Open gdhy9064 opened 2 years ago
Use cfg.n_rays_per_batch instead of default n_rays_per_batch 4096 in JNeRF/python/jnerf/models/position_encoders/hash_encoder/grid_encode.py which may cause cudaErrorIllegalAddress below when cfg.n_rays_per_batch much greater than 4096.
cfg.n_rays_per_batch
n_rays_per_batch
JNeRF/python/jnerf/models/position_encoders/hash_encoder/grid_encode.py
code=700( cudaErrorIllegalAddress ) cudaMemcpy(a.ptr, var->mem_ptr, var->size, cudaMemcpyDeviceToHost)
Thanks!
Use
cfg.n_rays_per_batch
instead of defaultn_rays_per_batch
4096 inJNeRF/python/jnerf/models/position_encoders/hash_encoder/grid_encode.py
which may cause cudaErrorIllegalAddress below whencfg.n_rays_per_batch
much greater than 4096.