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. 17dcc1e3ecSJeremy L Thompson- Warning added when compiling OCCA backend to alert users that this backend is experimental. 189a1d3511SJeremy L Thompson- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`. 1943e1b16fSJeremy L Thompson- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name. 209c774eddSJeremy L Thompson- Clarify documentation for {c:func}`CeedVectorTakeArray`; this function will error if {c:func}`CeedVectorSetArray` with `copy_mode == CEED_USE_POINTER` was not previously called for the corresponding `CeedMemType`. 219c774eddSJeremy L Thompson- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`. 22c38440baSJed Brown- Added {c:func}`CeedQFunctionContextRegisterDouble` and {c:func}`CeedQFunctionContextRegisterInt32` with {c:func}`CeedQFunctionContextSetDouble` and {c:func}`CeedQFunctionContextSetInt32` to facilitate easy updating of {c:struct}`CeedQFunctionContext` data by user defined field names. 23cdf32b93SJeremy L Thompson- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`. 247a06ec9fSJeremy L Thompson- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums. 25f4f98f9dSJeremy L Thompson- Added type `CeedSize` equivalent to `ptrdiff_t` for array sizes in {c:func}`CeedVectorCreate`, {c:func}`CeedVectorGetLength`, `CeedElemRestrictionCreate*`, {c:func}`CeedElemRestrictionGetLVectorSize`, and {c:func}`CeedOperatorLinearAssembleSymbolic`. This is a breaking change. 268b919e6bSJeremy L Thompson- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy. 27*c9366a6bSJeremy L Thompson- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator. 287e7773b5SJeremy L Thompson 29f479eb23SJeremy L Thompson### New features 30f479eb23SJeremy L Thompson 31f479eb23SJeremy L Thompson- `CeedScalar` can now be set as `float` or `double` at compile time. 3230601ac0SJeremy L Thompson- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends. 33fb3c7d02SJeremy L Thompson- Added support for JiT of QFunctions with `#include "relative/path/local-file.h"` statements for additional local files. Note that files included with `""` are searched relative to the current file first, then by compiler paths (as with `<>` includes). To use this feature, one should adhere to relative paths only, not compiler flags like `-I`, which the JiT will not be aware of. 3423dfbf5bSJeremy L Thompson- Remove need to guard library headers in QFunction source for code generation backends. 353f21f6b1SJeremy L Thompson- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present. 36f7e22acaSJeremy L Thompson- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally. 373451974fSJeremy L Thompson- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file. 387a06ec9fSJeremy L Thompson- Added support for non-tensor H(div) elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor. 39d34e270fSJeremy L Thompson- Added {c:func}`CeedQFunctionSetContextWritable` and read-only access to `CeedQFunctionContext` data as an optional feature to improve GPU performance. By default, calling the `CeedQFunctionUser` during {c:func}`CeedQFunctionApply` is assumed to write into the `CeedQFunctionContext` data, consistent with the previous behavior. Note that if a user asserts that their `CeedQFunctionUser` does not write into the `CeedQFunctionContext` data, they are responsible for the validity of this assertion. 40f479eb23SJeremy L Thompson 41bcb2dfaeSJed Brown### Maintainability 42bcb2dfaeSJed Brown 43bcb2dfaeSJed Brown- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage. 44db52d626SJeremy L Thompson- `Include-what-you-use` makefile target added as `make iwyu`. 45bf4cb664SJeremy L Thompson- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase. 463451974fSJeremy L Thompson- Put GPU JiTed kernel source code into separate files. 47f9996dfdSJeremy L Thompson- Dropped legacy version support in PETSc based examples to better utilize PETSc DMPlex and Mat updates to support libCEED; current minimum PETSc version for the examples is v3.17. 48bcb2dfaeSJed Brown 49bcb2dfaeSJed Brown(v0-9)= 50bcb2dfaeSJed Brown 51bcb2dfaeSJed Brown## v0.9 (Jul 6, 2021) 52bcb2dfaeSJed Brown 53bcb2dfaeSJed Brown### Interface changes 54bcb2dfaeSJed Brown 55bcb2dfaeSJed Brown- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact. 56bcb2dfaeSJed Brown 57bcb2dfaeSJed Brown### New features 58bcb2dfaeSJed Brown 59bcb2dfaeSJed Brown- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use. 60bcb2dfaeSJed Brown- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends. 61bcb2dfaeSJed Brown- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations. 