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- Remove unneeded pointer for `CeedElemRestrictionGetELayout`. 14- Require use of `Ceed*Destroy()` on Ceed objects returned from `CeedOperatorFieldGet*()`; 15 16### New features 17 18- Add `CeedOperatorCreateAtPoints` which evaluates the `CeedQFunction` at arbitrary locations in each element, for use in Particle in Cell, Material Point Method, and similar methods. 19- Add `CeedElemRestrictionGetLLayout` to provide L-vector layout for strided `CeedElemRestriction` created with `CEED_BACKEND_STRIDES`. 20- Add `CeedVectorReturnCeed` and similar when parent `Ceed` context for a libCEED object is only needed once in a calling scope. 21- Enable `#pragma once` for all JiT source; remove duplicate includes in JiT source string before compilation. 22- Allow user to set additional compiler options for CUDA and HIP JiT. 23Specifically, directories set with `CeedAddJitSourceRoot(ceed, "foo/bar")` will be used to set `-Ifoo/bar` and defines set with `CeedAddJitDefine(ceed, "foo=bar")` will be used to set `-Dfoo=bar`. 24 25### Examples 26 27- Add deal.II example with CEED BP suite. 28 29(v0-12)= 30 31## v0.12 (Oct 31, 2023) 32 33### Interface changes 34 35- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency. 36For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`. 37- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface. 38 39### New features 40 41- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name. 42- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions. 43- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases. 44- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`. 45- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements. 46- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`. 47- 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)). 48- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`. 49- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads. 50 51### Examples 52 53- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh. 54 55#### {ref}`example-petsc-bps` 56 57- Requires PETSc version 3.19 or later. 58 59#### {ref}`example-petsc-navier-stokes` 60 61- Updated restart and checkpointing interface. 62- Add data-driven subgrid-stress model. 63- Add differential filtering of solution. 64- Add turbulence statistics collection over spanwise-symmetric geometries. 65- Add Taylor-Green vortex initial condition. 66- Add Riemann-based outflow boundary conditions. 67- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer. 68- Add Internal Damping Layer (IDL) for helping turbulent simulation stability. 69- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates. 70 71(v0-11)= 72 73## v0.11 (Dec 24, 2022) 74 75### Interface changes 76 77- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output. 78- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`. 79- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface. 80- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity. 81Some 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. 82 83### New features 84 85- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user. 86- Improved support for $H(\text{div})$ bases. 87- Added `CeedInt_FMT` to support potential future use of larger integer sizes. 88- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`. 89- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled. 90Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend. 91This issue will be fixed in a future OCCA release. 92 93### Bugfix 94 95- Fix bug in setting device id for GPU backends. 96- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends. 97- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`. 98- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes. 99 100### Examples 101 102#### {ref}`example-petsc-navier-stokes` 103 104- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs. 105- Refactored to improve code reuse and modularity. 106- Support for primitive variables for more accurate boundary layers and all-speed flow. 107- Added $YZ\beta$ shock capturing scheme and Shock Tube example. 108- Added Channel example, with comparison to analytic solutions. 109- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations. 110- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions. 111- Added "freestream" boundary conditions based on HLLC Riemann solver. 112- Automated stabilization coefficients for different basis degree. 113 114#### {ref}`example-petsc-bps` 115 116- Support for convergence studies. 117 118### Maintainability 119 120- 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. 121- Enabled support for `p > 8` for `/gpu/*/shared` backends. 122- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`. 123- Improved test harness. 124 125(v0-10-1)= 126 127## v0.10.1 (Apr 11, 2022) 128 129### Interface changes 130 131- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application. 132 133### New features 134 135- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends). 136This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work. 137 138### Bugfix 139 140- Install JiT source files in install directory to fix GPU functionality for installed libCEED. 141 142(v0-10)= 143 144## v0.10 (Mar 21, 2022) 145 146### Interface changes 147 148- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields. 149- 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`. 150- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added. 151- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients. 152- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces. 153- Warning added when compiling OCCA backend to alert users that this backend is experimental. 154- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`. 155- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name. 156- 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`. 157- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`. 158- 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. 159- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`. 160- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums. 161- 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. 162- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy. 163- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator. 164 165### New features 166 167- `CeedScalar` can now be set as `float` or `double` at compile time. 168- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends. 169- 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. 170- Remove need to guard library headers in QFunction source for code generation backends. 171- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present. 172- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally. 173- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file. 174- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor. 175- 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. 176- Added support for element matrix assembly in GPU backends. 177 178### Maintainability 179 180- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage. 181- `Include-what-you-use` makefile target added as `make iwyu`. 182- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase. 183- Put GPU JiTed kernel source code into separate files. 184- 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. 185 186(v0-9)= 187 188## v0.9 (Jul 6, 2021) 189 190### Interface changes 191 192- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact. 193 194### New features 195 196- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use. 197- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends. 198- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations. 199- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed). 200 201### Performance improvements 202 203### Examples 204 205- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations. 206- Fluid mechanics example adds GPU support and improves modularity. 207 208### Deprecated backends 209 210- 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. 211 212(v0-8)= 213 214## v0.8 (Mar 31, 2021) 215 216### Interface changes 217 218- Error handling improved to include enumerated error codes for C interface return values. 219- 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. 220 221### New features 222 223- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features. 224- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly. 225- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators. 226 227### Performance improvements 228 229- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`. 230- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`. 231 232### Examples 233 234- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions. 235- {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. 236- {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. 237- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree. 238 239(v0-7)= 240 241## v0.7 (Sep 29, 2020) 242 243### Interface changes 244 245- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors. 246 As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`. 247 These changes improve support for mixed finite element methods. 248- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature. 249- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity. 250- 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. 251- 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. 252- 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. 253- 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. 254 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`. 255- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory. 256- 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. 257 258### New features 259 260- New HIP backend: `/gpu/hip/ref`. 261- CeedQFunction support for user `CUfunction`s in some backends 262 263### Performance improvements 264 265- OCCA backend rebuilt to facilitate future performance enhancements. 266- PETSc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`. 267 268### Examples 269 270- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions. 271 272### Deprecated backends 273 274- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`. 275 276(v0-6)= 277 278## v0.6 (Mar 29, 2020) 279 280libCEED v0.6 contains numerous new features and examples, as well as expanded 281documentation in [this new website](https://libceed.org). 282 283### New features 284 285- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly 286 1-1 correspondence with the C interface, plus some convenience features. For instance, 287 data stored in the {cpp:type}`CeedVector` structure are available without copy as 288 {py:class}`numpy.ndarray`. Short tutorials are provided in 289 [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/). 290- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point, 291 {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal 292 ({c:func}`CeedOperatorAssembleLinearDiagonal`). These operations are useful for 293 preconditioning ingredients and are used in the libCEED's multigrid examples. 294- The inverse of separable operators can be obtained using 295 {c:func}`CeedOperatorCreateFDMElementInverse` and applied with 296 {c:func}`CeedOperatorApply`. This is a useful preconditioning ingredient, 297 especially for Laplacians and related operators. 298- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`, 299 {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`. 