1# Changes/Release Notes 2 3On this page we provide a summary of the main API changes, new features and examples for each release of libCEED. 4 5(main)= 6 7## Current `main` branch 8 9### Interface changes 10 11- Add `bool` field type for `CeedQFunctionContext` and related interfaces to use `bool` fields. 12- `CEED_BASIS_COLLOCATED` removed; users should only use `CEED_BASIS_NONE`. 13 14### New features 15 16- Add `CeedOperatorCreateAtPoints` which evaluates the `CeedQFunction` at arbitrary locations in each element, for use in Particle in Cell, Material Point Method, and similar methods. 17 18### Examples 19 20- Add deal.II example with CEED BP suite. 21 22(v0-12)= 23 24## v0.12 (Oct 31, 2023) 25 26### Interface changes 27 28- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency. 29For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`. 30- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface. 31 32### New features 33 34- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name. 35- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions. 36- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases. 37- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`. 38- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements. 39- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`. 40- Added {c:func}`CeedElemRestrictionCreateCurlOriented`, similar to {c:func}`CeedElemRestrictionCreateOriented`, for element restrictions requiring more general element transformations such as those for high-order $H(\text{curl})$ spaces on tetrahedra (see [https://dl.acm.org/doi/pdf/10.1145/3524456](https://dl.acm.org/doi/pdf/10.1145/3524456)). 41- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`. 42- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads. 43 44### Examples 45 46- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh. 47 48#### {ref}`example-petsc-bps` 49 50- Requires PETSc version 3.19 or later. 51 52#### {ref}`example-petsc-navier-stokes` 53 54- Updated restart and checkpointing interface. 55- Add data-driven subgrid-stress model. 56- Add differential filtering of solution. 57- Add turbulence statistics collection over spanwise-symmetric geometries. 58- Add Taylor-Green vortex initial condition. 59- Add Riemann-based outflow boundary conditions. 60- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer. 61- Add Internal Damping Layer (IDL) for helping turbulent simulation stability. 62- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates. 63 64(v0-11)= 65 66## v0.11 (Dec 24, 2022) 67 68### Interface changes 69 70- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output. 71- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`. 72- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface. 73- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity. 74Some users previously misinterpreted a `CeedOperator` field using `CEED_BASIS_COLLOCATED` as meaning that the entire `CeedOperator` used a quadrature space that is collocated with the nodal space of the active bases. 75 76### New features 77 78- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user. 79- Improved support for $H(\text{div})$ bases. 80- Added `CeedInt_FMT` to support potential future use of larger integer sizes. 81- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`. 82- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled. 83Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend. 84This issue will be fixed in a future OCCA release. 85 86### Bugfix 87 88- Fix bug in setting device id for GPU backends. 89- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends. 90- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`. 91- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes. 92 93### Examples 94 95#### {ref}`example-petsc-navier-stokes` 96 97- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs. 98- Refactored to improve code reuse and modularity. 99- Support for primitive variables for more accurate boundary layers and all-speed flow. 100- Added $YZ\beta$ shock capturing scheme and Shock Tube example. 101- Added Channel example, with comparison to analytic solutions. 102- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations. 103- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions. 104- Added "freestream" boundary conditions based on HLLC Riemann solver. 105- Automated stabilization coefficients for different basis degree. 106 107#### {ref}`example-petsc-bps` 108 109- Support for convergence studies. 110 111### Maintainability 112 113- Refactored `/gpu/cuda/shared` and `/gpu/cuda/gen` as well as `/gpu/hip/shared` and `/gpu/hip/gen` backend to improve maintainablity and reduce duplicated code. 114- Enabled support for `p > 8` for `/gpu/*/shared` backends. 115- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`. 116- Improved test harness. 117 118(v0-10-1)= 119 120## v0.10.1 (Apr 11, 2022) 121 122### Interface changes 123 124- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application. 125 126### New features 127 128- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends). 129This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work. 130 131### Bugfix 132 133- Install JiT source files in install directory to fix GPU functionality for installed libCEED. 134 135(v0-10)= 136 137## v0.10 (Mar 21, 2022) 138 139### Interface changes 140 141- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields. 142- 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`. 143- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added. 144- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients. 145- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces. 146- Warning added when compiling OCCA backend to alert users that this backend is experimental. 147- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`. 148- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name. 149- 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`. 150- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`. 151- 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. 152- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`. 153- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums. 154- 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. 155- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy. 156- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator. 157 158### New features 159 160- `CeedScalar` can now be set as `float` or `double` at compile time. 161- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends. 162- 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. 163- Remove need to guard library headers in QFunction source for code generation backends. 164- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present. 165- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally. 166- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file. 167- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor. 168- 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. 169- Added support for element matrix assembly in GPU backends. 170 171### Maintainability 172 173- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage. 174- `Include-what-you-use` makefile target added as `make iwyu`. 175- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase. 176- Put GPU JiTed kernel source code into separate files. 177- 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. 178 179(v0-9)= 180 181## v0.9 (Jul 6, 2021) 182 183### Interface changes 184 185- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact. 186 187### New features 188 189- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use. 190- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends. 191- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations. 192- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed). 193 194### Performance improvements 195 196### Examples 197 198- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations. 199- Fluid mechanics example adds GPU support and improves modularity. 200 201### Deprecated backends 202 203- 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. 204 205(v0-8)= 206 207## v0.8 (Mar 31, 2021) 208 209### Interface changes 210 211- Error handling improved to include enumerated error codes for C interface return values. 212- 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. 213 214### New features 215 216- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features. 217- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly. 218- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators. 219 220### Performance improvements 221 222- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`. 223- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`. 224 225### Examples 226 227- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions. 228- {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. 229- {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. 230- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree. 231 232(v0-7)= 233 234## v0.7 (Sep 29, 2020) 235 236### Interface changes 237 238- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors. 239 As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`. 240 These changes improve support for mixed finite element methods. 241- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature. 242- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity. 243- 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. 244- 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. 245- 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. 246- 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. 247 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`. 248- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory. 249- 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. 250 251### New features 252 253- New HIP backend: `/gpu/hip/ref`. 254- CeedQFunction support for user `CUfunction`s in some backends 255 256### Performance improvements 257 258- OCCA backend rebuilt to facilitate future performance enhancements. 259- PETSc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`. 260 261### Examples 262 263- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions. 264 265### Deprecated backends 266 267- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`. 268 269(v0-6)= 270 271## v0.6 (Mar 29, 2020) 272 273libCEED v0.6 contains numerous new features and examples, as well as expanded 274documentation in [this new website](https://libceed.org). 275 276### New features 277 278- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly 279 1-1 correspondence with the C interface, plus some convenience features. For instance, 280 data stored in the {cpp:type}`CeedVector` structure are available without copy as 281 {py:class}`numpy.ndarray`. Short tutorials are provided in 282 [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/). 283- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point, 284 {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal 285 ({c:func}`CeedOperatorAssembleLinearDiagonal`). These operations are useful for 286 preconditioning ingredients and are used in the libCEED's multigrid examples. 287- The inverse of separable operators can be obtained using 288 {c:func}`CeedOperatorCreateFDMElementInverse` and applied with 289 {c:func}`CeedOperatorApply`. This is a useful preconditioning ingredient, 290 especially for Laplacians and related operators. 291- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`, 292 {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`. 