1bcb2dfaeSJed Brown# Changes/Release Notes 2bcb2dfaeSJed Brown 3bcb2dfaeSJed BrownOn this page we provide a summary of the main API changes, new features and examples 4bcb2dfaeSJed Brownfor each release of libCEED. 5bcb2dfaeSJed Brown 6bcb2dfaeSJed Brown(main)= 7bcb2dfaeSJed Brown 8bcb2dfaeSJed Brown## Current `main` branch 9bcb2dfaeSJed Brown 107e7773b5SJeremy L Thompson### Interface changes 117e7773b5SJeremy L Thompson 127e7773b5SJeremy L Thompson- Update {c:func} `CeedQFunctionGetFields` and {c:func} `CeedOperatorGetFields` to include number of fields. 13ce4822f6SJeremy L Thompson- Promote to the public API: QFunction and Operator field objects, `CeedQFunctionField` and `CeedOperatorField`, and associated getters, {c:func}`CeedQFunctionGetFields`; {c:func}`CeedQFunctionFieldGetName`; {c:func}`CeedQFunctionFieldGetSize`; {c:func}`CeedQFunctionFieldGetEvalMode`; {c:func}`CeedOperatorGetFields`; {c:func}`CeedOperatorFieldGetElemRestriction`; {c:func}`CeedOperatorFieldGetBasis`; and {c:func}`CeedOperatorFieldGetVector`. 14f04ea552SJeremy L Thompson- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added. 1570a7ffb3SJeremy L Thompson- Add {c:func} `CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients. 164db537f9SJeremy L Thompson- Promote {c:func} `CeedOperatorCheckReady`to the public API to facilitate interactive interfaces. 17*dcc1e3ecSJeremy L Thompson- Warning added when compiling OCCA backend to alert users that this backend is experimental. 187e7773b5SJeremy L Thompson 19f479eb23SJeremy L Thompson### New features 20f479eb23SJeremy L Thompson 21f479eb23SJeremy L Thompson- `CeedScalar` can now be set as `float` or `double` at compile time. 22f479eb23SJeremy L Thompson 23bcb2dfaeSJed Brown### Maintainability 24bcb2dfaeSJed Brown 25bcb2dfaeSJed Brown- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage. 26bcb2dfaeSJed Brown 27bcb2dfaeSJed Brown(v0-9)= 28bcb2dfaeSJed Brown 29bcb2dfaeSJed Brown## v0.9 (Jul 6, 2021) 30bcb2dfaeSJed Brown 31bcb2dfaeSJed Brown### Interface changes 32bcb2dfaeSJed Brown 33bcb2dfaeSJed Brown- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact. 34bcb2dfaeSJed Brown 35bcb2dfaeSJed Brown### New features 36bcb2dfaeSJed Brown 37bcb2dfaeSJed Brown- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use. 38bcb2dfaeSJed Brown- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends. 39bcb2dfaeSJed Brown- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations. 40bcb2dfaeSJed Brown- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed). 41bcb2dfaeSJed Brown 42bcb2dfaeSJed Brown### Performance improvements 43bcb2dfaeSJed Brown 44bcb2dfaeSJed Brown### Examples 45bcb2dfaeSJed Brown 46bcb2dfaeSJed Brown- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations. 47bcb2dfaeSJed Brown- Fluid mechanics example adds GPU support and improves modularity. 48bcb2dfaeSJed Brown 49bcb2dfaeSJed Brown### Deprecated backends 50bcb2dfaeSJed Brown 51bcb2dfaeSJed Brown- The `/cpu/self/tmpl` and `/cpu/self/tmpl/sub` backends have been removed. These backends were intially added to test the backend inheritance mechanism, but this mechanism is now widely used and tested in multiple backends. 52bcb2dfaeSJed Brown 53bcb2dfaeSJed Brown(v0-8)= 54bcb2dfaeSJed Brown 55bcb2dfaeSJed Brown## v0.8 (Mar 31, 2021) 56bcb2dfaeSJed Brown 57bcb2dfaeSJed Brown### Interface changes 58bcb2dfaeSJed Brown 59bcb2dfaeSJed Brown- Error handling improved to include enumerated error codes for C interface return values. 60bcb2dfaeSJed Brown- Installed headers that will follow semantic versioning were moved to {code}`include/ceed` directory. These headers have been renamed from {code}`ceed-*.h` to {code}`ceed/*.h`. Placeholder headers with the old naming schema are currently provided, but these headers will be removed in the libCEED v0.9 release. 61bcb2dfaeSJed Brown 62bcb2dfaeSJed Brown### New features 63bcb2dfaeSJed Brown 64bcb2dfaeSJed Brown- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features. 65bcb2dfaeSJed Brown- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly. 66bcb2dfaeSJed Brown- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators. 67bcb2dfaeSJed Brown 68bcb2dfaeSJed Brown### Performance improvements 69bcb2dfaeSJed Brown 70bcb2dfaeSJed Brown- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`. 