: "C:\Users\Alessandro\AppData\Local\NVIDIA\OptixCache\cache7.db"
[LuxRays][1.125] [Optix][4][DISK CACHE] Cache data size: "17.3 KiB"
[LuxCore][1.125] CUDA devices used:
[LuxCore][1.125] [GeForce GTX 970 CUDAIntersect]
[LuxCore][1.125] OpenCL devices used:
[LuxCore][1.125] Native devices used: 0
[LuxCore][1.125] Configuring 1 OpenCL render threads
[LuxCore][1.125] Configuring 0 native render threads
[LuxRays][1.125] Preprocessing DataSet
[LuxRays][1.125] Total vertex count: 3229390
[LuxRays][1.125] Total triangle count: 4973406
[LuxRays][1.125] Preprocessing DataSet done
[LuxRays][1.125] Adding DataSet accelerator: OPTIX
[LuxRays][1.125] Total vertex count: 3229390
[LuxRays][1.125] Total triangle count: 4973406
[LuxRays][1.125] Building Optix accelerator
[LuxRays][1.469] Optix accelerator leafs: 496
[LuxRays][1.469] OptixAccel total build time: 343ms
[LuxRays][1.469] [OptixAccel] Compiler options: -D LUXRAYS_OPENCL_KERNEL -D PARAM_RAY_EPSILON_MIN=1e-09f -D PARAM_RAY_EPSILON_MAX=0.1f -D LUXRAYS_CUDA_DEVICE -D LUXRAYS_OS_WINDOWS --use_fast_math
[LuxRays][1.469] [OptixAccel] Program cached
[LuxRays][1.469] [Optix][4][DISKCACHE] Cache hit for key: ptx-2547-keyd4ee84a65eb13d6eecc98490c7e715b1-sm_52-rtc0-drv451.67
[LuxRays][1.484] [Optix][4][COMPILE FEEDBACK] Info: Pipeline has 1 module(s), 3 entry function(s), 1 trace call(s), 0 continuation callable call(s), 0 direct callable call(s), 10 basic block(s) in entry functions, 95 instruction(s) in entry functions, 7 non-entry function(s), 42 basic block(s) in non-entry functions, 616 instruction(s) in non-entry functions
[LuxRays][1.484] Total Optix memory usage: 278237Kbytes
[LuxRays][1.484] Adding DataSet accelerator: EMBREE
[LuxRays][1.484] Total vertex count: 3229390
[LuxRays][1.484] Total triangle count: 4973406
[LuxRays][1.844] EmbreeAccel build time: 359ms
[LuxCore][1.844] [PathOCLRenderEngine] OpenCL task count: 131072
[LuxCore][1.844] [PathOCLBaseRenderEngine] OpenCL max. page memory size: 18014398509481983Kbytes
[LuxCore][1.844] Compile Geometry
[LuxCore][1.969] Scene geometry compilation time: 125ms
[LuxCore][1.969] Compile 256 Textures
[LuxCore][1.969] Texture evaluation ops count: 872
[LuxCore][1.969] Texture evaluation max. stack size: 11
[LuxCore][1.969] Textures compilation time: 0ms
[LuxCore][1.969] Compile 53 Materials
[LuxCore][1.969] Material evaluation ops count: 371
[LuxCore][1.969] Material evaluation max. stack size: 8
[LuxCore][1.969] Material compilation time: 0ms
[LuxCore][1.969] Compile Lights
[LuxCore][1.969] Lights compilation time: 0ms
[LuxCore][1.969] Compile ImageMaps
[LuxCore][2.031] Image maps page(s) count: 1
[LuxCore][2.031] RGB channel page 0 size: 50176Kbytes
[LuxCore][2.031] Image maps compilation time: 61ms
[LuxCore][2.031] Always enabled OpenCL code:
[LuxCore][2.031] Compile Geometry
[LuxCore][2.062] Scene geometry compilation time: 31ms
[LuxCore][2.062] Compile 256 Textures
[LuxCore][2.062] Texture evaluation ops count: 872
[LuxCore][2.062] Texture evaluation max. stack size: 11
[LuxCore][2.062] Textures compilation time: 0ms
[LuxCore][2.062] Compile 53 Materials
[LuxCore][2.062] Material evaluation ops count: 371
[LuxCore][2.062] Material evaluation max. stack size: 8
[LuxCore][2.062] Material compilation time: 0ms
[LuxCore][2.062] Compile Lights
[LuxCore][2.062] Lights compilation time: 0ms
[LuxCore][2.062] Compile ImageMaps
[LuxCore][2.125] Image maps page(s) count: 1
[LuxCore][2.125] RGB channel page 0 size: 50176Kbytes
[LuxCore][2.125] Image maps compilation time: 63ms