293- Make public accessors for various attributes to facilitate writing composable code. 294- New backend: `/cpu/self/memcheck/serial`. 295- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA 296 backends. (Single source is coming soon for OCCA backends.) 297- Fix some missing edge cases in CUDA backend. 298 299### Performance Improvements 300 301- MAGMA backend performance optimization and non-tensor bases. 302- No-copy optimization in {c:func}`CeedOperatorApply`. 303 304### Interface changes 305 306- Replace {code}`CeedElemRestrictionCreateIdentity` and 307 {code}`CeedElemRestrictionCreateBlocked` with more flexible 308 {c:func}`CeedElemRestrictionCreateStrided` and 309 {c:func}`CeedElemRestrictionCreateBlockedStrided`. 310- Add arguments to {c:func}`CeedQFunctionCreateIdentity`. 311- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification 312 with {cpp:enum}`CeedInterlaceMode`. This is now an attribute of the 313 {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no 314 longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and 315 {c:func}`CeedElemRestrictionApply`. 316 317### Examples 318 319libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`. 320Notable additions include: 321 322- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of 323 a domain in 1, 2, and 3 dimensions by applying a Laplacian. 324 325- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area 326 of domains (like the cube and sphere) by direct integration on a surface mesh; 327 demonstrates geometric dimension different from topological dimension. 328 329- PETSc {ref}`example-petsc-bps`: 330 331 - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support. 332 - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements 333 and transparent CUDA support. 334 - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`): 335 generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric 336 dimension different from topological dimension. 337 338- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid 339 solver with algebraic multigrid coarse solve. 340 341- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly 342 `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`), 343 implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and 344 quasi-2D computational domain support. 345 346- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for 347 linear elasticity, small-strain hyperelasticity, and globalized finite-strain 348 hyperelasticity using p-multigrid with algebraic multigrid coarse solve. 349 350(v0-5)= 351 352## v0.5 (Sep 18, 2019) 353 354For this release, several improvements were made. Two new CUDA backends were added to 355the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art 356performance using single-source {ref}`CeedQFunction`. From this release, users 357can define Q-Functions in a single source code independently of the targeted backend 358with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU 359compilation of the user provided {ref}`CeedQFunction` code. To allow a unified 360declaration, the {ref}`CeedQFunction` API has undergone a slight change: 361the `QFunctionField` parameter `ncomp` has been changed to `size`. This change 362requires setting the previous value of `ncomp` to `ncomp*dim` when adding a 363`QFunctionField` with eval mode `CEED EVAL GRAD`. 364 365Additionally, new CPU backends 366were included in this release, such as the `/cpu/self/opt/*` backends (which are 367written in pure C and use partial **E-vectors** to improve performance) and the 368`/cpu/self/ref/memcheck` backend (which relies upon the 369[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user 370{ref}`CeedQFunction` have no undefined values). 371This release also included various performance improvements, bug fixes, new examples, 372and improved tests. Among these improvements, vectorized instructions for 373{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD` 374instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and 375identity Q-Functions were introduced, and the PETSc benchmark problems were expanded 376to include unstructured meshes handling were. For this expansion, the prior version of 377the PETSc BPs, which only included data associated with structured geometries, were 378renamed `bpsraw`, and the new version of the BPs, which can handle data associated 379with any unstructured geometry, were called `bps`. Additionally, other benchmark 380problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively), 381and BP5 and BP6 (the collocated versions---for which the quadrature points are the same 382as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc 383examples. Furthermoew, another standalone libCEED example, called `ex2`, which 384computes the surface area of a given mesh was added to this release. 385 386Backends available in this release: 387 388| CEED resource (`-ceed`) | Backend | 389|--------------------------|-----------------------------------------------------| 390| `/cpu/self/ref/serial` | Serial reference implementation | 391| `/cpu/self/ref/blocked` | Blocked reference implementation | 392| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks | 393| `/cpu/self/opt/serial` | Serial optimized C implementation | 394| `/cpu/self/opt/blocked` | Blocked optimized C implementation | 395| `/cpu/self/avx/serial` | Serial AVX implementation | 396| `/cpu/self/avx/blocked` | Blocked AVX implementation | 397| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 398| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 399| `/cpu/occa` | Serial OCCA kernels | 400| `/gpu/occa` | CUDA OCCA kernels | 401| `/omp/occa` | OpenMP OCCA kernels | 402| `/ocl/occa` | OpenCL OCCA kernels | 403| `/gpu/cuda/ref` | Reference pure CUDA kernels | 404| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 