71bcb2dfaeSJed Brown- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`. 72bcb2dfaeSJed Brown 73bcb2dfaeSJed Brown### Examples 74bcb2dfaeSJed Brown 75bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions. 76bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with Neo-Hookean hyperelasticity in current configuration as well as improved Neo-Hookean hyperelasticity exploring storage vs computation tradeoffs. 77bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` example updated with isentropic traveling vortex test case, an analytical solution to the Euler equations that is useful for testing boundary conditions, discretization stability, and order of accuracy. 78bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree. 79bcb2dfaeSJed Brown 80bcb2dfaeSJed Brown(v0-7)= 81bcb2dfaeSJed Brown 82bcb2dfaeSJed Brown## v0.7 (Sep 29, 2020) 83bcb2dfaeSJed Brown 84bcb2dfaeSJed Brown### Interface changes 85bcb2dfaeSJed Brown 86bcb2dfaeSJed Brown- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors. 87bcb2dfaeSJed Brown As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`. 88bcb2dfaeSJed Brown These changes improve support for mixed finite element methods. 89bcb2dfaeSJed Brown- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature. 90bcb2dfaeSJed Brown- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity. 91bcb2dfaeSJed Brown- Linear Operators can be assembled as point-block diagonal matrices with {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`, provided in row-major form in a {code}`ncomp` by {code}`ncomp` block per node. 92bcb2dfaeSJed Brown- Diagonal assemble interface changed to accept a {ref}`CeedVector` instead of a pointer to a {ref}`CeedVector` to reduce memory movement when interfacing with calling code. 93bcb2dfaeSJed Brown- Added {c:func}`CeedOperatorLinearAssembleAddDiagonal` and {c:func}`CeedOperatorLinearAssembleAddPointBlockDiagonal` for improved future integration with codes such as MFEM that compose the action of {ref}`CeedOperator`s external to libCEED. 94bcb2dfaeSJed Brown- Added {c:func}`CeedVectorTakeAray` to sync and remove libCEED read/write access to an allocated array and pass ownership of the array to the caller. 95bcb2dfaeSJed Brown This function is recommended over {c:func}`CeedVectorSyncArray` when the {code}`CeedVector` has an array owned by the caller that was set by {c:func}`CeedVectorSetArray`. 96bcb2dfaeSJed Brown- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory. 97bcb2dfaeSJed Brown- Added {c:func}`CeedOperatorMultigridLevelCreate`, {c:func}`CeedOperatorMultigridLevelCreateTensorH1`, and {c:func}`CeedOperatorMultigridLevelCreateH1` to facilitate creation of multigrid prolongation, restriction, and coarse grid operators using a common quadrature space. 98bcb2dfaeSJed Brown 99bcb2dfaeSJed Brown### New features 100bcb2dfaeSJed Brown 101bcb2dfaeSJed Brown- New HIP backend: `/gpu/hip/ref`. 102bcb2dfaeSJed Brown- CeedQFunction support for user `CUfunction`s in some backends 103bcb2dfaeSJed Brown 104bcb2dfaeSJed Brown### Performance improvements 105bcb2dfaeSJed Brown 106bcb2dfaeSJed Brown- OCCA backend rebuilt to facilitate future performance enhancements. 107bcb2dfaeSJed Brown- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`. 108bcb2dfaeSJed Brown 109bcb2dfaeSJed Brown### Examples 110bcb2dfaeSJed Brown 111bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions. 112bcb2dfaeSJed Brown 113bcb2dfaeSJed Brown### Deprecated backends 114bcb2dfaeSJed Brown 115bcb2dfaeSJed Brown- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`. 116bcb2dfaeSJed Brown 117bcb2dfaeSJed Brown(v0-6)= 118bcb2dfaeSJed Brown 119bcb2dfaeSJed Brown## v0.6 (Mar 29, 2020) 120bcb2dfaeSJed Brown 121bcb2dfaeSJed BrownlibCEED v0.6 contains numerous new features and examples, as well as expanded 122bcb2dfaeSJed Browndocumentation in [this new website](https://libceed.readthedocs.io). 123bcb2dfaeSJed Brown 124bcb2dfaeSJed Brown### New features 125bcb2dfaeSJed Brown 126bcb2dfaeSJed Brown- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly 127bcb2dfaeSJed Brown 1-1 correspondence with the C interface, plus some convenience features. For instance, 128bcb2dfaeSJed Brown data stored in the {cpp:type}`CeedVector` structure are available without copy as 129bcb2dfaeSJed Brown {py:class}`numpy.ndarray`. Short tutorials are provided in 130bcb2dfaeSJed Brown [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/). 