Open TK0627 opened 1 year ago
hardware configuration too low
@superhandishere what would be the minimum configuration or recommended configuration.
20 images dataset which is a reasonable figure, how about the resolution of each image?
20images with a resolution of 1080x720 also tried 300x300 with aab scale 4 and 1 respectively but same error
have you tried originial fox folder of images? does that work?
No, not yet tried the fox images.
try that first, if you can't run successfully then you have to upgrade hardware, recommend rtx30x
okay i will try to render with fox image and see, Thankyou.
I Have a 1070 with 8GB of VRAM and had the same issue with my own dataset. It turns out i had to decrease the "--aabb_scale" now the custom dataset works too.
@FGay
What did you lower it to?
Normally aabb scale to 4 is optimal (cuz fox data is set so)
okay i will try to render with fox image and see, Thankyou.
Did it work? I have the same config and I couldn't get it to work.
No it is not worked for me as well.
Sent from Outlook for Androidhttps://aka.ms/AAb9ysg
From: doctorjana @.> Sent: Saturday, December 17, 2022 9:22:42 PM To: NVlabs/instant-ngp @.> Cc: Thirukumaran Sankar @.>; Author @.> Subject: Re: [NVlabs/instant-ngp] failed with error CUDA_ERROR_OUT_OF_MEMORY (Issue #1108)
okay i will try to render with fox image and see, Thankyou.
Did it work? I have the same config and I couldn't get it to work.
— Reply to this email directly, view it on GitHubhttps://github.com/NVlabs/instant-ngp/issues/1108#issuecomment-1356316022, or unsubscribehttps://github.com/notifications/unsubscribe-auth/A4CVBALX5DYFOONV3XFJA53WNXOUVANCNFSM6AAAAAASRWGOAQ. You are receiving this because you authored the thread.Message ID: @.***>
No it is not worked for me as well. Sent from Outlook for Androidhttps://aka.ms/AAb9ysg … ____ From: doctorjana @.> Sent: Saturday, December 17, 2022 9:22:42 PM To: NVlabs/instant-ngp @.> Cc: Thirukumaran Sankar @.>; Author @.> Subject: Re: [NVlabs/instant-ngp] failed with error CUDA_ERROR_OUT_OF_MEMORY (Issue #1108) okay i will try to render with fox image and see, Thankyou. Did it work? I have the same config and I couldn't get it to work. — Reply to this email directly, view it on GitHub<#1108 (comment)>, or unsubscribehttps://github.com/notifications/unsubscribe-auth/A4CVBALX5DYFOONV3XFJA53WNXOUVANCNFSM6AAAAAASRWGOAQ. You are receiving this because you authored the thread.Message ID: @.***>
Looks like a common graphic card issue. Just hope someone finds a solution.
14:16:08 SUCCESS Loaded 20 images after 0s 14:16:08 INFO cam_aabb=[min=[0.617572,-0.877056,0.500298], max=[1.81439,1.80589,0.529203]] 14:16:08 INFO Loading network config from: configs\nerf\base.json 14:16:08 INFO GridEncoding: Nmin=16 b=1.51572 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+. 14:16:08 INFO Density model: 3--[HashGrid]-->32--[FullyFusedMLP(neurons=64,layers=3)]-->1 14:16:08 INFO Color model: 3--[Composite]-->16+16--[FullyFusedMLP(neurons=64,layers=4)]-->3 14:16:08 INFO total_encoding_params=13074912 total_network_params=10240 Attempting to free memory arena while it is still in use. Could not free memory arena: C:\ngp\instant-ngp\dependencies\tiny-cuda-nn\include\tiny-cuda-nn/gpu_memory.h:489 cuMemAddressFree(m_base_address, m_max_size) failed with error CUDA_ERROR_INVALID_VALUE 14:16:08 ERROR Uncaught exception: C:\ngp\instant-ngp\dependencies\tiny-cuda-nn\include\tiny-cuda-nn/gpu_memory.h:592 cuMemSetAccess(m_base_address + m_size, n_bytes_to_allocate, &access_desc, 1) failed with error CUDA_ERROR_OUT_OF_MEMORY
kindly help on this one
GPU: