rkierulf
rkierulf
### Feature Request It's possible that the run_spin_precession!() and run_spin_excitation!() functions for the current Bloch simulation method could be made to run faster. For example, inside the main for loop...
### Feature Request KomaMRI currently uses an array-based approach when simulations are run on the GPU. It's likely that a kernel-based simulation method using KernelAbstractions.jl would run faster, since some...
### Feature Request Note: this would require writing kernel-based GPU simulation functions first, so #353 is a pre-requisite. Use of the KernelAbstractions.jl @localmem API could help with faster memory access...
### Feature Request In some cases, it may be necessary to run MRI simulations across multiple compute nodes if the problem is too large to fit in memory for a...
### Feature Request A similar approach to what Flux.jl does [here](https://github.com/FluxML/Flux.jl/tree/master/ext) could be used to extend GPU support to all backends supported by KernelAbstractions.jl. #179 looks like a good starting...
### Feature Request Part 1) Tracking for objectives in GSOC Proposal (May 29 - July 12): - [x] #351 - [x] #147 - [x] #352 - [ ] #353 Part...
### Feature Request Part 2) Tracking for objectives in GSOC proposal (July 15 - August 26): Note: the tasks below besides the first and last bullet point fall into the...
### Feature Request In some cases, it may be beneficial to run MRI simulations on multiple GPUs if the problem is too large for single-GPU memory. KomaMRI does not have...
### Feature Request Currently we are working around the lack of sincos support in oneAPI.jl. However, sincos was recently implemented in https://github.com/JuliaGPU/oneAPI.jl/commit/260a4dda0ea223dbf0893de7b4a13d994ae27bd1. Once oneAPI tags a version with this commit,...
Logical indexing of a matrix using .== doesn't work for oneAPI: ``` using oneAPI A = oneArray(rand(Float32, (1000, 1000))) A[A .== 0] .= 1.0f0 ``` ``` ERROR: LoadError: GPU compilation...