131bcb2dfaeSJed Brown- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point, 132bcb2dfaeSJed Brown {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal 133bcb2dfaeSJed Brown ({c:func}`CeedOperatorAssembleLinearDiagonal`). These operations are useful for 134bcb2dfaeSJed Brown preconditioning ingredients and are used in the libCEED's multigrid examples. 135bcb2dfaeSJed Brown- The inverse of separable operators can be obtained using 136bcb2dfaeSJed Brown {c:func}`CeedOperatorCreateFDMElementInverse` and applied with 137bcb2dfaeSJed Brown {c:func}`CeedOperatorApply`. This is a useful preconditioning ingredient, 138bcb2dfaeSJed Brown especially for Laplacians and related operators. 139bcb2dfaeSJed Brown- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`, 140bcb2dfaeSJed Brown {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`. 141bcb2dfaeSJed Brown- Make public accessors for various attributes to facilitate writing composable code. 142bcb2dfaeSJed Brown- New backend: `/cpu/self/memcheck/serial`. 143bcb2dfaeSJed Brown- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA 144bcb2dfaeSJed Brown backends. (Single source is coming soon for OCCA backends.) 145bcb2dfaeSJed Brown- Fix some missing edge cases in CUDA backend. 146bcb2dfaeSJed Brown 147bcb2dfaeSJed Brown### Performance Improvements 148bcb2dfaeSJed Brown 149bcb2dfaeSJed Brown- MAGMA backend performance optimization and non-tensor bases. 150bcb2dfaeSJed Brown- No-copy optimization in {c:func}`CeedOperatorApply`. 151bcb2dfaeSJed Brown 152bcb2dfaeSJed Brown### Interface changes 153bcb2dfaeSJed Brown 154bcb2dfaeSJed Brown- Replace {code}`CeedElemRestrictionCreateIdentity` and 155bcb2dfaeSJed Brown {code}`CeedElemRestrictionCreateBlocked` with more flexible 156bcb2dfaeSJed Brown {c:func}`CeedElemRestrictionCreateStrided` and 157bcb2dfaeSJed Brown {c:func}`CeedElemRestrictionCreateBlockedStrided`. 158bcb2dfaeSJed Brown- Add arguments to {c:func}`CeedQFunctionCreateIdentity`. 159bcb2dfaeSJed Brown- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification 160bcb2dfaeSJed Brown with {cpp:enum}`CeedInterlaceMode`. This is now an attribute of the 161bcb2dfaeSJed Brown {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no 162bcb2dfaeSJed Brown longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and 163bcb2dfaeSJed Brown {c:func}`CeedElemRestrictionApply`. 164bcb2dfaeSJed Brown 165bcb2dfaeSJed Brown### Examples 166bcb2dfaeSJed Brown 167bcb2dfaeSJed BrownlibCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`. 168bcb2dfaeSJed BrownNotable additions include: 169bcb2dfaeSJed Brown 170bcb2dfaeSJed Brown- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of 171bcb2dfaeSJed Brown a domain in 1, 2, and 3 dimensions by applying a Laplacian. 172bcb2dfaeSJed Brown 173bcb2dfaeSJed Brown- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area 174bcb2dfaeSJed Brown of domains (like the cube and sphere) by direct integration on a surface mesh; 175bcb2dfaeSJed Brown demonstrates geometric dimension different from topological dimension. 176bcb2dfaeSJed Brown 177bcb2dfaeSJed Brown- PETSc {ref}`example-petsc-bps`: 178bcb2dfaeSJed Brown 179bcb2dfaeSJed Brown - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support. 180bcb2dfaeSJed Brown - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements 181bcb2dfaeSJed Brown and transparent CUDA support. 182bcb2dfaeSJed Brown - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`): 183bcb2dfaeSJed Brown generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric 184bcb2dfaeSJed Brown dimension different from topological dimension. 185bcb2dfaeSJed Brown 186bcb2dfaeSJed Brown- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid 187bcb2dfaeSJed Brown solver with algebraic multigrid coarse solve. 188bcb2dfaeSJed Brown 189bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly 190bcb2dfaeSJed Brown `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`), 191bcb2dfaeSJed Brown implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and 192bcb2dfaeSJed Brown quasi-2D computational domain support. 