300- Make public accessors for various attributes to facilitate writing composable code. 301- New backend: `/cpu/self/memcheck/serial`. 302- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA 303 backends. (Single source is coming soon for OCCA backends.) 304- Fix some missing edge cases in CUDA backend. 305 306### Performance Improvements 307 308- MAGMA backend performance optimization and non-tensor bases. 309- No-copy optimization in {c:func}`CeedOperatorApply`. 310 311### Interface changes 312 313- Replace {code}`CeedElemRestrictionCreateIdentity` and 314 {code}`CeedElemRestrictionCreateBlocked` with more flexible 315 {c:func}`CeedElemRestrictionCreateStrided` and 316 {c:func}`CeedElemRestrictionCreateBlockedStrided`. 317- Add arguments to {c:func}`CeedQFunctionCreateIdentity`. 318- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification 319 with {cpp:enum}`CeedInterlaceMode`. This is now an attribute of the 320 {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no 321 longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and 322 {c:func}`CeedElemRestrictionApply`. 323 324### Examples 325 326libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`. 327Notable additions include: 328 329- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of 330 a domain in 1, 2, and 3 dimensions by applying a Laplacian. 331 332- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area 333 of domains (like the cube and sphere) by direct integration on a surface mesh; 334 demonstrates geometric dimension different from topological dimension. 335 336- PETSc {ref}`example-petsc-bps`: 337 338 - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support. 339 - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements 340 and transparent CUDA support. 341 - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`): 342 generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric 343 dimension different from topological dimension. 344 345- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid 346 solver with algebraic multigrid coarse solve. 347 348- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly 349 `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`), 350 implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and 351 quasi-2D computational domain support. 352 353- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for 354 linear elasticity, small-strain hyperelasticity, and globalized finite-strain 355 hyperelasticity using p-multigrid with algebraic multigrid coarse solve. 356 357(v0-5)= 358 359## v0.5 (Sep 18, 2019) 360 361For this release, several improvements were made. Two new CUDA backends were added to 362the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art 363performance using single-source {ref}`CeedQFunction`. From this release, users 364can define Q-Functions in a single source code independently of the targeted backend 365with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU 366compilation of the user provided {ref}`CeedQFunction` code. To allow a unified 367declaration, the {ref}`CeedQFunction` API has undergone a slight change: 368the `QFunctionField` parameter `ncomp` has been changed to `size`. This change 369requires setting the previous value of `ncomp` to `ncomp*dim` when adding a 370`QFunctionField` with eval mode `CEED EVAL GRAD`. 371 372Additionally, new CPU backends 373were included in this release, such as the `/cpu/self/opt/*` backends (which are 374written in pure C and use partial **E-vectors** to improve performance) and the 375`/cpu/self/ref/memcheck` backend (which relies upon the 376[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user 377{ref}`CeedQFunction` have no undefined values). 378This release also included various performance improvements, bug fixes, new examples, 379and improved tests. Among these improvements, vectorized instructions for 380{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD` 381instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and 382identity Q-Functions were introduced, and the PETSc benchmark problems were expanded 383to include unstructured meshes handling were. For this expansion, the prior version of 384the PETSc BPs, which only included data associated with structured geometries, were 385renamed `bpsraw`, and the new version of the BPs, which can handle data associated 386with any unstructured geometry, were called `bps`. Additionally, other benchmark 387problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively), 388and BP5 and BP6 (the collocated versions---for which the quadrature points are the same 389as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc 390examples. Furthermoew, another standalone libCEED example, called `ex2`, which 391computes the surface area of a given mesh was added to this release. 392 393Backends available in this release: 394 395| CEED resource (`-ceed`) | Backend | 396|--------------------------|-----------------------------------------------------| 397| `/cpu/self/ref/serial` | Serial reference implementation | 398| `/cpu/self/ref/blocked` | Blocked reference implementation | 399| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks | 400| `/cpu/self/opt/serial` | Serial optimized C implementation | 401| `/cpu/self/opt/blocked` | Blocked optimized C implementation | 402| `/cpu/self/avx/serial` | Serial AVX implementation | 403| `/cpu/self/avx/blocked` | Blocked AVX implementation | 404| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 405| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 406| `/cpu/occa` | Serial OCCA kernels | 407| `/gpu/occa` | CUDA OCCA kernels | 408| `/omp/occa` | OpenMP OCCA kernels | 409| `/ocl/occa` | OpenCL OCCA kernels | 410| `/gpu/cuda/ref` | Reference pure CUDA kernels | 411| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 