Some cleanup (trixi-framework#1408)

* link to SummationByPartsOperators.jl package * clarify that static_length is from StaticArrayInterface * describe options to save precompilation time * Apply suggestions from code review Co-authored-by: Michael Schlottke-Lakemper <[email protected]> --------- Co-authored-by: Michael Schlottke-Lakemper <[email protected]>
Xinyu-Li-1997 · Apr 24, 2023 · d952106 · d952106
1 parent 7b5b0bd
commit d952106
Show file tree

Hide file tree

Showing 5 changed files with 47 additions and 5 deletions.
diff --git a/Project.toml b/Project.toml
@@ -30,6 +30,7 @@ SimpleUnPack = "ce78b400-467f-4804-87d8-8f486da07d0a"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 StartUpDG = "472ebc20-7c99-4d4b-9470-8fde4e9faa0f"
 Static = "aedffcd0-7271-4cad-89d0-dc628f76c6d3"
+StaticArrayInterface = "0d7ed370-da01-4f52-bd93-41d350b8b718"
 StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
 StrideArrays = "d1fa6d79-ef01-42a6-86c9-f7c551f8593b"
 StructArrays = "09ab397b-f2b6-538f-b94a-2f83cf4a842a"
@@ -63,6 +64,7 @@ Setfield = "0.8, 1"
 SimpleUnPack = "1.1"
 StartUpDG = "0.16"
 Static = "0.3, 0.4, 0.5, 0.6, 0.7, 0.8"
+StaticArrayInterface = "1.4"
 StaticArrays = "1"
 StrideArrays = "0.1.18"
 StructArrays = "0.6"

diff --git a/README.md b/README.md
@@ -35,7 +35,7 @@ installation and postprocessing procedures. Its features include:
   * Kinetic energy-preserving and entropy-stable methods based on flux differencing
   * Entropy-stable shock capturing
   * Positivity-preserving limiting
-  * Finite difference summation by parts (SBP) methods
+  * [Finite difference summation by parts (SBP) methods](https://github.com/ranocha/SummationByPartsOperators.jl)
 * Compatible with the [SciML ecosystem for ordinary differential equations](https://diffeq.sciml.ai/latest/)
   * [Explicit low-storage Runge-Kutta time integration](https://diffeq.sciml.ai/latest/solvers/ode_solve/#Low-Storage-Methods)
   * [Strong stability preserving methods](https://diffeq.sciml.ai/latest/solvers/ode_solve/#Explicit-Strong-Stability-Preserving-Runge-Kutta-Methods-for-Hyperbolic-PDEs-(Conservation-Laws))

diff --git a/docs/src/index.md b/docs/src/index.md
@@ -29,7 +29,7 @@ installation and postprocessing procedures. Its features include:
   * Kinetic energy-preserving and entropy-stable methods based on flux differencing
   * Entropy-stable shock capturing
   * Positivity-preserving limiting
-  * Finite difference summation by parts (SBP) methods
+  * [Finite difference summation by parts (SBP) methods](https://github.com/ranocha/SummationByPartsOperators.jl)
 * Compatible with the [SciML ecosystem for ordinary differential equations](https://diffeq.sciml.ai/latest/)
   * [Explicit low-storage Runge-Kutta time integration](https://diffeq.sciml.ai/latest/solvers/ode_solve/#Low-Storage-Methods)
   * [Strong stability preserving methods](https://diffeq.sciml.ai/latest/solvers/ode_solve/#Explicit-Strong-Stability-Preserving-Runge-Kutta-Methods-for-Hyperbolic-PDEs-(Conservation-Laws))

diff --git a/docs/src/troubleshooting.md b/docs/src/troubleshooting.md
@@ -6,7 +6,7 @@ Julia packages via the package manager, e.g., by running
 ```julia
 julia> import Pkg; Pkg.update()
 ```
-If you do  not use the latest stable release of Julia from the 
+If you do not use the latest stable release of Julia from the
 [official website](https://julialang.org/downloads/#current_stable_release),
 consider updating your Julia installation.
 