193bcb2dfaeSJed Brown 194bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for 195bcb2dfaeSJed Brown linear elasticity, small-strain hyperelasticity, and globalized finite-strain 196bcb2dfaeSJed Brown hyperelasticity using p-multigrid with algebraic multigrid coarse solve. 197bcb2dfaeSJed Brown 198bcb2dfaeSJed Brown(v0-5)= 199bcb2dfaeSJed Brown 200bcb2dfaeSJed Brown## v0.5 (Sep 18, 2019) 201bcb2dfaeSJed Brown 202bcb2dfaeSJed BrownFor this release, several improvements were made. Two new CUDA backends were added to 203bcb2dfaeSJed Brownthe family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art 204bcb2dfaeSJed Brownperformance using single-source {ref}`CeedQFunction`. From this release, users 205bcb2dfaeSJed Browncan define Q-Functions in a single source code independently of the targeted backend 206bcb2dfaeSJed Brownwith the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU 207bcb2dfaeSJed Browncompilation of the user provided {ref}`CeedQFunction` code. To allow a unified 208bcb2dfaeSJed Browndeclaration, the {ref}`CeedQFunction` API has undergone a slight change: 209bcb2dfaeSJed Brownthe `QFunctionField` parameter `ncomp` has been changed to `size`. This change 210bcb2dfaeSJed Brownrequires setting the previous value of `ncomp` to `ncomp*dim` when adding a 211bcb2dfaeSJed Brown`QFunctionField` with eval mode `CEED EVAL GRAD`. 212bcb2dfaeSJed Brown 213bcb2dfaeSJed BrownAdditionally, new CPU backends 214bcb2dfaeSJed Brownwere included in this release, such as the `/cpu/self/opt/*` backends (which are 215bcb2dfaeSJed Brownwritten in pure C and use partial **E-vectors** to improve performance) and the 216bcb2dfaeSJed Brown`/cpu/self/ref/memcheck` backend (which relies upon the 217bcb2dfaeSJed Brown[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user 218bcb2dfaeSJed Brown{ref}`CeedQFunction` have no undefined values). 219bcb2dfaeSJed BrownThis release also included various performance improvements, bug fixes, new examples, 220bcb2dfaeSJed Brownand improved tests. Among these improvements, vectorized instructions for 221bcb2dfaeSJed Brown{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD` 222bcb2dfaeSJed Browninstead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and 223bcb2dfaeSJed Brownidentity Q-Functions were introduced, and the PETSc benchmark problems were expanded 224bcb2dfaeSJed Brownto include unstructured meshes handling were. For this expansion, the prior version of 225bcb2dfaeSJed Brownthe PETSc BPs, which only included data associated with structured geometries, were 226bcb2dfaeSJed Brownrenamed `bpsraw`, and the new version of the BPs, which can handle data associated 227bcb2dfaeSJed Brownwith any unstructured geometry, were called `bps`. Additionally, other benchmark 228bcb2dfaeSJed Brownproblems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively), 229bcb2dfaeSJed Brownand BP5 and BP6 (the collocated versions---for which the quadrature points are the same 230bcb2dfaeSJed Brownas the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc 231bcb2dfaeSJed Brownexamples. Furthermoew, another standalone libCEED example, called `ex2`, which 232bcb2dfaeSJed Browncomputes the surface area of a given mesh was added to this release. 233bcb2dfaeSJed Brown 234bcb2dfaeSJed BrownBackends available in this release: 235bcb2dfaeSJed Brown 23668e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 23768e843eeSJed Brown|--------------------------|-----------------------------------------------------| 23868e843eeSJed Brown| `/cpu/self/ref/serial` | Serial reference implementation | 23968e843eeSJed Brown| `/cpu/self/ref/blocked` | Blocked reference implementation | 24068e843eeSJed Brown| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks | 24168e843eeSJed Brown| `/cpu/self/opt/serial` | Serial optimized C implementation | 24268e843eeSJed Brown| `/cpu/self/opt/blocked` | Blocked optimized C implementation | 24368e843eeSJed Brown| `/cpu/self/avx/serial` | Serial AVX implementation | 24468e843eeSJed Brown| `/cpu/self/avx/blocked` | Blocked AVX implementation | 24568e843eeSJed Brown| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 24668e843eeSJed Brown| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 24768e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 