412| `/gpu/cuda/shared` | Optimized pure CUDA kernels using shared memory | 413| `/gpu/cuda/gen` | Optimized pure CUDA kernels using code generation | 414| `/gpu/magma` | CUDA MAGMA kernels | 415 416Examples available in this release: 417 418:::{list-table} 419:header-rows: 1 420:widths: auto 421* - User code 422 - Example 423* - `ceed` 424 - * ex1 (volume) 425 * ex2 (surface) 426* - `mfem` 427 - * BP1 (scalar mass operator) 428 * BP3 (scalar Laplace operator) 429* - `petsc` 430 - * BP1 (scalar mass operator) 431 * BP2 (vector mass operator) 432 * BP3 (scalar Laplace operator) 433 * BP4 (vector Laplace operator) 434 * BP5 (collocated scalar Laplace operator) 435 * BP6 (collocated vector Laplace operator) 436 * Navier-Stokes 437* - `nek5000` 438 - * BP1 (scalar mass operator) 439 * BP3 (scalar Laplace operator) 440::: 441 442(v0-4)= 443 444## v0.4 (Apr 1, 2019) 445 446libCEED v0.4 was made again publicly available in the second full CEED software 447distribution, release CEED 2.0. This release contained notable features, such as 448four new CPU backends, two new GPU backends, CPU backend optimizations, initial 449support for operator composition, performance benchmarking, and a Navier-Stokes demo. 450The new CPU backends in this release came in two families. The `/cpu/self/*/serial` 451backends process one element at a time and are intended for meshes with a smaller number 452of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of 453eight interlaced elements and are intended for meshes with higher numbers of elements. 454The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU 455performance. The `/cpu/self/xsmm/*` backends rely upon the 456[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU 457performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA. 458The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable 459performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple 460parallelization approach, where each thread treats a finite element. Using just in time 461compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this 462backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend 463achieve good peak performance for 1D, 2D, and low order 3D problems, but performance 464deteriorates very quickly when threads run out of registers. 465 466A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED 467examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`). 468This example solves the time-dependent Navier-Stokes equations of compressible gas 469dynamics in a static Eulerian three-dimensional frame, using structured high-order 470finite/spectral element spatial discretizations and explicit high-order time-stepping 471(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc, 472so that the pointwise physics (defined at quadrature points) is separated from the 473parallelization and meshing concerns. 474 475Backends available in this release: 476 477| CEED resource (`-ceed`) | Backend | 478|--------------------------|-----------------------------------------------------| 479| `/cpu/self/ref/serial` | Serial reference implementation | 480| `/cpu/self/ref/blocked` | Blocked reference implementation | 481| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 482| `/cpu/self/avx/serial` | Serial AVX implementation | 483| `/cpu/self/avx/blocked` | Blocked AVX implementation | 484| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 485| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 486| `/cpu/occa` | Serial OCCA kernels | 487| `/gpu/occa` | CUDA OCCA kernels | 488| `/omp/occa` | OpenMP OCCA kernels | 489| `/ocl/occa` | OpenCL OCCA kernels | 490| `/gpu/cuda/ref` | Reference pure CUDA kernels | 491| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 492| `/gpu/magma` | CUDA MAGMA kernels | 493 494Examples available in this release: 495 496:::{list-table} 497:header-rows: 1 498:widths: auto 499* - User code 500 - Example 501* - `ceed` 502 - * ex1 (volume) 503* - `mfem` 504 - * BP1 (scalar mass operator) 505 * BP3 (scalar Laplace operator) 506* - `petsc` 507 - * BP1 (scalar mass operator) 508 * BP3 (scalar Laplace operator) 509 * Navier-Stokes 510* - `nek5000` 511 - * BP1 (scalar mass operator) 512 * BP3 (scalar Laplace operator) 513::: 514 515(v0-3)= 516 517## v0.3 (Sep 30, 2018) 518 519Notable features in this release include active/passive field interface, support for 520non-tensor bases, backend optimization, and improved Fortran interface. This release 521also focused on providing improved continuous integration, and many new tests with code 522coverage reports of about 90%. This release also provided a significant change to the 523public interface: a {ref}`CeedQFunction` can take any number of named input and output 524arguments while {ref}`CeedOperator` connects them to the actual data, which may be 525supplied explicitly to `CeedOperatorApply()` (active) or separately via 526`CeedOperatorSetField()` (passive). This interface change enables reusable libraries 527of CeedQFunctions and composition of block solvers constructed using 528{ref}`CeedOperator`. A concept of blocked restriction was added to this release and 529used in an optimized CPU backend. Although this is typically not visible to the user, 530it enables effective use of arbitrary-length SIMD while maintaining cache locality. 531This CPU backend also implements an algebraic factorization of tensor product gradients 532to perform fewer operations than standard application of interpolation and 533differentiation from nodes to quadrature points. This algebraic formulation 534automatically supports non-polynomial and non-interpolatory bases, thus is more general 535than the more common derivation in terms of Lagrange polynomials on the quadrature points. 