@@ -148,3 +148,43 @@ are kept on the same MPI rank to be able to coarsen them easily. This might caus
 distribution of cells on different ranks. For 2D meshes, this also means that *initially* each rank will
 at least own 4 cells, and for 3D meshes, *initially* each rank will at least own 8 cells.
 See [issue #1329](https://github.com/trixi-framework/Trixi.jl/issues/1329).
+
+
+
+## Installing and updating everything takes a lot of time
+
+Julia compiles code to get good (C/Fortran-like) performance. At the same time,
+Julia provides a dynamic environment and usually compiles code just before using
+it. Over time, Julia has improved its caching infrastructure, allowing to store
+and reuse more results from (pre-)compilation. This often results in an
+increased time to install/update packages, in particular when updating
+to Julia v1.8 or v1.9 from older versions.
+
+Some packages used together with [Trixi.jl](https://github.com/trixi-framework/Trixi.jl)
+provide options to configure the amount of precompilation. For example,
+[OrdinaryDiffEq.jl](https://github.com/SciML/OrdinaryDiffEq.jl) precompiles
+many ODE solvers for a good runtime experience of average users. Currently,
+[Trixi.jl](https://github.com/trixi-framework/Trixi.jl) does not use all of
+the available solvers. Thus, you can save some time at every update by setting
+their [precompilation options](https://docs.sciml.ai/DiffEqDocs/stable/features/low_dep/).
+
+At the time of writing, this could look as follows. First, you need to activate
+the environment where you have installed
+[OrdinaryDiffEq.jl](https://github.com/SciML/OrdinaryDiffEq.jl). Then, you need
+to execute the following Julia code.
+
+```julia
+using Preferences, UUIDs
+set_preferences!(UUID("1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"), "PrecompileNonStiff" => true)
+set_preferences!(UUID("1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"), "PrecompileStiff" => false)
+set_preferences!(UUID("1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"), "PrecompileAutoSwitch" => false)
+set_preferences!(UUID("1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"), "PrecompileLowStorage" => true)
+set_preferences!(UUID("1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"), "PrecompileDefaultSpecialize" => true)
+set_preferences!(UUID("1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"), "PrecompileAutoSpecialize" => false)
+set_preferences!(UUID("1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"), "PrecompileFunctionWrapperSpecialize" => false)
+set_preferences!(UUID("1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"), "PrecompileNoSpecialize" => false)
+```
+
+This disables precompilation of all implicit methods. This should usually not affect
+the runtime latency with [Trixi.jl](https://github.com/trixi-framework/Trixi.jl)
+since most setups use explicit time integration methods.
diff --git a/src/Trixi.jl b/src/Trixi.jl
@@ -41,7 +41,7 @@ using HDF5: h5open, attributes
 using IfElse: ifelse
 using LinearMaps: LinearMap
 using LoopVectorization: LoopVectorization, @turbo, indices
-using LoopVectorization.ArrayInterface: static_length
+using StaticArrayInterface: static_length # used by LoopVectorization
 using MPI: MPI
 using MuladdMacro: @muladd
 using Octavian: Octavian, matmul!
@@ -133,7 +133,7 @@ export AcousticPerturbationEquations2D,
        LinearScalarAdvectionEquation1D, LinearScalarAdvectionEquation2D, LinearScalarAdvectionEquation3D,
        InviscidBurgersEquation1D,
        LatticeBoltzmannEquations2D, LatticeBoltzmannEquations3D,
-       ShallowWaterEquations1D, ShallowWaterEquations2D, 
+       ShallowWaterEquations1D, ShallowWaterEquations2D,
        ShallowWaterTwoLayerEquations1D, ShallowWaterTwoLayerEquations2D,
        LinearizedEulerEquations2D