62bcb2dfaeSJed Brown- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed). 63bcb2dfaeSJed Brown 64bcb2dfaeSJed Brown### Performance improvements 65bcb2dfaeSJed Brown 66bcb2dfaeSJed Brown### Examples 67bcb2dfaeSJed Brown 68bcb2dfaeSJed Brown- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations. 69bcb2dfaeSJed Brown- Fluid mechanics example adds GPU support and improves modularity. 70bcb2dfaeSJed Brown 71bcb2dfaeSJed Brown### Deprecated backends 72bcb2dfaeSJed Brown 73bcb2dfaeSJed 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. 74bcb2dfaeSJed Brown 75bcb2dfaeSJed Brown(v0-8)= 76bcb2dfaeSJed Brown 77bcb2dfaeSJed Brown## v0.8 (Mar 31, 2021) 78bcb2dfaeSJed Brown 79bcb2dfaeSJed Brown### Interface changes 80bcb2dfaeSJed Brown 81bcb2dfaeSJed Brown- Error handling improved to include enumerated error codes for C interface return values. 82bcb2dfaeSJed 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. 83bcb2dfaeSJed Brown 84bcb2dfaeSJed Brown### New features 85bcb2dfaeSJed Brown 86bcb2dfaeSJed Brown- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features. 87bcb2dfaeSJed Brown- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly. 88bcb2dfaeSJed Brown- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators. 89bcb2dfaeSJed Brown 90bcb2dfaeSJed Brown### Performance improvements 91bcb2dfaeSJed Brown 92bcb2dfaeSJed Brown- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`. 93bcb2dfaeSJed Brown- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`. 94bcb2dfaeSJed Brown 95bcb2dfaeSJed Brown### Examples 96bcb2dfaeSJed Brown 97bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions. 98bcb2dfaeSJed 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. 99bcb2dfaeSJed 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. 100bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree. 101bcb2dfaeSJed Brown 102bcb2dfaeSJed Brown(v0-7)= 103bcb2dfaeSJed Brown 104bcb2dfaeSJed Brown## v0.7 (Sep 29, 2020) 105bcb2dfaeSJed Brown 106bcb2dfaeSJed Brown### Interface changes 107bcb2dfaeSJed Brown 108bcb2dfaeSJed Brown- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors. 109bcb2dfaeSJed 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`. 110bcb2dfaeSJed Brown These changes improve support for mixed finite element methods. 111bcb2dfaeSJed Brown- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature. 112bcb2dfaeSJed Brown- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity. 113bcb2dfaeSJed 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. 114bcb2dfaeSJed 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. 115bcb2dfaeSJed 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. 116bcb2dfaeSJed 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. 117bcb2dfaeSJed 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`. 118bcb2dfaeSJed Brown- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory. 119bcb2dfaeSJed 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. 120bcb2dfaeSJed Brown 121bcb2dfaeSJed Brown### New features 122bcb2dfaeSJed Brown 123bcb2dfaeSJed Brown- New HIP backend: `/gpu/hip/ref`. 124bcb2dfaeSJed Brown- CeedQFunction support for user `CUfunction`s in some backends 125bcb2dfaeSJed Brown 126bcb2dfaeSJed Brown### Performance improvements 127bcb2dfaeSJed Brown 128bcb2dfaeSJed Brown- OCCA backend rebuilt to facilitate future performance enhancements. 129bcb2dfaeSJed Brown- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`. 130bcb2dfaeSJed Brown 131bcb2dfaeSJed Brown### Examples 132bcb2dfaeSJed Brown 133bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions. 134bcb2dfaeSJed Brown 135bcb2dfaeSJed Brown### Deprecated backends 136bcb2dfaeSJed Brown 137bcb2dfaeSJed Brown- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`. 138bcb2dfaeSJed Brown 139bcb2dfaeSJed Brown(v0-6)= 140bcb2dfaeSJed Brown 141bcb2dfaeSJed Brown## v0.6 (Mar 29, 2020) 142bcb2dfaeSJed Brown 143bcb2dfaeSJed BrownlibCEED v0.6 contains numerous new features and examples, as well as expanded 14413964f07SJed Browndocumentation in [this new website](https://libceed.org). 