NVlabs / instant-ngp

Instant neural graphics primitives: lightning fast NeRF and more
https://nvlabs.github.io/instant-ngp
Other
16.01k stars 1.93k forks source link

Vulkan error : build\testbed --scene data/fox #1082

Open 565ee opened 1 year ago

565ee commented 1 year ago

when run build\testbed --scene data/fox : • Vulkan error: loader_validate_device_extensions: Device extension VK_NVX_binary_import not supported by selected physical device or enabled layers. • Vulkan error: vkCreateDevice: Failed to validate extensions in list • WARNING Could not initialize Vulkan and NGX. DLSS not supported

windows 10 x64
nvdia gtx 1080ti
Python  3.9
cuda 11.6
instant-ngp  version=2022-11-16
cmake 3.22
VulkanSDK 1.3.231.1
(py39) D:\Program\Anaconda3\envs\py39\Lib\site-packages\instant-ngp>build\testbed --scene data/fox
04:51:54 INFO     Loading NeRF dataset from
04:51:54 INFO       data\fox\transforms.json
04:51:54 SUCCESS  Loaded 50 images after 0s
04:51:54 INFO       cam_aabb=[min=[1.12716,-0.673166,-0.135153], max=[2.06779,1.13592,1.24793]]
04:51:55 INFO     Loading network config from: configs\nerf\base.json
04:51:55 INFO     GridEncoding:  Nmin=16 b=1.66248 F=2 T=2^19 L=16
Warning: FullyFusedMLP is not supported for the selected architecture 61. Falling back to CutlassMLP. For maximum performance, raise the target GPU architecture to 75+.
Warning: FullyFusedMLP is not supported for the selected architecture 61. Falling back to CutlassMLP. For maximum performance, raise the target GPU architecture to 75+.
Warning: FullyFusedMLP is not supported for the selected architecture 61. Falling back to CutlassMLP. For maximum performance, raise the target GPU architecture to 75+.
04:51:55 INFO     Density model: 3--[HashGrid]-->32--[FullyFusedMLP(neurons=64,layers=3)]-->1
04:51:55 INFO     Color model:   3--[Composite]-->16+16--[FullyFusedMLP(neurons=64,layers=4)]-->3
04:51:55 INFO       total_encoding_params=13623184 total_network_params=10240

04:51:55 WARNING  Vulkan error: loader_validate_device_extensions: Device extension VK_NVX_binary_import not supported by selected physical device or enabled layers.

04:51:55 WARNING  Vulkan error: vkCreateDevice: Failed to validate extensions in list

04:51:55 WARNING  Could not initialize Vulkan and NGX. DLSS not supported. (D:\Program\Anaconda3\envs\py39\Lib\site-packages\instant-ngp\src\dlss.cu:358 vkCreateDevice(vk_physical_device, &device_create_info, nullptr, &vk_device) failed)

04:51:55 ERROR    Uncaught exception: D:\Program\Anaconda3\envs\py39\Lib\site-packages\instant-ngp\dependencies\tiny-cuda-nn\include\tiny-cuda-nn/cutlass_matmul.h:332 status failed with error Error Internal
yenchenlin commented 1 year ago

May I know what's your goal here?

1) Solve Vulkan error to enable DLSS 2) Solve the Uncaught exception which causes the program to crash

If the goal is 1), you need to download the Vulkan SDK here and install it. Importantly, make sure your CUDA driver is up-to-date (>= 520) otherwise it will not be detected when you re-build instant-ngp.

565ee commented 1 year ago

May I know what's your goal here?

  1. Solve Vulkan error to enable DLSS
  2. Solve the Uncaught exception which causes the program to crash

If the goal is 1), you need to download the Vulkan SDK here and install it. Importantly, make sure your CUDA driver is up-to-date (>= 520) otherwise it will not be detected when you re-build instant-ngp.

i had install
VulkanSDK 1.3.231.1
nvdia Graphics driver 526.98

Dandelionym commented 1 year ago

Don't worry about that. It's worth mentioning that this is just a warning when you do not successfully export Nvidia OptiX in ~/.bashrc. Check your env first, please. :-)

xiehousen commented 1 year ago

s worth mentioning that this is just a warning

Hi @Dandelionym , when I build this project by cmake , the output log shows system found the Vulkan_path. However, in actual use, the log shows as follows : Could not initialize Vulkan and NGX. DLSS not supported. Thanks a lot for your reply.