536 537Backends available in this release: 538 539| CEED resource (`-ceed`) | Backend | 540|-------------------------|-----------------------------------------------------| 541| `/cpu/self/blocked` | Blocked reference implementation | 542| `/cpu/self/ref` | Serial reference implementation | 543| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 544| `/cpu/occa` | Serial OCCA kernels | 545| `/gpu/occa` | CUDA OCCA kernels | 546| `/omp/occa` | OpenMP OCCA kernels | 547| `/ocl/occa` | OpenCL OCCA kernels | 548| `/gpu/magma` | CUDA MAGMA kernels | 549 550Examples available in this release: 551 552:::{list-table} 553:header-rows: 1 554:widths: auto 555* - User code 556 - Example 557* - `ceed` 558 - * ex1 (volume) 559* - `mfem` 560 - * BP1 (scalar mass operator) 561 * BP3 (scalar Laplace operator) 562* - `petsc` 563 - * BP1 (scalar mass operator) 564 * BP3 (scalar Laplace operator) 565* - `nek5000` 566 - * BP1 (scalar mass operator) 567 * BP3 (scalar Laplace operator) 568::: 569 570(v0-21)= 571 572## v0.21 (Sep 30, 2018) 573 574A MAGMA backend (which relies upon the 575[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this 576release. This initial integration set up the framework of using MAGMA and provided the 577libCEED functionality through MAGMA kernels as one of libCEED’s computational backends. 578As any other backend, the MAGMA backend provides extended basic data structures for 579{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements 580the fundamental CEED building blocks to work with the new data structures. 581In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data 582but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled 583internally, and thus seamlessly to the user, through the functions/methods that are 584provided to support them. 585 586Backends available in this release: 587 588| CEED resource (`-ceed`) | Backend | 589|-------------------------|---------------------------------| 590| `/cpu/self` | Serial reference implementation | 591| `/cpu/occa` | Serial OCCA kernels | 592| `/gpu/occa` | CUDA OCCA kernels | 593| `/omp/occa` | OpenMP OCCA kernels | 594| `/ocl/occa` | OpenCL OCCA kernels | 595| `/gpu/magma` | CUDA MAGMA kernels | 596 597Examples available in this release: 598 599:::{list-table} 600:header-rows: 1 601:widths: auto 602* - User code 603 - Example 604* - `ceed` 605 - * ex1 (volume) 606* - `mfem` 607 - * BP1 (scalar mass operator) 608 * BP3 (scalar Laplace operator) 609* - `petsc` 610 - * BP1 (scalar mass operator) 611* - `nek5000` 612 - * BP1 (scalar mass operator) 613::: 614 615(v0-2)= 616 617## v0.2 (Mar 30, 2018) 618 619libCEED was made publicly available the first full CEED software distribution, release 620CEED 1.0. The distribution was made available using the Spack package manager to provide 621a common, easy-to-use build environment, where the user can build the CEED distribution 622with all dependencies. This release included a new Fortran interface for the library. 623This release also contained major improvements in the OCCA backend (including a new 624`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to 625compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an 626`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application 627of the Laplace operator) was also added to this release. 628 629Backends available in this release: 630 631| CEED resource (`-ceed`) | Backend | 632|-------------------------|---------------------------------| 633| `/cpu/self` | Serial reference implementation | 634| `/cpu/occa` | Serial OCCA kernels | 635| `/gpu/occa` | CUDA OCCA kernels | 636| `/omp/occa` | OpenMP OCCA kernels | 637| `/ocl/occa` | OpenCL OCCA kernels | 638 639Examples available in this release: 640 641:::{list-table} 642:header-rows: 1 643:widths: auto 644* - User code 645 - Example 646* - `ceed` 647 - * ex1 (volume) 648* - `mfem` 649 - * BP1 (scalar mass operator) 650 * BP3 (scalar Laplace operator) 651* - `petsc` 652 - * BP1 (scalar mass operator) 653* - `nek5000` 654 - * BP1 (scalar mass operator) 655::: 656 657(v0-1)= 658 659## v0.1 (Jan 3, 2018) 660 661Initial low-level API of the CEED project. The low-level API provides a set of Finite 662Elements kernels and components for writing new low-level kernels. Examples include: 663vector and sparse linear algebra, element matrix assembly over a batch of elements, 664partial assembly and action for efficient high-order operators like mass, diffusion, 665advection, etc. The main goal of the low-level API is to establish the basis for the 666high-level API. Also, identifying such low-level kernels and providing a reference 667implementation for them serves as the basis for specialized backend implementations. 668This release contained several backends: `/cpu/self`, and backends which rely upon the 669[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`, 670`/gpu/occa`, and `/omp/occa`. 671It also included several examples, in the `examples` folder: 672A standalone code that shows the usage of libCEED (with no external 673dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1 674(with the application of the mass operator); and a `petsc` example to perform BP1 675(with the application of the mass operator). 676 677Backends available in this release: 678 679| CEED resource (`-ceed`) | Backend | 680|-------------------------|---------------------------------| 681| `/cpu/self` | Serial reference implementation | 682| `/cpu/occa` | Serial OCCA kernels | 683| `/gpu/occa` | CUDA OCCA kernels | 684| `/omp/occa` | OpenMP OCCA kernels | 685 686Examples available in this release: 687 688| User code | Example | 689|-----------------------|-----------------------------------| 690| `ceed` | ex1 (scalar Laplace operator) | 691| `mfem` | BP1 (scalar mass operator) | 692| `petsc` | BP1 (scalar mass operator) | 693``` 694