145bcb2dfaeSJed Brown 146bcb2dfaeSJed Brown### New features 147bcb2dfaeSJed Brown 148bcb2dfaeSJed Brown- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly 149bcb2dfaeSJed Brown 1-1 correspondence with the C interface, plus some convenience features. For instance, 150bcb2dfaeSJed Brown data stored in the {cpp:type}`CeedVector` structure are available without copy as 151bcb2dfaeSJed Brown {py:class}`numpy.ndarray`. Short tutorials are provided in 152bcb2dfaeSJed Brown [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/). 153bcb2dfaeSJed Brown- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point, 154bcb2dfaeSJed Brown {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal 155bcb2dfaeSJed Brown ({c:func}`CeedOperatorAssembleLinearDiagonal`). These operations are useful for 156bcb2dfaeSJed Brown preconditioning ingredients and are used in the libCEED's multigrid examples. 157bcb2dfaeSJed Brown- The inverse of separable operators can be obtained using 158bcb2dfaeSJed Brown {c:func}`CeedOperatorCreateFDMElementInverse` and applied with 159bcb2dfaeSJed Brown {c:func}`CeedOperatorApply`. This is a useful preconditioning ingredient, 160bcb2dfaeSJed Brown especially for Laplacians and related operators. 161bcb2dfaeSJed Brown- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`, 162bcb2dfaeSJed Brown {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`. 163bcb2dfaeSJed Brown- Make public accessors for various attributes to facilitate writing composable code. 164bcb2dfaeSJed Brown- New backend: `/cpu/self/memcheck/serial`. 165bcb2dfaeSJed Brown- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA 166bcb2dfaeSJed Brown backends. (Single source is coming soon for OCCA backends.) 167bcb2dfaeSJed Brown- Fix some missing edge cases in CUDA backend. 168bcb2dfaeSJed Brown 169bcb2dfaeSJed Brown### Performance Improvements 170bcb2dfaeSJed Brown 171bcb2dfaeSJed Brown- MAGMA backend performance optimization and non-tensor bases. 172bcb2dfaeSJed Brown- No-copy optimization in {c:func}`CeedOperatorApply`. 173bcb2dfaeSJed Brown 174bcb2dfaeSJed Brown### Interface changes 175bcb2dfaeSJed Brown 176bcb2dfaeSJed Brown- Replace {code}`CeedElemRestrictionCreateIdentity` and 177bcb2dfaeSJed Brown {code}`CeedElemRestrictionCreateBlocked` with more flexible 178bcb2dfaeSJed Brown {c:func}`CeedElemRestrictionCreateStrided` and 179bcb2dfaeSJed Brown {c:func}`CeedElemRestrictionCreateBlockedStrided`. 180bcb2dfaeSJed Brown- Add arguments to {c:func}`CeedQFunctionCreateIdentity`. 181bcb2dfaeSJed Brown- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification 182bcb2dfaeSJed Brown with {cpp:enum}`CeedInterlaceMode`. This is now an attribute of the 183bcb2dfaeSJed Brown {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no 184bcb2dfaeSJed Brown longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and 185bcb2dfaeSJed Brown {c:func}`CeedElemRestrictionApply`. 186bcb2dfaeSJed Brown 187bcb2dfaeSJed Brown### Examples 188bcb2dfaeSJed Brown 189bcb2dfaeSJed BrownlibCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`. 190bcb2dfaeSJed BrownNotable additions include: 191bcb2dfaeSJed Brown 192bcb2dfaeSJed Brown- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of 193bcb2dfaeSJed Brown a domain in 1, 2, and 3 dimensions by applying a Laplacian. 194bcb2dfaeSJed Brown 195bcb2dfaeSJed Brown- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area 196bcb2dfaeSJed Brown of domains (like the cube and sphere) by direct integration on a surface mesh; 197bcb2dfaeSJed Brown demonstrates geometric dimension different from topological dimension. 198bcb2dfaeSJed Brown 199bcb2dfaeSJed Brown- PETSc {ref}`example-petsc-bps`: 200bcb2dfaeSJed Brown 201bcb2dfaeSJed Brown - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support. 202bcb2dfaeSJed Brown - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements 203bcb2dfaeSJed Brown and transparent CUDA support. 204bcb2dfaeSJed Brown - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`): 205bcb2dfaeSJed Brown generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric 206bcb2dfaeSJed Brown dimension different from topological dimension. 207bcb2dfaeSJed Brown 208bcb2dfaeSJed Brown- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid 209bcb2dfaeSJed Brown solver with algebraic multigrid coarse solve. 