24868e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 24968e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 25068e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 25168e843eeSJed Brown| `/gpu/cuda/ref` | Reference pure CUDA kernels | 25268e843eeSJed Brown| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 25368e843eeSJed Brown| `/gpu/cuda/shared` | Optimized pure CUDA kernels using shared memory | 25468e843eeSJed Brown| `/gpu/cuda/gen` | Optimized pure CUDA kernels using code generation | 25568e843eeSJed Brown| `/gpu/magma` | CUDA MAGMA kernels | 256bcb2dfaeSJed Brown 257bcb2dfaeSJed BrownExamples available in this release: 258bcb2dfaeSJed Brown 25968e843eeSJed Brown:::{list-table} 26068e843eeSJed Brown:header-rows: 1 26168e843eeSJed Brown:widths: auto 26268e843eeSJed Brown* - User code 26368e843eeSJed Brown - Example 26468e843eeSJed Brown* - `ceed` 26568e843eeSJed Brown - * ex1 (volume) 26668e843eeSJed Brown * ex2 (surface) 26768e843eeSJed Brown* - `mfem` 26868e843eeSJed Brown - * BP1 (scalar mass operator) 26968e843eeSJed Brown * BP3 (scalar Laplace operator) 27068e843eeSJed Brown* - `petsc` 27168e843eeSJed Brown - * BP1 (scalar mass operator) 27268e843eeSJed Brown * BP2 (vector mass operator) 27368e843eeSJed Brown * BP3 (scalar Laplace operator) 27468e843eeSJed Brown * BP4 (vector Laplace operator) 27568e843eeSJed Brown * BP5 (collocated scalar Laplace operator) 27668e843eeSJed Brown * BP6 (collocated vector Laplace operator) 27768e843eeSJed Brown * Navier-Stokes 27868e843eeSJed Brown* - `nek5000` 27968e843eeSJed Brown - * BP1 (scalar mass operator) 28068e843eeSJed Brown * BP3 (scalar Laplace operator) 28168e843eeSJed Brown::: 282bcb2dfaeSJed Brown 283bcb2dfaeSJed Brown(v0-4)= 284bcb2dfaeSJed Brown 285bcb2dfaeSJed Brown## v0.4 (Apr 1, 2019) 286bcb2dfaeSJed Brown 287bcb2dfaeSJed BrownlibCEED v0.4 was made again publicly available in the second full CEED software 288bcb2dfaeSJed Browndistribution, release CEED 2.0. This release contained notable features, such as 289bcb2dfaeSJed Brownfour new CPU backends, two new GPU backends, CPU backend optimizations, initial 290bcb2dfaeSJed Brownsupport for operator composition, performance benchmarking, and a Navier-Stokes demo. 291bcb2dfaeSJed BrownThe new CPU backends in this release came in two families. The `/cpu/self/*/serial` 292bcb2dfaeSJed Brownbackends process one element at a time and are intended for meshes with a smaller number 293bcb2dfaeSJed Brownof high order elements. The `/cpu/self/*/blocked` backends process blocked batches of 294bcb2dfaeSJed Browneight interlaced elements and are intended for meshes with higher numbers of elements. 295bcb2dfaeSJed BrownThe `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU 296bcb2dfaeSJed Brownperformance. The `/cpu/self/xsmm/*` backends rely upon the 297bcb2dfaeSJed Brown[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU 298bcb2dfaeSJed Brownperformance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA. 299bcb2dfaeSJed BrownThe `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable 300bcb2dfaeSJed Brownperformance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple 301bcb2dfaeSJed Brownparallelization approach, where each thread treats a finite element. Using just in time 302bcb2dfaeSJed Browncompilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this 303bcb2dfaeSJed Brownbackend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend 304bcb2dfaeSJed Brownachieve good peak performance for 1D, 2D, and low order 3D problems, but performance 305bcb2dfaeSJed Browndeteriorates very quickly when threads run out of registers. 306bcb2dfaeSJed Brown 307bcb2dfaeSJed BrownA new explicit time-stepping Navier-Stokes solver was added to the family of libCEED 308bcb2dfaeSJed Brownexamples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`). 309bcb2dfaeSJed BrownThis example solves the time-dependent Navier-Stokes equations of compressible gas 310bcb2dfaeSJed Browndynamics in a static Eulerian three-dimensional frame, using structured high-order 311bcb2dfaeSJed Brownfinite/spectral element spatial discretizations and explicit high-order time-stepping 312bcb2dfaeSJed Brown(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc, 313bcb2dfaeSJed Brownso that the pointwise physics (defined at quadrature points) is separated from the 314bcb2dfaeSJed Brownparallelization and meshing concerns. 