[LuxCore][2.125] Starting 1 OpenCL render threads
[LuxRays][2.140] [Device GeForce GTX 970 CUDAIntersect] RADIANCE_PER_PIXEL_NORMALIZEDs[0] buffer size: 9216Kbytes
[LuxRays][2.156] [Device GeForce GTX 970 CUDAIntersect] NOISE buffer size: 2304Kbytes
[LuxRays][2.156] [Device GeForce GTX 970 CUDAIntersect] RADIANCE_PER_PIXEL_NORMALIZEDs[0] buffer size: 9216Kbytes
[LuxRays][2.156] [Device GeForce GTX 970 CUDAIntersect] Camera buffer size: 5488bytes
[LuxRays][2.156] [Device GeForce GTX 970 CUDAIntersect] Normals buffer size: 16359Kbytes
[LuxRays][2.172] [Device GeForce GTX 970 CUDAIntersect] UVs buffer size: 8487Kbytes
[LuxRays][2.172] [Device GeForce GTX 970 CUDAIntersect] Triangle normals buffer size: 26930Kbytes
[LuxRays][2.172] [Device GeForce GTX 970 CUDAIntersect] Vertices buffer size: 16359Kbytes
[LuxRays][2.187] [Device GeForce GTX 970 CUDAIntersect] Triangles buffer size: 26930Kbytes
[LuxRays][2.187] [Device GeForce GTX 970 CUDAIntersect] Mesh description buffer size: 151Kbytes
[LuxRays][2.187] [Device GeForce GTX 970 CUDAIntersect] ImageMap descriptions buffer size: 608bytes
[LuxRays][2.187] [Device GeForce GTX 970 CUDAIntersect] ImageMaps buffer size: 50176Kbytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Textures buffer size: 74Kbytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Texture evaluation ops buffer size: 6976bytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Texture evaluation stacks buffer size: 5632Kbytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Materials buffer size: 11Kbytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Material evaluation ops buffer size: 4452bytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Material evaluation stacks buffer size: 4096Kbytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Scene objects buffer size: 11Kbytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Lights buffer size: 664bytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Env. light indices buffer size: 8bytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Light offsets (Part I) buffer size: 1984bytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Env. light distributions buffer size: 1028Kbytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] LightsDistribution buffer size: 24bytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] InfiniteLightSourcesDistribution buffer size: 24bytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] Ray buffer size: 6144Kbytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] RayHit buffer size: 2560Kbytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] GPUTaskConfiguration buffer size: 288bytes
[LuxRays][2.203] [Device GeForce GTX 970 CUDAIntersect] GPUTask buffer size: 84992Kbytes
[LuxRays][2.219] [Device GeForce GTX 970 CUDAIntersect] GPUTaskDirectLight buffer size: 7680Kbytes
[LuxRays][2.219] [Device GeForce GTX 970 CUDAIntersect] GPUTaskState buffer size: 50176Kbytes
[LuxRays][2.219] [Device GeForce GTX 970 CUDAIntersect] GPUTask Stats buffer size: 512Kbytes
[LuxRays][2.219] [Device GeForce GTX 970 CUDAIntersect] SamplerSharedData buffer size: 2312Kbytes
[LuxCore][2.219] [PathOCLBaseRenderThread::0] Size of a Sample: 40bytes
[LuxRays][2.219] [Device GeForce GTX 970 CUDAIntersect] Sample buffer size: 5120Kbytes
[LuxCore][2.219] [PathOCLBaseRenderThread::0] Size of a SampleData: 8bytes
[LuxRays][2.219] [Device GeForce GTX 970 CUDAIntersect] SampleData buffer size: 1024Kbytes
[LuxCore][2.234] [PathOCLBaseRenderThread::0] Size of a SampleResult: 308bytes