210bcb2dfaeSJed Brown 211bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly 212bcb2dfaeSJed Brown `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`), 213bcb2dfaeSJed Brown implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and 214bcb2dfaeSJed Brown quasi-2D computational domain support. 215bcb2dfaeSJed Brown 216bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for 217bcb2dfaeSJed Brown linear elasticity, small-strain hyperelasticity, and globalized finite-strain 218bcb2dfaeSJed Brown hyperelasticity using p-multigrid with algebraic multigrid coarse solve. 219bcb2dfaeSJed Brown 220bcb2dfaeSJed Brown(v0-5)= 221bcb2dfaeSJed Brown 222bcb2dfaeSJed Brown## v0.5 (Sep 18, 2019) 223bcb2dfaeSJed Brown 224bcb2dfaeSJed BrownFor this release, several improvements were made. Two new CUDA backends were added to 225bcb2dfaeSJed Brownthe family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art 226bcb2dfaeSJed Brownperformance using single-source {ref}`CeedQFunction`. From this release, users 227bcb2dfaeSJed Browncan define Q-Functions in a single source code independently of the targeted backend 228bcb2dfaeSJed Brownwith the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU 229bcb2dfaeSJed Browncompilation of the user provided {ref}`CeedQFunction` code. To allow a unified 230bcb2dfaeSJed Browndeclaration, the {ref}`CeedQFunction` API has undergone a slight change: 231bcb2dfaeSJed Brownthe `QFunctionField` parameter `ncomp` has been changed to `size`. This change 232bcb2dfaeSJed Brownrequires setting the previous value of `ncomp` to `ncomp*dim` when adding a 233bcb2dfaeSJed Brown`QFunctionField` with eval mode `CEED EVAL GRAD`. 234bcb2dfaeSJed Brown 235bcb2dfaeSJed BrownAdditionally, new CPU backends 236bcb2dfaeSJed Brownwere included in this release, such as the `/cpu/self/opt/*` backends (which are 237bcb2dfaeSJed Brownwritten in pure C and use partial **E-vectors** to improve performance) and the 238bcb2dfaeSJed Brown`/cpu/self/ref/memcheck` backend (which relies upon the 239bcb2dfaeSJed Brown[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user 240bcb2dfaeSJed Brown{ref}`CeedQFunction` have no undefined values). 241bcb2dfaeSJed BrownThis release also included various performance improvements, bug fixes, new examples, 242bcb2dfaeSJed Brownand improved tests. Among these improvements, vectorized instructions for 243bcb2dfaeSJed Brown{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD` 244bcb2dfaeSJed Browninstead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and 245bcb2dfaeSJed Brownidentity Q-Functions were introduced, and the PETSc benchmark problems were expanded 246bcb2dfaeSJed Brownto include unstructured meshes handling were. For this expansion, the prior version of 247bcb2dfaeSJed Brownthe PETSc BPs, which only included data associated with structured geometries, were 248bcb2dfaeSJed Brownrenamed `bpsraw`, and the new version of the BPs, which can handle data associated 249bcb2dfaeSJed Brownwith any unstructured geometry, were called `bps`. Additionally, other benchmark 250bcb2dfaeSJed Brownproblems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively), 251bcb2dfaeSJed Brownand BP5 and BP6 (the collocated versions---for which the quadrature points are the same 252bcb2dfaeSJed Brownas the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc 253bcb2dfaeSJed Brownexamples. Furthermoew, another standalone libCEED example, called `ex2`, which 254bcb2dfaeSJed Browncomputes the surface area of a given mesh was added to this release. 255bcb2dfaeSJed Brown 256bcb2dfaeSJed BrownBackends available in this release: 257bcb2dfaeSJed Brown 25868e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 25968e843eeSJed Brown|--------------------------|-----------------------------------------------------| 26068e843eeSJed Brown| `/cpu/self/ref/serial` | Serial reference implementation | 26168e843eeSJed Brown| `/cpu/self/ref/blocked` | Blocked reference implementation | 26268e843eeSJed Brown| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks | 26368e843eeSJed Brown| `/cpu/self/opt/serial` | Serial optimized C implementation | 26468e843eeSJed Brown| `/cpu/self/opt/blocked` | Blocked optimized C implementation | 26568e843eeSJed Brown| `/cpu/self/avx/serial` | Serial AVX implementation | 26668e843eeSJed Brown| `/cpu/self/avx/blocked` | Blocked AVX implementation | 26768e843eeSJed Brown| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 26868e843eeSJed Brown| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 26968e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 27068e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 27168e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 27268e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 27368e843eeSJed Brown| `/gpu/cuda/ref` | Reference pure CUDA kernels | 27468e843eeSJed Brown| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 27568e843eeSJed Brown| `/gpu/cuda/shared` | Optimized pure CUDA kernels using shared memory | 27668e843eeSJed Brown| `/gpu/cuda/gen` | Optimized pure CUDA kernels using code generation | 27768e843eeSJed Brown| `/gpu/magma` | CUDA MAGMA kernels | 278bcb2dfaeSJed Brown 279bcb2dfaeSJed BrownExamples available in this release: 280bcb2dfaeSJed Brown 28168e843eeSJed Brown:::{list-table} 28268e843eeSJed Brown:header-rows: 1 28368e843eeSJed Brown:widths: auto 28468e843eeSJed Brown* - User code 28568e843eeSJed Brown - Example 28668e843eeSJed Brown* - `ceed` 28768e843eeSJed Brown - * ex1 (volume) 28868e843eeSJed Brown * ex2 (surface) 28968e843eeSJed Brown* - `mfem` 29068e843eeSJed Brown - * BP1 (scalar mass operator) 29168e843eeSJed Brown * BP3 (scalar Laplace operator) 29268e843eeSJed Brown* - `petsc` 29368e843eeSJed Brown - * BP1 (scalar mass operator) 29468e843eeSJed Brown * BP2 (vector mass operator) 29568e843eeSJed Brown * BP3 (scalar Laplace operator) 29668e843eeSJed Brown * BP4 (vector Laplace operator) 29768e843eeSJed Brown * BP5 (collocated scalar Laplace operator) 29868e843eeSJed Brown * BP6 (collocated vector Laplace operator) 29968e843eeSJed Brown * Navier-Stokes 30068e843eeSJed Brown* - `nek5000` 30168e843eeSJed Brown - * BP1 (scalar mass operator) 30268e843eeSJed Brown * BP3 (scalar Laplace operator) 30368e843eeSJed Brown::: 304bcb2dfaeSJed Brown 305bcb2dfaeSJed Brown(v0-4)= 306bcb2dfaeSJed Brown 307bcb2dfaeSJed Brown## v0.4 (Apr 1, 2019) 308bcb2dfaeSJed Brown 309bcb2dfaeSJed BrownlibCEED v0.4 was made again publicly available in the second full CEED software 310bcb2dfaeSJed Browndistribution, release CEED 2.0. This release contained notable features, such as 311bcb2dfaeSJed Brownfour new CPU backends, two new GPU backends, CPU backend optimizations, initial 312bcb2dfaeSJed Brownsupport for operator composition, performance benchmarking, and a Navier-Stokes demo. 313bcb2dfaeSJed BrownThe new CPU backends in this release came in two families. The `/cpu/self/*/serial` 314bcb2dfaeSJed Brownbackends process one element at a time and are intended for meshes with a smaller number 315bcb2dfaeSJed Brownof high order elements. The `/cpu/self/*/blocked` backends process blocked batches of 316bcb2dfaeSJed Browneight interlaced elements and are intended for meshes with higher numbers of elements. 317bcb2dfaeSJed BrownThe `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU 318bcb2dfaeSJed Brownperformance. The `/cpu/self/xsmm/*` backends rely upon the 319bcb2dfaeSJed Brown[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU 320bcb2dfaeSJed Brownperformance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA. 321bcb2dfaeSJed BrownThe `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable 322bcb2dfaeSJed Brownperformance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple 323bcb2dfaeSJed Brownparallelization approach, where each thread treats a finite element. Using just in time 324bcb2dfaeSJed Browncompilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this 325bcb2dfaeSJed Brownbackend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend 326bcb2dfaeSJed Brownachieve good peak performance for 1D, 2D, and low order 3D problems, but performance 327bcb2dfaeSJed Browndeteriorates very quickly when threads run out of registers. 328bcb2dfaeSJed Brown 329bcb2dfaeSJed BrownA new explicit time-stepping Navier-Stokes solver was added to the family of libCEED 330bcb2dfaeSJed Brownexamples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`). 331bcb2dfaeSJed BrownThis example solves the time-dependent Navier-Stokes equations of compressible gas 332bcb2dfaeSJed Browndynamics in a static Eulerian three-dimensional frame, using structured high-order 333bcb2dfaeSJed Brownfinite/spectral element spatial discretizations and explicit high-order time-stepping 334bcb2dfaeSJed Brown(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc, 335bcb2dfaeSJed Brownso that the pointwise physics (defined at quadrature points) is separated from the 336bcb2dfaeSJed Brownparallelization and meshing concerns. 