315bcb2dfaeSJed Brown 316bcb2dfaeSJed BrownBackends available in this release: 317bcb2dfaeSJed Brown 31868e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 31968e843eeSJed Brown|--------------------------|-----------------------------------------------------| 32068e843eeSJed Brown| `/cpu/self/ref/serial` | Serial reference implementation | 32168e843eeSJed Brown| `/cpu/self/ref/blocked` | Blocked reference implementation | 32268e843eeSJed Brown| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 32368e843eeSJed Brown| `/cpu/self/avx/serial` | Serial AVX implementation | 32468e843eeSJed Brown| `/cpu/self/avx/blocked` | Blocked AVX implementation | 32568e843eeSJed Brown| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 32668e843eeSJed Brown| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 32768e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 32868e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 32968e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 33068e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 33168e843eeSJed Brown| `/gpu/cuda/ref` | Reference pure CUDA kernels | 33268e843eeSJed Brown| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 33368e843eeSJed Brown| `/gpu/magma` | CUDA MAGMA kernels | 334bcb2dfaeSJed Brown 335bcb2dfaeSJed BrownExamples available in this release: 336bcb2dfaeSJed Brown 33768e843eeSJed Brown:::{list-table} 33868e843eeSJed Brown:header-rows: 1 33968e843eeSJed Brown:widths: auto 34068e843eeSJed Brown* - User code 34168e843eeSJed Brown - Example 34268e843eeSJed Brown* - `ceed` 34368e843eeSJed Brown - * ex1 (volume) 34468e843eeSJed Brown* - `mfem` 34568e843eeSJed Brown - * BP1 (scalar mass operator) 34668e843eeSJed Brown * BP3 (scalar Laplace operator) 34768e843eeSJed Brown* - `petsc` 34868e843eeSJed Brown - * BP1 (scalar mass operator) 34968e843eeSJed Brown * BP3 (scalar Laplace operator) 35068e843eeSJed Brown * Navier-Stokes 35168e843eeSJed Brown* - `nek5000` 35268e843eeSJed Brown - * BP1 (scalar mass operator) 35368e843eeSJed Brown * BP3 (scalar Laplace operator) 35468e843eeSJed Brown::: 355bcb2dfaeSJed Brown 356bcb2dfaeSJed Brown(v0-3)= 357bcb2dfaeSJed Brown 358bcb2dfaeSJed Brown## v0.3 (Sep 30, 2018) 359bcb2dfaeSJed Brown 360bcb2dfaeSJed BrownNotable features in this release include active/passive field interface, support for 361bcb2dfaeSJed Brownnon-tensor bases, backend optimization, and improved Fortran interface. This release 362bcb2dfaeSJed Brownalso focused on providing improved continuous integration, and many new tests with code 363bcb2dfaeSJed Browncoverage reports of about 90%. This release also provided a significant change to the 364bcb2dfaeSJed Brownpublic interface: a {ref}`CeedQFunction` can take any number of named input and output 365bcb2dfaeSJed Brownarguments while {ref}`CeedOperator` connects them to the actual data, which may be 366bcb2dfaeSJed Brownsupplied explicitly to `CeedOperatorApply()` (active) or separately via 367bcb2dfaeSJed Brown`CeedOperatorSetField()` (passive). This interface change enables reusable libraries 368bcb2dfaeSJed Brownof CeedQFunctions and composition of block solvers constructed using 369bcb2dfaeSJed Brown{ref}`CeedOperator`. A concept of blocked restriction was added to this release and 370bcb2dfaeSJed Brownused in an optimized CPU backend. Although this is typically not visible to the user, 371bcb2dfaeSJed Brownit enables effective use of arbitrary-length SIMD while maintaining cache locality. 372bcb2dfaeSJed BrownThis CPU backend also implements an algebraic factorization of tensor product gradients 373bcb2dfaeSJed Brownto perform fewer operations than standard application of interpolation and 374bcb2dfaeSJed Browndifferentiation from nodes to quadrature points. This algebraic formulation 375bcb2dfaeSJed Brownautomatically supports non-polynomial and non-interpolatory bases, thus is more general 376bcb2dfaeSJed Brownthan the more common derivation in terms of Lagrange polynomials on the quadrature points. 