Pradeep Garigipati

I don't think there is host memory involvement anymore from what I see in CUDA 9, 10 and 11 documentation. Minimum required version for ArrayFire is CUDA 9 and the...

What @jmont-dev said is correct about the route we will take for implementing the eigen solver in CUDA backend. For OpenCL however, it is different scenario and we have to...

@pavanky Which machine has R9 380 ? Can we test this with latest releases and see if it's an issue.

CPU Backend issue has been addressed via #2529 We are still looking into CUDA backend.

Okay, found the relevant info on CUDA documentation. linking here for future references https://docs.nvidia.com/cuda/cufft/index.html#data-layout > Note that in-place complex-to-real FFTs may overwrite arbitrary imaginary input point values when non-unit input...

@umar456 @gaika Can you please help verify if this is still happening with latest release and newer drivers. I tried it on my setup (shared below) and I don't see...

> Hmm. same thing seems to be happening on the v3.6.2 tag. I guess its a regression upstream? > > ``` > TEST(FFT, issue) { > > af::info(); > array...

Ran those commented out tests and they are green as of today. ``` [ RUN ] Inputs2D/FFTC2R2D.Complex32ToRealInputsPreserved_CommentedTestsNowVerifyingWithCUDA102/d0_513_d1_512_d2_1_d3_1_even_norm_0_5 [ OK ] Inputs2D/FFTC2R2D.Complex32ToRealInputsPreserved_CommentedTestsNowVerifyingWithCUDA102/d0_513_d1_512_d2_1_d3_1_even_norm_0_5 (260 ms) [ RUN ] Inputs2D/FFTC2R2D.Complex32ToRealInputsPreserved_CommentedTestsNowVerifyingWithCUDA102/d0_1025_d1_1024_d2_1_d3_1_even_norm_0_5 [ OK ]...

@syurkevi Can you verify if [these](https://github.com/arrayfire/arrayfire/blob/master/test/fft.cpp#L790) tests are failing or passing with latest CUDA toolkit + driver combo. I will verify this on my Windows box today.

@syurkevi Tested on the following devices/toolkit combo and the commented tests pass. ``` +-----------------------------------------------------------------------------+ | NVIDIA-SMI 440.33.01 Driver Version: 440.33.01 CUDA Version: 10.2 | |-------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id...