337bcb2dfaeSJed Brown 338bcb2dfaeSJed BrownBackends available in this release: 339bcb2dfaeSJed Brown 34068e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 34168e843eeSJed Brown|--------------------------|-----------------------------------------------------| 34268e843eeSJed Brown| `/cpu/self/ref/serial` | Serial reference implementation | 34368e843eeSJed Brown| `/cpu/self/ref/blocked` | Blocked reference implementation | 34468e843eeSJed Brown| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 34568e843eeSJed Brown| `/cpu/self/avx/serial` | Serial AVX implementation | 34668e843eeSJed Brown| `/cpu/self/avx/blocked` | Blocked AVX implementation | 34768e843eeSJed Brown| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 34868e843eeSJed Brown| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 34968e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 35068e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 35168e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 35268e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 35368e843eeSJed Brown| `/gpu/cuda/ref` | Reference pure CUDA kernels | 35468e843eeSJed Brown| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 35568e843eeSJed Brown| `/gpu/magma` | CUDA MAGMA kernels | 356bcb2dfaeSJed Brown 357bcb2dfaeSJed BrownExamples available in this release: 358bcb2dfaeSJed Brown 35968e843eeSJed Brown:::{list-table} 36068e843eeSJed Brown:header-rows: 1 36168e843eeSJed Brown:widths: auto 36268e843eeSJed Brown* - User code 36368e843eeSJed Brown - Example 36468e843eeSJed Brown* - `ceed` 36568e843eeSJed Brown - * ex1 (volume) 36668e843eeSJed Brown* - `mfem` 36768e843eeSJed Brown - * BP1 (scalar mass operator) 36868e843eeSJed Brown * BP3 (scalar Laplace operator) 36968e843eeSJed Brown* - `petsc` 37068e843eeSJed Brown - * BP1 (scalar mass operator) 37168e843eeSJed Brown * BP3 (scalar Laplace operator) 37268e843eeSJed Brown * Navier-Stokes 37368e843eeSJed Brown* - `nek5000` 37468e843eeSJed Brown - * BP1 (scalar mass operator) 37568e843eeSJed Brown * BP3 (scalar Laplace operator) 37668e843eeSJed Brown::: 377bcb2dfaeSJed Brown 378bcb2dfaeSJed Brown(v0-3)= 379bcb2dfaeSJed Brown 380bcb2dfaeSJed Brown## v0.3 (Sep 30, 2018) 381bcb2dfaeSJed Brown 382bcb2dfaeSJed BrownNotable features in this release include active/passive field interface, support for 383bcb2dfaeSJed Brownnon-tensor bases, backend optimization, and improved Fortran interface. This release 384bcb2dfaeSJed Brownalso focused on providing improved continuous integration, and many new tests with code 385bcb2dfaeSJed Browncoverage reports of about 90%. This release also provided a significant change to the 386bcb2dfaeSJed Brownpublic interface: a {ref}`CeedQFunction` can take any number of named input and output 387bcb2dfaeSJed Brownarguments while {ref}`CeedOperator` connects them to the actual data, which may be 388bcb2dfaeSJed Brownsupplied explicitly to `CeedOperatorApply()` (active) or separately via 389bcb2dfaeSJed Brown`CeedOperatorSetField()` (passive). This interface change enables reusable libraries 390bcb2dfaeSJed Brownof CeedQFunctions and composition of block solvers constructed using 391bcb2dfaeSJed Brown{ref}`CeedOperator`. A concept of blocked restriction was added to this release and 392bcb2dfaeSJed Brownused in an optimized CPU backend. Although this is typically not visible to the user, 393bcb2dfaeSJed Brownit enables effective use of arbitrary-length SIMD while maintaining cache locality. 394bcb2dfaeSJed BrownThis CPU backend also implements an algebraic factorization of tensor product gradients 395bcb2dfaeSJed Brownto perform fewer operations than standard application of interpolation and 396bcb2dfaeSJed Browndifferentiation from nodes to quadrature points. This algebraic formulation 397bcb2dfaeSJed Brownautomatically supports non-polynomial and non-interpolatory bases, thus is more general 398bcb2dfaeSJed Brownthan the more common derivation in terms of Lagrange polynomials on the quadrature points. 