377bcb2dfaeSJed Brown 378bcb2dfaeSJed BrownBackends available in this release: 379bcb2dfaeSJed Brown 38068e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 38168e843eeSJed Brown|-------------------------|-----------------------------------------------------| 38268e843eeSJed Brown| `/cpu/self/blocked` | Blocked reference implementation | 38368e843eeSJed Brown| `/cpu/self/ref` | Serial reference implementation | 38468e843eeSJed Brown| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 38568e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 38668e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 38768e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 38868e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 38968e843eeSJed Brown| `/gpu/magma` | CUDA MAGMA kernels | 390bcb2dfaeSJed Brown 391bcb2dfaeSJed BrownExamples available in this release: 392bcb2dfaeSJed Brown 39368e843eeSJed Brown:::{list-table} 39468e843eeSJed Brown:header-rows: 1 39568e843eeSJed Brown:widths: auto 39668e843eeSJed Brown* - User code 39768e843eeSJed Brown - Example 39868e843eeSJed Brown* - `ceed` 39968e843eeSJed Brown - * ex1 (volume) 40068e843eeSJed Brown* - `mfem` 40168e843eeSJed Brown - * BP1 (scalar mass operator) 40268e843eeSJed Brown * BP3 (scalar Laplace operator) 40368e843eeSJed Brown* - `petsc` 40468e843eeSJed Brown - * BP1 (scalar mass operator) 40568e843eeSJed Brown * BP3 (scalar Laplace operator) 40668e843eeSJed Brown* - `nek5000` 40768e843eeSJed Brown - * BP1 (scalar mass operator) 40868e843eeSJed Brown * BP3 (scalar Laplace operator) 40968e843eeSJed Brown::: 410bcb2dfaeSJed Brown 411bcb2dfaeSJed Brown(v0-21)= 412bcb2dfaeSJed Brown 413bcb2dfaeSJed Brown## v0.21 (Sep 30, 2018) 414bcb2dfaeSJed Brown 415bcb2dfaeSJed BrownA MAGMA backend (which relies upon the 416bcb2dfaeSJed Brown[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this 417bcb2dfaeSJed Brownrelease. This initial integration set up the framework of using MAGMA and provided the 418bcb2dfaeSJed BrownlibCEED functionality through MAGMA kernels as one of libCEED’s computational backends. 419bcb2dfaeSJed BrownAs any other backend, the MAGMA backend provides extended basic data structures for 420bcb2dfaeSJed Brown{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements 421bcb2dfaeSJed Brownthe fundamental CEED building blocks to work with the new data structures. 422bcb2dfaeSJed BrownIn general, the MAGMA-specific data structures keep the libCEED pointers to CPU data 423bcb2dfaeSJed Brownbut also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled 424bcb2dfaeSJed Browninternally, and thus seamlessly to the user, through the functions/methods that are 425bcb2dfaeSJed Brownprovided to support them. 426bcb2dfaeSJed Brown 427bcb2dfaeSJed BrownBackends available in this release: 428bcb2dfaeSJed Brown 42968e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 43068e843eeSJed Brown|-------------------------|---------------------------------| 43168e843eeSJed Brown| `/cpu/self` | Serial reference implementation | 43268e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 43368e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 43468e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 43568e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 43668e843eeSJed Brown| `/gpu/magma` | CUDA MAGMA kernels | 437bcb2dfaeSJed Brown 438bcb2dfaeSJed BrownExamples available in this release: 439bcb2dfaeSJed Brown 44068e843eeSJed Brown:::{list-table} 44168e843eeSJed Brown:header-rows: 1 44268e843eeSJed Brown:widths: auto 44368e843eeSJed Brown* - User code 44468e843eeSJed Brown - Example 44568e843eeSJed Brown* - `ceed` 44668e843eeSJed Brown - * ex1 (volume) 44768e843eeSJed Brown* - `mfem` 44868e843eeSJed Brown - * BP1 (scalar mass operator) 44968e843eeSJed Brown * BP3 (scalar Laplace operator) 45068e843eeSJed Brown* - `petsc` 45168e843eeSJed Brown - * BP1 (scalar mass operator) 45268e843eeSJed Brown* - `nek5000` 45368e843eeSJed Brown - * BP1 (scalar mass operator) 45468e843eeSJed Brown::: 455bcb2dfaeSJed Brown 456bcb2dfaeSJed Brown(v0-2)= 457bcb2dfaeSJed Brown 458bcb2dfaeSJed Brown## v0.2 (Mar 30, 2018) 459bcb2dfaeSJed Brown 460bcb2dfaeSJed BrownlibCEED was made publicly available the first full CEED software distribution, release 461bcb2dfaeSJed BrownCEED 1.0. The distribution was made available using the Spack package manager to provide 462bcb2dfaeSJed Browna common, easy-to-use build environment, where the user can build the CEED distribution 463bcb2dfaeSJed Brownwith all dependencies. This release included a new Fortran interface for the library. 