405| `/gpu/cuda/shared` | Optimized pure CUDA kernels using shared memory | 406| `/gpu/cuda/gen` | Optimized pure CUDA kernels using code generation | 407| `/gpu/magma` | CUDA MAGMA kernels | 408 409Examples available in this release: 410 411:::{list-table} 412:header-rows: 1 413:widths: auto 414* - User code 415 - Example 416* - `ceed` 417 - * ex1 (volume) 418 * ex2 (surface) 419* - `mfem` 420 - * BP1 (scalar mass operator) 421 * BP3 (scalar Laplace operator) 422* - `petsc` 423 - * BP1 (scalar mass operator) 424 * BP2 (vector mass operator) 425 * BP3 (scalar Laplace operator) 426 * BP4 (vector Laplace operator) 427 * BP5 (collocated scalar Laplace operator) 428 * BP6 (collocated vector Laplace operator) 429 * Navier-Stokes 430* - `nek5000` 431 - * BP1 (scalar mass operator) 432 * BP3 (scalar Laplace operator) 433::: 434 435(v0-4)= 436 437## v0.4 (Apr 1, 2019) 438 439libCEED v0.4 was made again publicly available in the second full CEED software 440distribution, release CEED 2.0. This release contained notable features, such as 441four new CPU backends, two new GPU backends, CPU backend optimizations, initial 442support for operator composition, performance benchmarking, and a Navier-Stokes demo. 443The new CPU backends in this release came in two families. The `/cpu/self/*/serial` 444backends process one element at a time and are intended for meshes with a smaller number 445of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of 446eight interlaced elements and are intended for meshes with higher numbers of elements. 447The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU 448performance. The `/cpu/self/xsmm/*` backends rely upon the 449[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU 450performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA. 451The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable 452performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple 453parallelization approach, where each thread treats a finite element. Using just in time 454compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this 455backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend 456achieve good peak performance for 1D, 2D, and low order 3D problems, but performance 457deteriorates very quickly when threads run out of registers. 458 459A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED 460examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`). 461This example solves the time-dependent Navier-Stokes equations of compressible gas 462dynamics in a static Eulerian three-dimensional frame, using structured high-order 463finite/spectral element spatial discretizations and explicit high-order time-stepping 464(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc, 465so that the pointwise physics (defined at quadrature points) is separated from the 466parallelization and meshing concerns. 467 468Backends available in this release: 469 470| CEED resource (`-ceed`) | Backend | 471|--------------------------|-----------------------------------------------------| 472| `/cpu/self/ref/serial` | Serial reference implementation | 473| `/cpu/self/ref/blocked` | Blocked reference implementation | 474| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 475| `/cpu/self/avx/serial` | Serial AVX implementation | 476| `/cpu/self/avx/blocked` | Blocked AVX implementation | 477| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 478| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 479| `/cpu/occa` | Serial OCCA kernels | 480| `/gpu/occa` | CUDA OCCA kernels | 481| `/omp/occa` | OpenMP OCCA kernels | 482| `/ocl/occa` | OpenCL OCCA kernels | 483| `/gpu/cuda/ref` | Reference pure CUDA kernels | 484| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 485| `/gpu/magma` | CUDA MAGMA kernels | 486 487Examples available in this release: 488 489:::{list-table} 490:header-rows: 1 491:widths: auto 492* - User code 493 - Example 494* - `ceed` 495 - * ex1 (volume) 496* - `mfem` 497 - * BP1 (scalar mass operator) 498 * BP3 (scalar Laplace operator) 499* - `petsc` 500 - * BP1 (scalar mass operator) 501 * BP3 (scalar Laplace operator) 502 * Navier-Stokes 503* - `nek5000` 504 - * BP1 (scalar mass operator) 505 * BP3 (scalar Laplace operator) 506::: 507 508(v0-3)= 509 510## v0.3 (Sep 30, 2018) 511 512Notable features in this release include active/passive field interface, support for 513non-tensor bases, backend optimization, and improved Fortran interface. This release 514also focused on providing improved continuous integration, and many new tests with code 515coverage reports of about 90%. This release also provided a significant change to the 516public interface: a {ref}`CeedQFunction` can take any number of named input and output 517arguments while {ref}`CeedOperator` connects them to the actual data, which may be 518supplied explicitly to `CeedOperatorApply()` (active) or separately via 519`CeedOperatorSetField()` (passive). This interface change enables reusable libraries 520of CeedQFunctions and composition of block solvers constructed using 521{ref}`CeedOperator`. A concept of blocked restriction was added to this release and 522used in an optimized CPU backend. Although this is typically not visible to the user, 523it enables effective use of arbitrary-length SIMD while maintaining cache locality. 524This CPU backend also implements an algebraic factorization of tensor product gradients 525to perform fewer operations than standard application of interpolation and 526differentiation from nodes to quadrature points. This algebraic formulation 527automatically supports non-polynomial and non-interpolatory bases, thus is more general 528than the more common derivation in terms of Lagrange polynomials on the quadrature points. 