399bcb2dfaeSJed Brown 400bcb2dfaeSJed BrownBackends available in this release: 401bcb2dfaeSJed Brown 40268e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 40368e843eeSJed Brown|-------------------------|-----------------------------------------------------| 40468e843eeSJed Brown| `/cpu/self/blocked` | Blocked reference implementation | 40568e843eeSJed Brown| `/cpu/self/ref` | Serial reference implementation | 40668e843eeSJed Brown| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 40768e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 40868e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 40968e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 41068e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 41168e843eeSJed Brown| `/gpu/magma` | CUDA MAGMA kernels | 412bcb2dfaeSJed Brown 413bcb2dfaeSJed BrownExamples available in this release: 414bcb2dfaeSJed Brown 41568e843eeSJed Brown:::{list-table} 41668e843eeSJed Brown:header-rows: 1 41768e843eeSJed Brown:widths: auto 41868e843eeSJed Brown* - User code 41968e843eeSJed Brown - Example 42068e843eeSJed Brown* - `ceed` 42168e843eeSJed Brown - * ex1 (volume) 42268e843eeSJed Brown* - `mfem` 42368e843eeSJed Brown - * BP1 (scalar mass operator) 42468e843eeSJed Brown * BP3 (scalar Laplace operator) 42568e843eeSJed Brown* - `petsc` 42668e843eeSJed Brown - * BP1 (scalar mass operator) 42768e843eeSJed Brown * BP3 (scalar Laplace operator) 42868e843eeSJed Brown* - `nek5000` 42968e843eeSJed Brown - * BP1 (scalar mass operator) 43068e843eeSJed Brown * BP3 (scalar Laplace operator) 43168e843eeSJed Brown::: 432bcb2dfaeSJed Brown 433bcb2dfaeSJed Brown(v0-21)= 434bcb2dfaeSJed Brown 435bcb2dfaeSJed Brown## v0.21 (Sep 30, 2018) 436bcb2dfaeSJed Brown 437bcb2dfaeSJed BrownA MAGMA backend (which relies upon the 438bcb2dfaeSJed Brown[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this 439bcb2dfaeSJed Brownrelease. This initial integration set up the framework of using MAGMA and provided the 440bcb2dfaeSJed BrownlibCEED functionality through MAGMA kernels as one of libCEED’s computational backends. 441bcb2dfaeSJed BrownAs any other backend, the MAGMA backend provides extended basic data structures for 442bcb2dfaeSJed Brown{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements 443bcb2dfaeSJed Brownthe fundamental CEED building blocks to work with the new data structures. 444bcb2dfaeSJed BrownIn general, the MAGMA-specific data structures keep the libCEED pointers to CPU data 445bcb2dfaeSJed Brownbut also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled 446bcb2dfaeSJed Browninternally, and thus seamlessly to the user, through the functions/methods that are 447bcb2dfaeSJed Brownprovided to support them. 448bcb2dfaeSJed Brown 449bcb2dfaeSJed BrownBackends available in this release: 450bcb2dfaeSJed Brown 45168e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 45268e843eeSJed Brown|-------------------------|---------------------------------| 45368e843eeSJed Brown| `/cpu/self` | Serial reference implementation | 45468e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 45568e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 45668e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 45768e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 45868e843eeSJed Brown| `/gpu/magma` | CUDA MAGMA kernels | 459bcb2dfaeSJed Brown 460bcb2dfaeSJed BrownExamples available in this release: 461bcb2dfaeSJed Brown 46268e843eeSJed Brown:::{list-table} 46368e843eeSJed Brown:header-rows: 1 46468e843eeSJed Brown:widths: auto 46568e843eeSJed Brown* - User code 46668e843eeSJed Brown - Example 46768e843eeSJed Brown* - `ceed` 46868e843eeSJed Brown - * ex1 (volume) 46968e843eeSJed Brown* - `mfem` 47068e843eeSJed Brown - * BP1 (scalar mass operator) 47168e843eeSJed Brown * BP3 (scalar Laplace operator) 47268e843eeSJed Brown* - `petsc` 47368e843eeSJed Brown - * BP1 (scalar mass operator) 47468e843eeSJed Brown* - `nek5000` 47568e843eeSJed Brown - * BP1 (scalar mass operator) 47668e843eeSJed Brown::: 477bcb2dfaeSJed Brown 478bcb2dfaeSJed Brown(v0-2)= 479bcb2dfaeSJed Brown 480bcb2dfaeSJed Brown## v0.2 (Mar 30, 2018) 481bcb2dfaeSJed Brown 482bcb2dfaeSJed BrownlibCEED was made publicly available the first full CEED software distribution, release 483bcb2dfaeSJed BrownCEED 1.0. The distribution was made available using the Spack package manager to provide 484bcb2dfaeSJed Browna common, easy-to-use build environment, where the user can build the CEED distribution 485bcb2dfaeSJed Brownwith all dependencies. This release included a new Fortran interface for the library. 