Releases: SciML/DiffEqGPU.jl
Releases · SciML/DiffEqGPU.jl
v2.4.0
DiffEqGPU v2.4.0
Closed issues:
EnsembleGPUArray
compatibility with different time intervals (#283)- Error when running example code (#289)
Merged pull requests:
- Add support for different tspans in EnsembleGPUArray (#285) (@utkarsh530)
- fix doc + AD tests (#286) (@utkarsh530)
- Allow large problems to work with stiff ODE solvers in
EnsembleGPUKernel
(#287) (@utkarsh530)
v2.3.1
DiffEqGPU v2.3.1
Merged pull requests:
- Update to ForwardDiff support in EnsembleGPUKernel (#278) (@utkarsh530)
- CompatHelper: bump compat for SafeTestsets to 0.1 for package docs, (keep existing compat) (#282) (@github-actions[bot])
- Update DiffEqGPU.jl for LinearSolve v2 (#284) (@ChrisRackauckas)
v2.3.0
DiffEqGPU v2.3.0
Closed issues:
- RuntimeGeneratedFunctions are not compatible with CUDA.jl (ModelingToolkit.jl generated functions) (#169)
Merged pull requests:
- Add Rodas5P method for EnsembleGPUKernel (#275) (@utkarsh530)
- Add Kvaerno methods for EnsembleGPUKernel (#276) (@utkarsh530)
- CompatHelper: add new compat entry for Setfield at version 1, (keep existing compat) (#277) (@github-actions[bot])
- Fix docstring examples by adding
using CUDA
(#279) (@ChrisRackauckas) - WIP: Mass matrix support for rosenbrock23 (#280) (@tshort)
- allow linearsolve v2 (#281) (@ChrisRackauckas)
v2.2.1
v2.2.0
v2.1.1
DiffEqGPU v2.1.1
Closed issues:
- EnsembleMPGOS (#77)
- Recent benchmarks of Kernelized GPU ODE solvers with MPGOS (#187)
- Example in docs fails (#243)
- Support non-array parameter types (#248)
- Problem with EnsembleGPUArray() when updating library (#254)
Merged pull requests:
- Add Rosenbrock23 with EnsembleGPUKernel (#252) (@utkarsh530)
- Add doctests and add information about backends to the docs (#255) (@ChrisRackauckas)
- support non-array parameters (#256) (@pepijndevos)
- Update compat for KA and CUDA (#270) (@utkarsh530)
v2.1.0
DiffEqGPU v2.1.0
Closed issues:
- Add AMDGPU support (#96)
Merged pull requests:
- Add lower level API for EnsembleGPUArray (#253) (@utkarsh530)
v2.0.0
DiffEqGPU v2.0.0
Merged pull requests:
- Simple translation to KernelAbstractions (#220) (@vchuravy)
- [skip ci] LanguageTool (#223) (@ArnoStrouwen)
- Update lower API example + fix (#225) (@utkarsh530)
- Move SDE solvers to KernelAbstractions.jl (#226) (@utkarsh530)
- Generalize GPU code to more than one backend (#227) (@vchuravy)
- CompatHelper: bump compat for CUDA to 4, (keep existing compat) (#228) (@github-actions[bot])
- format markdown (#230) (@ArnoStrouwen)
- CompatHelper: add new compat entry for AMDGPU at version 0.4, (keep existing compat) (#231) (@github-actions[bot])
- CompatHelper: add new compat entry for ROCKernels at version 0.3, (keep existing compat) (#232) (@github-actions[bot])
- Test support for oneAPI (#233) (@utkarsh530)
- CompatHelper: add new compat entry for oneAPI at version 1, (keep existing compat) (#236) (@github-actions[bot])
- CompatHelper: add new compat entry for oneAPIKernels at version 0.1, (keep existing compat) (#237) (@github-actions[bot])
- Add warning if kernel runs on CPU (#238) (@utkarsh530)
- update for ArrayInterface v7 (#240) (@ChrisRackauckas)
- Add MetalKernels (#241) (@vchuravy)
- enable dependabot for GitHub actions (#244) (@ranocha)
- Bump actions/checkout from 1 to 3 (#245) (@dependabot[bot])
- destats -> stats (#246) (@ChrisRackauckas)
v1.26.0
DiffEqGPU v1.26.0
Merged pull requests:
- ci: update encrypted key for documentation builds (#222) (@thazhemadam)
v1.25.0
DiffEqGPU v1.25.0
Closed issues:
- Direct Compilation methods (#3)
- color vector handling and sparse representation (#18)
- Newton-Krylov handling (#19)
- Support terminate! (#43)
- Iteration doesn't work in RHS function (#55)
- GPU Parallel Ensemble simulation with random time span (#141)
- EnsembleGPUArray performance vs EnsembleSerial (#147)
- Scope of improvements in EnsembleGPUKernel (#171)
- Ensemble simulation v1.18 wrong solution for adaptive=false & save_everystep=false (#177)
- ContinuousCallback does not work correctly with EnsembleGPUArray() (#191)
- Just wonder how faster DiffEqGPU.jl is compared to DifferentialEquations.jl (#204)
- Add a section to the manual on how to use the lower level API for lower overhead (#205)
- Out of dynamic GPU memory in EnsembleGPUKernel for higher number of threads (#206)
Merged pull requests:
- Fix dynamic allocation overflow in GPUs for higher thread count (#207) (@utkarsh530)
- Add GPU Euler-Maruyama SDE solver (#208) (@utkarsh530)
- Build and deploy documentation using Buildkite (#209) (@thazhemadam)
- Add support for non-diagonal noise and other fixes (#210) (@utkarsh530)
- ci: update encrypted key for documentation builds (#211) (@thazhemadam)
- Add a warning to the documentation of EnsembleGPUArray for terminate (#212) (@ChrisRackauckas)
- Improve lower level documentation (#213) (@ChrisRackauckas)
- ci: update encrypted key for documentation builds (#214) (@thazhemadam)
- ci: update encrypted key for documentation builds (#215) (@thazhemadam)
- Add support for random time span within trajectories (#216) (@utkarsh530)
- EnsembleGPUKernel Weak Order 2.0 SDE solver (SIEA) (#217) (@utkarsh530)
- Fixes and updates on SDE solvers (#218) (@utkarsh530)