529 530Backends available in this release: 531 532| CEED resource (`-ceed`) | Backend | 533|-------------------------|-----------------------------------------------------| 534| `/cpu/self/blocked` | Blocked reference implementation | 535| `/cpu/self/ref` | Serial reference implementation | 536| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 537| `/cpu/occa` | Serial OCCA kernels | 538| `/gpu/occa` | CUDA OCCA kernels | 539| `/omp/occa` | OpenMP OCCA kernels | 540| `/ocl/occa` | OpenCL OCCA kernels | 541| `/gpu/magma` | CUDA MAGMA kernels | 542 543Examples available in this release: 544 545:::{list-table} 546:header-rows: 1 547:widths: auto 548* - User code 549 - Example 550* - `ceed` 551 - * ex1 (volume) 552* - `mfem` 553 - * BP1 (scalar mass operator) 554 * BP3 (scalar Laplace operator) 555* - `petsc` 556 - * BP1 (scalar mass operator) 557 * BP3 (scalar Laplace operator) 558* - `nek5000` 559 - * BP1 (scalar mass operator) 560 * BP3 (scalar Laplace operator) 561::: 562 563(v0-21)= 564 565## v0.21 (Sep 30, 2018) 566 567A MAGMA backend (which relies upon the 568[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this 569release. This initial integration set up the framework of using MAGMA and provided the 570libCEED functionality through MAGMA kernels as one of libCEED’s computational backends. 571As any other backend, the MAGMA backend provides extended basic data structures for 572{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements 573the fundamental CEED building blocks to work with the new data structures. 574In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data 575but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled 576internally, and thus seamlessly to the user, through the functions/methods that are 577provided to support them. 578 579Backends available in this release: 580 581| CEED resource (`-ceed`) | Backend | 582|-------------------------|---------------------------------| 583| `/cpu/self` | Serial reference implementation | 584| `/cpu/occa` | Serial OCCA kernels | 585| `/gpu/occa` | CUDA OCCA kernels | 586| `/omp/occa` | OpenMP OCCA kernels | 587| `/ocl/occa` | OpenCL OCCA kernels | 588| `/gpu/magma` | CUDA MAGMA kernels | 589 590Examples available in this release: 591 592:::{list-table} 593:header-rows: 1 594:widths: auto 595* - User code 596 - Example 597* - `ceed` 598 - * ex1 (volume) 599* - `mfem` 600 - * BP1 (scalar mass operator) 601 * BP3 (scalar Laplace operator) 602* - `petsc` 603 - * BP1 (scalar mass operator) 604* - `nek5000` 605 - * BP1 (scalar mass operator) 606::: 607 608(v0-2)= 609 610## v0.2 (Mar 30, 2018) 611 612libCEED was made publicly available the first full CEED software distribution, release 613CEED 1.0. The distribution was made available using the Spack package manager to provide 614a common, easy-to-use build environment, where the user can build the CEED distribution 615with all dependencies. This release included a new Fortran interface for the library. 616This release also contained major improvements in the OCCA backend (including a new 617`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to 618compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an 619`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application 620of the Laplace operator) was also added to this release. 621 622Backends available in this release: 623 624| CEED resource (`-ceed`) | Backend | 625|-------------------------|---------------------------------| 626| `/cpu/self` | Serial reference implementation | 627| `/cpu/occa` | Serial OCCA kernels | 628| `/gpu/occa` | CUDA OCCA kernels | 629| `/omp/occa` | OpenMP OCCA kernels | 630| `/ocl/occa` | OpenCL OCCA kernels | 631 632Examples available in this release: 633 634:::{list-table} 635:header-rows: 1 636:widths: auto 637* - User code 638 - Example 639* - `ceed` 640 - * ex1 (volume) 641* - `mfem` 642 - * BP1 (scalar mass operator) 643 * BP3 (scalar Laplace operator) 644* - `petsc` 645 - * BP1 (scalar mass operator) 646* - `nek5000` 647 - * BP1 (scalar mass operator) 648::: 649 650(v0-1)= 651 652## v0.1 (Jan 3, 2018) 653 654Initial low-level API of the CEED project. The low-level API provides a set of Finite 655Elements kernels and components for writing new low-level kernels. Examples include: 656vector and sparse linear algebra, element matrix assembly over a batch of elements, 657partial assembly and action for efficient high-order operators like mass, diffusion, 658advection, etc. The main goal of the low-level API is to establish the basis for the 659high-level API. Also, identifying such low-level kernels and providing a reference 660implementation for them serves as the basis for specialized backend implementations. 661This release contained several backends: `/cpu/self`, and backends which rely upon the 662[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`, 663`/gpu/occa`, and `/omp/occa`. 664It also included several examples, in the `examples` folder: 665A standalone code that shows the usage of libCEED (with no external 666dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1 667(with the application of the mass operator); and a `petsc` example to perform BP1 668(with the application of the mass operator). 669 670Backends available in this release: 671 672| CEED resource (`-ceed`) | Backend | 673|-------------------------|---------------------------------| 674| `/cpu/self` | Serial reference implementation | 675| `/cpu/occa` | Serial OCCA kernels | 676| `/gpu/occa` | CUDA OCCA kernels | 677| `/omp/occa` | OpenMP OCCA kernels | 678 679Examples available in this release: 680 681| User code | Example | 682|-----------------------|-----------------------------------| 683| `ceed` | ex1 (scalar Laplace operator) | 684| `mfem` | BP1 (scalar mass operator) | 685| `petsc` | BP1 (scalar mass operator) | 686``` 687