486bcb2dfaeSJed BrownThis release also contained major improvements in the OCCA backend (including a new 487bcb2dfaeSJed Brown`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to 488bcb2dfaeSJed Browncompute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an 489bcb2dfaeSJed Brown`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application 490bcb2dfaeSJed Brownof the Laplace operator) was also added to this release. 491bcb2dfaeSJed Brown 492bcb2dfaeSJed BrownBackends available in this release: 493bcb2dfaeSJed Brown 49468e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 49568e843eeSJed Brown|-------------------------|---------------------------------| 49668e843eeSJed Brown| `/cpu/self` | Serial reference implementation | 49768e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 49868e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 49968e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 50068e843eeSJed Brown| `/ocl/occa` | OpenCL OCCA kernels | 501bcb2dfaeSJed Brown 502bcb2dfaeSJed BrownExamples available in this release: 503bcb2dfaeSJed Brown 50468e843eeSJed Brown:::{list-table} 50568e843eeSJed Brown:header-rows: 1 50668e843eeSJed Brown:widths: auto 50768e843eeSJed Brown* - User code 50868e843eeSJed Brown - Example 50968e843eeSJed Brown* - `ceed` 51068e843eeSJed Brown - * ex1 (volume) 51168e843eeSJed Brown* - `mfem` 51268e843eeSJed Brown - * BP1 (scalar mass operator) 51368e843eeSJed Brown * BP3 (scalar Laplace operator) 51468e843eeSJed Brown* - `petsc` 51568e843eeSJed Brown - * BP1 (scalar mass operator) 51668e843eeSJed Brown* - `nek5000` 51768e843eeSJed Brown - * BP1 (scalar mass operator) 51868e843eeSJed Brown::: 519bcb2dfaeSJed Brown 520bcb2dfaeSJed Brown(v0-1)= 521bcb2dfaeSJed Brown 522bcb2dfaeSJed Brown## v0.1 (Jan 3, 2018) 523bcb2dfaeSJed Brown 524bcb2dfaeSJed BrownInitial low-level API of the CEED project. The low-level API provides a set of Finite 525bcb2dfaeSJed BrownElements kernels and components for writing new low-level kernels. Examples include: 526bcb2dfaeSJed Brownvector and sparse linear algebra, element matrix assembly over a batch of elements, 527bcb2dfaeSJed Brownpartial assembly and action for efficient high-order operators like mass, diffusion, 528bcb2dfaeSJed Brownadvection, etc. The main goal of the low-level API is to establish the basis for the 529bcb2dfaeSJed Brownhigh-level API. Also, identifying such low-level kernels and providing a reference 530bcb2dfaeSJed Brownimplementation for them serves as the basis for specialized backend implementations. 531bcb2dfaeSJed BrownThis release contained several backends: `/cpu/self`, and backends which rely upon the 532bcb2dfaeSJed Brown[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`, 533bcb2dfaeSJed Brown`/gpu/occa`, and `/omp/occa`. 534bcb2dfaeSJed BrownIt also included several examples, in the `examples` folder: 535bcb2dfaeSJed BrownA standalone code that shows the usage of libCEED (with no external 536bcb2dfaeSJed Browndependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1 537bcb2dfaeSJed Brown(with the application of the mass operator); and a `petsc` example to perform BP1 538bcb2dfaeSJed Brown(with the application of the mass operator). 539bcb2dfaeSJed Brown 540bcb2dfaeSJed BrownBackends available in this release: 541bcb2dfaeSJed Brown 54268e843eeSJed Brown| CEED resource (`-ceed`) | Backend | 54368e843eeSJed Brown|-------------------------|---------------------------------| 54468e843eeSJed Brown| `/cpu/self` | Serial reference implementation | 54568e843eeSJed Brown| `/cpu/occa` | Serial OCCA kernels | 54668e843eeSJed Brown| `/gpu/occa` | CUDA OCCA kernels | 54768e843eeSJed Brown| `/omp/occa` | OpenMP OCCA kernels | 548bcb2dfaeSJed Brown 549bcb2dfaeSJed BrownExamples available in this release: 550bcb2dfaeSJed Brown 551bcb2dfaeSJed Brown| User code | Example | 55268e843eeSJed Brown|-----------------------|-----------------------------------| 55368e843eeSJed Brown| `ceed` | ex1 (scalar Laplace operator) | 55468e843eeSJed Brown| `mfem` | BP1 (scalar mass operator) | 55568e843eeSJed Brown| `petsc` | BP1 (scalar mass operator) | 556bcb2dfaeSJed Brown``` 557