3D convolution non contiguous for 256x256x256

gtm2122 · March 27, 2019, 1:51pm

Hi, I am trying to do 3D convolution on a 256x256x256 volume, I noticed that this error happens on only 256x256x256 and not below that. I wanted to know if there is any workaround ? I have called the .contiguous() method for all the tensors but still it gives me a cuDNN error: CUDNN_STATUS_EXECUTION_FAILED

below link has the minimal example where I reproduce the bug

github.com

gtm2122/TAVR/blob/master/try.py

import numpy as np
import os
import matplotlib.pyplot as plt
import torch
from torch.utils.data import DataLoader
from torchvision import transforms
from PIL import Image
from models import *

import torch.optim as optim

import torch.nn.functional as F
import matplotlib.pyplot as plt

import pickle
 

def conv_block_3d(in_dim,out_dim,act_fn):
    model = nn.Sequential(
        nn.Conv3d(in_dim,out_dim, kernel_size=3, stride=1, padding=1),

This file has been truncated. show original

Can someone please suggest a workaround ?