ATS-GPU Change Log
This file contains all important changes to ATS-GPU
[4.1.0] - 2020-3-4
- API to query compute capability.
- API call to return libraries version number.
- ATS-CUDA-OCT requires an AlazarTech board in order to be used.
- ATS-CUDA-NUFFT low-level library to perform non-uniform FFTs.
- ATS-GPU-NUFFT high-level library to perform non-uniform FFTs using various
- ats-gpu-oct-sample.vi generates window function.
- MATLAB and Python code samples support complex window functions.
- Prevent ATS_GPU_OCT_GetBuffer() from doing a stream synchronize by default.
- Renamed input data types enum.
[4.0.1] - 2019-4-23
- ATS_GPU_OCT result buffers take number of outputs into account
- AcqToGPU_OCT_CUDA sample code uses bytesPerDmaBuffer for buffer posting and allocation.
- NVCC options supports all GPU architectures.
[4.0.0] - 2019-1-22
- Separate GPU processing stages, and make them available through new ATS-CUDA
and ATS-CUDA-OCT interfaces.
- Support for unpacking and de-interleaving.
- Support GPUs with compute capabilities 7.5
- ATS-GPU library built with CUDA toolkit v10.0
- GPUs with compute capability <3.0 due to CUDA toolkit 10.0 update.
[3.7.0] - 2018-8-22
- Modification in ATS-GPU-BASE unpacking and de-interleaving kernel to correct
for inaccurate number of blocks.
- Lock-in amplification library.
[3.6.0] - 2018-4-23
- Modification in Acquisition thread to have the ability to pass non-infinite
values of records per acquisition to `ATS_GPU_Setup()`
- Support for deinterleaving and unpacking digitizer data on the GPU prior to
delivering data buffers to users.
[3.5.0] - 2017-6-29
- Completely reworked ATS-GPU with a new API
- ATS-GPU 3.0.0 and earlier interface
[3.0.0] - 2017-4-13
- FFT processing library `ATS_GPU_FFT_CUDA`.
- Averaging option in `ATS_GPU_DLL_CUDA`.
- Example codes in C++, MATLAB, Python and LabVIEW.
- Directory architecture since OpenCL is deprecated.
[2.0.0] - 2016-11-4
- Support for CUDA with a sample inverting kernel and example program.
- Support for Linux with CMake for the CUDA library and example program.
- Directory architecture to improve clarity between OpenCL and CUDA implementations.
[1.1.0] - 2016-9-13
- support for amplitude, phase, real and imaginary values output.
- kernel code changed for easier readability.