Tolmachev Dmitrii

Hello, The changes to VKFFT_BACKEND look cosmetic to me but if they can make the code more clear to others I am happy to add them. However, instead of adding...

Hello, Something is off. While it is true that at times this test gives high error values for big systems - they come from the randomness of input data, some...

Hello, So this is quite an old GPU, so I will need more results to investigate the issue. Can you run the small systems and print kernels and results? I...

There were missing floating-point suffix literals in v1.2.30 which was fixed in v1.2.31. Other than that identical code runs on modern GPUs, so I am not sure how to debug...

Hello, I believe this is the same request as https://github.com/vincefn/pyvkfft/issues/33#issuecomment-1950252081. Shouldn't be hard to add and seems to be really useful. I will try to add this functionality in the...

I have added a sample (53) that shows how the requested functionality can be implemented in the new develop version of VkFFT. I haven't fully tested it yet apart from...

> Not exactly related, but are there any requirements for dimension sizes when it comes to convolution? Yes, there is an issue that the convolution optimization has not been implemented...

> Any hint what it should be aligned to then, to not encounter such problems for now? Convolutions should be safe now for C2C sequences decomposable as a multiplication of...

> I must have missed it. So R2C convolutions aren't supported specifially for 1D case? Cause I think there's a 2D R2C convolution in sample 52. Exactly. The last axis...

Hello, Thank you for the change, it will also make the code more clear. It missed the 0x1027f00 Apple vendor ID definition and I am not sure yet whether the...