Michał Zientkiewicz

We have found the root cause; a fix will be available shortly.

Hello @proevgenii This is orders of magnitude faster approach (I think it's actually zero-copy). ```Python np.frombuffer(img, dtype=np.uint8) ``` where `img` is your `bytes` object.

@romanmaznikov1 Paste has very limited functionality. It's better to use `crop` or `slice` operators with a padding option (`out_of_bounds_policy="pad"`).

@5had3z Thanks for pointing this out. Some of the types were probably overlooked and we can take a look at what we can do. Some others were trimmed to limit...

The tweaks have been used to establish optimum settings. The adjustment is merged as #5382 .

@tadejsv Thanks for reporting. It looks indeed like an omission. We'll look into that.

If you're already rotating the images, you can pass the size explicitly to `fn.rotate` - you can make it fill the borders with a constant value (monochrome!) or replicate the...

Also, if you're fine with bilinear resizing without antialiasing, then you can do all those transforms in one go with `fn.warp_affine`: ```Python import nvidia.dali as dali import nvidia.dali.fn as fn...

Hello @SuperJarvis > Additionally, with persistent_workers=True in the dataloader, the speed in the second epoch noticeably increased, while DALI seems to have consistent runtime for each epoch This indicates that...

@sirosen Bumping up the issue. We introduced timing into our CI so we can see how long it takes for given point in the tests to be reached (this way...