464bcb2dfaeSJed BrownThis release also contained major improvements in the OCCA backend (including a new 465bcb2dfaeSJed Brown`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to 466bcb2dfaeSJed Browncompute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an 467bcb2dfaeSJed Brown`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application 468bcb2dfaeSJed Brownof the Laplace operator) was also added to this release. 469bcb2dfaeSJed Brown 470bcb2dfaeSJed BrownBackends available in this release: 471bcb2dfaeSJed Brown 47268e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 47368e843eeSJed Brown|-------------------------|---------------------------------| 47468e843eeSJed Brown| `/cpu/self` | Serial reference implementation | 47568e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 47668e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 47768e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 47868e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 479bcb2dfaeSJed Brown 480bcb2dfaeSJed BrownExamples available in this release: 481bcb2dfaeSJed Brown 48268e843eeSJed Brown:::{list-table} 48368e843eeSJed Brown:header-rows: 1 48468e843eeSJed Brown:widths: auto 48568e843eeSJed Brown* - User code 48668e843eeSJed Brown - Example 48768e843eeSJed Brown* - `ceed` 48868e843eeSJed Brown - * ex1 (volume) 48968e843eeSJed Brown* - `mfem` 49068e843eeSJed Brown - * BP1 (scalar mass operator) 49168e843eeSJed Brown * BP3 (scalar Laplace operator) 49268e843eeSJed Brown* - `petsc` 49368e843eeSJed Brown - * BP1 (scalar mass operator) 49468e843eeSJed Brown* - `nek5000` 49568e843eeSJed Brown - * BP1 (scalar mass operator) 49668e843eeSJed Brown::: 497bcb2dfaeSJed Brown 498bcb2dfaeSJed Brown(v0-1)= 499bcb2dfaeSJed Brown 500bcb2dfaeSJed Brown## v0.1 (Jan 3, 2018) 501bcb2dfaeSJed Brown 502bcb2dfaeSJed BrownInitial low-level API of the CEED project. The low-level API provides a set of Finite 503bcb2dfaeSJed BrownElements kernels and components for writing new low-level kernels. Examples include: 504bcb2dfaeSJed Brownvector and sparse linear algebra, element matrix assembly over a batch of elements, 505bcb2dfaeSJed Brownpartial assembly and action for efficient high-order operators like mass, diffusion, 506bcb2dfaeSJed Brownadvection, etc. The main goal of the low-level API is to establish the basis for the 507bcb2dfaeSJed Brownhigh-level API. Also, identifying such low-level kernels and providing a reference 508bcb2dfaeSJed Brownimplementation for them serves as the basis for specialized backend implementations. 509bcb2dfaeSJed BrownThis release contained several backends: `/cpu/self`, and backends which rely upon the 510bcb2dfaeSJed Brown[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`, 511bcb2dfaeSJed Brown`/gpu/occa`, and `/omp/occa`. 512bcb2dfaeSJed BrownIt also included several examples, in the `examples` folder: 513bcb2dfaeSJed BrownA standalone code that shows the usage of libCEED (with no external 514bcb2dfaeSJed Browndependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1 515bcb2dfaeSJed Brown(with the application of the mass operator); and a `petsc` example to perform BP1 516bcb2dfaeSJed Brown(with the application of the mass operator). 517bcb2dfaeSJed Brown 518bcb2dfaeSJed BrownBackends available in this release: 519bcb2dfaeSJed Brown 52068e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 52168e843eeSJed Brown|-------------------------|---------------------------------| 52268e843eeSJed Brown| `/cpu/self` | Serial reference implementation | 52368e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 52468e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 52568e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 526bcb2dfaeSJed Brown 527bcb2dfaeSJed BrownExamples available in this release: 528bcb2dfaeSJed Brown 529bcb2dfaeSJed Brown| User code | Example | 53068e843eeSJed Brown|-----------------------|-----------------------------------| 53168e843eeSJed Brown| `ceed` | ex1 (scalar Laplace operator) | 53268e843eeSJed Brown| `mfem` | BP1 (scalar mass operator) | 53368e843eeSJed Brown| `petsc` | BP1 (scalar mass operator) | 534bcb2dfaeSJed Brown``` 535