[LuxRays][2.234] [Device GeForce GTX 970 CUDAIntersect] Sample buffer size: 39424Kbytes
[LuxRays][2.234] [Device GeForce GTX 970 CUDAIntersect] PathInfo buffer size: 13824Kbytes
[LuxRays][2.234] [Device GeForce GTX 970 CUDAIntersect] DirectLightVolumeInfo buffer size: 5632Kbytes
[LuxRays][2.234] [Device GeForce GTX 970 CUDAIntersect] Pixel Filter Distribution buffer size: 33Kbytes
[LuxCore][2.234] [PathOCLBaseRenderThread::0] Compiling kernels
[LuxRays][2.234] [PathOCL kernel] Compiler options: -D LUXRAYS_OPENCL_KERNEL -D SLG_OPENCL_KERNEL -D RENDER_ENGINE_PATHOCL -D PARAM_RAY_EPSILON_MIN=1e-09f -D PARAM_RAY_EPSILON_MAX=0.1f -D LUXRAYS_CUDA_DEVICE -D LUXRAYS_OS_WINDOWS --use_fast_math
[LuxRays][2.234] [PathOCL kernel] Compiling kernels
[LuxRays][2.281] [PathOCL kernel] Program cached
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling Film_Clear Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling InitSeed Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling Init Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_RT_NEXT_VERTEX Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_HIT_NOTHING Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_HIT_OBJECT Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_RT_DL Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_DL_ILLUMINATE Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_DL_SAMPLE_BSDF Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_GENERATE_NEXT_VERTEX_RAY Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_SPLAT_SAMPLE Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_NEXT_SAMPLE Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Compiling AdvancePaths_MK_GENERATE_CAMERA_RAY Kernel
[LuxCore][2.281] [PathOCLBaseRenderThread::0] AdvancePaths_MK_* workgroup size: 32
[LuxCore][2.281] [PathOCLBaseRenderThread::0] Kernels compilation time: 46ms
[LuxCore][2.281] Starting 0 native render threads
F:\Ale\Devel\LuxCoreRender\LuxCoreTestScenes>
Code: Select all
[LuxRays][1.000] OpenCL support: enabled
[LuxRays][1.015] OpenCL Platform 0: NVIDIA CUDA
[LuxRays][1.015] CUDA support: enabled
[LuxRays][1.015] CUDA support: available
[LuxRays][1.015] CUDA driver version: 11.0
[LuxRays][1.015] CUDA device count: 1
[LuxRays][1.015] Optix support: available
[LuxRays][1.015] Device 0 name: Native
Code: Select all
[LuxRays][0.829] OpenCL support: enabled
[LuxRays][0.844] OpenCL Platform 0: NVIDIA CUDA
[LuxRays][0.844] CUDA support: enabled
[LuxRays][0.844] CUDA support: available
[LuxRays][0.844] CUDA driver version: 11.0
[LuxRays][0.844] CUDA device count: 1
[LuxRays][0.844] Device 0 name: Native
Check the code: https://github.com/LuxCoreRender/LuxCor ... t.cpp#L115acasta69 wrote: ↑Tue Aug 04, 2020 4:44 pm In both cases I use the CUDA_GPU, the only difference seems to me that I have disabled native threads for LuxMark-Hotel.
LuxMark-Hotel:LuxMark-Mic:Code: Select all
[LuxRays][1.000] OpenCL support: enabled [LuxRays][1.015] OpenCL Platform 0: NVIDIA CUDA [LuxRays][1.015] CUDA support: enabled [LuxRays][1.015] CUDA support: available [LuxRays][1.015] CUDA driver version: 11.0 [LuxRays][1.015] CUDA device count: 1 [LuxRays][1.015] Optix support: available [LuxRays][1.015] Device 0 name: Native
Code: Select all
[LuxRays][0.829] OpenCL support: enabled [LuxRays][0.844] OpenCL Platform 0: NVIDIA CUDA [LuxRays][0.844] CUDA support: enabled [LuxRays][0.844] CUDA support: available [LuxRays][0.844] CUDA driver version: 11.0 [LuxRays][0.844] CUDA device count: 1 [LuxRays][0.844] Device 0 name: Native