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- Change QFunction source include file handling in JiT compilers 15 - Add `CEED_RUNNING_JIT_PASS` compiler definition for wrapping header files that device JiT compilers cannot read 16 - Users should now prefer `#include <ceed/types.h>` rather than `#include <ceed.h>` in QFunction source files 17- Require use of `Ceed*Destroy()` on Ceed objects returned from `Ceed*Get*()`. 18- Rename `CeedCompositeOperatorCreate()` to `CeedOperatorCreateComposite()` for uniformity. 19- Rename `CeedCompositeOperator*()` to `CeedOperatorComposite*()` for uniformity. 20 21### New features 22 23- Add `CeedOperatorCreateAtPoints` which evaluates the `CeedQFunction` at arbitrary locations in each element, for use in Particle in Cell, Material Point Method, and similar methods. 24- Add `CeedElemRestrictionGetLLayout` to provide L-vector layout for strided `CeedElemRestriction` created with `CEED_BACKEND_STRIDES`. 25- Add `CeedVectorReturnCeed` and similar when parent `Ceed` context for a libCEED object is only needed once in a calling scope. 26- Enable `#pragma once` for all JiT source; remove duplicate includes in JiT source string before compilation. 27- Allow user to set additional compiler options for CUDA and HIP JiT. 28Specifically, 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`. 29- Added non-tensor basis support to code generation backends `/gpu/cuda/gen` and `/gpu/hip/gen`. 30- Added support to code generation backends `/gpu/cuda/gen` and `/gpu/hip/gen` for operators with both tensor and non-tensor bases. 31- Add `CeedGetGitVersion()` to access the Git commit and dirty state of the repository at build time. 32- Add `CeedGetBuildConfiguration()` to access compilers, flags, and related information about the build environment. 33 34### Examples 35 36- Add deal.II example with CEED BP suite. 37 38### Maintainability 39 40- OCCA backends were retired. 41 42(v0-12)= 43 44## v0.12 (Oct 31, 2023) 45 46### Interface changes 47 48- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency. 49For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`. 50- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface. 51 52### New features 53 54- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name. 55- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions. 56- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases. 57- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`. 58- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements. 59- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`. 60- 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)). 61- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`. 62- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads. 63 64### Examples 65 66- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh. 67 68#### {ref}`example-petsc-bps` 69 70- Requires PETSc version 3.19 or later. 71 72#### {ref}`example-petsc-navier-stokes` 73 74- Updated restart and checkpointing interface. 75- Add data-driven subgrid-stress model. 76- Add differential filtering of solution. 77- Add turbulence statistics collection over spanwise-symmetric geometries. 78- Add Taylor-Green vortex initial condition. 79- Add Riemann-based outflow boundary conditions. 80- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer. 81- Add Internal Damping Layer (IDL) for helping turbulent simulation stability. 82- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates. 83 84(v0-11)= 85 86## v0.11 (Dec 24, 2022) 87 88### Interface changes 89 90- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output. 91- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`. 92- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface. 93- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity. 94Some 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. 95 96### New features 97 98- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user. 99- Improved support for $H(\text{div})$ bases. 100- Added `CeedInt_FMT` to support potential future use of larger integer sizes. 101- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`. 102- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled. 103Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend. 104This issue will be fixed in a future OCCA release. 105 106### Bugfix 107 108- Fix bug in setting device id for GPU backends. 109- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends. 110- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`. 111- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes. 112 113### Examples 114 115#### {ref}`example-petsc-navier-stokes` 116 117- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs. 118- Refactored to improve code reuse and modularity. 119- Support for primitive variables for more accurate boundary layers and all-speed flow. 120- Added $YZ\beta$ shock capturing scheme and Shock Tube example. 121- Added Channel example, with comparison to analytic solutions. 122- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations. 123- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions. 124- Added "freestream" boundary conditions based on HLLC Riemann solver. 125- Automated stabilization coefficients for different basis degree. 126 127#### {ref}`example-petsc-bps` 128 129- Support for convergence studies. 130 131### Maintainability 132 133- 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. 134- Enabled support for `p > 8` for `/gpu/*/shared` backends. 135- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`. 136- Improved test harness. 137 138(v0-10-1)= 139 140## v0.10.1 (Apr 11, 2022) 141 142### Interface changes 143 144- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application. 145 146### New features 147 148- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends). 149This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work. 150 151### Bugfix 152 153- Install JiT source files in install directory to fix GPU functionality for installed libCEED. 154 155(v0-10)= 156 157## v0.10 (Mar 21, 2022) 158 159### Interface changes 160 161- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields. 162- 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`. 163- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added. 164- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients. 165- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces. 166- Warning added when compiling OCCA backend to alert users that this backend is experimental. 167- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`. 168- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name. 169- 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`. 170- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`. 171- 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. 172- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`. 173- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums. 174- 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. 175- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy. 176- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator. 177 178### New features 179 180- `CeedScalar` can now be set as `float` or `double` at compile time. 181- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends. 182- 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. 183- Remove need to guard library headers in QFunction source for code generation backends. 184- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present. 185- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally. 186- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file. 187- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor. 188- 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. 189- Added support for element matrix assembly in GPU backends. 190 191### Maintainability 192 193- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage. 194- `Include-what-you-use` makefile target added as `make iwyu`. 195- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase. 196- Put GPU JiTed kernel source code into separate files. 197- 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. 198 199(v0-9)= 200 201## v0.9 (Jul 6, 2021) 202 203### Interface changes 204 205- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact. 206 207### New features 208 209- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use. 210- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends. 211- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations. 212- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed). 213 214### Performance improvements 215 216### Examples 217 218- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations. 219- Fluid mechanics example adds GPU support and improves modularity. 220 221### Deprecated backends 222 223- 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. 224 225(v0-8)= 226 227## v0.8 (Mar 31, 2021) 228 229### Interface changes 230 231- Error handling improved to include enumerated error codes for C interface return values. 232- 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. 233 234### New features 235 236- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features. 237- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly. 238- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators. 239 240### Performance improvements 241 242- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`. 243- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`. 244 245### Examples 246 247- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions. 248- {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. 249- {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. 250- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree. 251 252(v0-7)= 253 254## v0.7 (Sep 29, 2020) 255 256### Interface changes 257 258- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors. 259 As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`. 260 These changes improve support for mixed finite element methods. 261- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature. 262- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity. 263- 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. 264- 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. 265- 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. 266- 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. 267 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`. 268- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory. 269- 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. 270 271### New features 272 273- New HIP backend: `/gpu/hip/ref`. 274- CeedQFunction support for user `CUfunction`s in some backends 275 276### Performance improvements 277 278- OCCA backend rebuilt to facilitate future performance enhancements. 279- PETSc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`. 280 281### Examples 282 283- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions. 284 285### Deprecated backends 286 287- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`. 288 289(v0-6)= 290 291## v0.6 (Mar 29, 2020) 292 293libCEED v0.6 contains numerous new features and examples, as well as expanded 294documentation in [this new website](https://libceed.org). 295 296### New features 297 298- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly 299 1-1 correspondence with the C interface, plus some convenience features. For instance, 300 data stored in the {cpp:type}`CeedVector` structure are available without copy as 301 {py:class}`numpy.ndarray`. Short tutorials are provided in 302 [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/). 303- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point, 304 {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal 305 ({c:func}`CeedOperatorAssembleLinearDiagonal`). These operations are useful for 306 preconditioning ingredients and are used in the libCEED's multigrid examples. 307- The inverse of separable operators can be obtained using 308 {c:func}`CeedOperatorCreateFDMElementInverse` and applied with 309 {c:func}`CeedOperatorApply`. This is a useful preconditioning ingredient, 310 especially for Laplacians and related operators. 311- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`, 312 {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`. 313- Make public accessors for various attributes to facilitate writing composable code. 314- New backend: `/cpu/self/memcheck/serial`. 315- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA 316 backends. (Single source is coming soon for OCCA backends.) 317- Fix some missing edge cases in CUDA backend. 318 319### Performance Improvements 320 321- MAGMA backend performance optimization and non-tensor bases. 322- No-copy optimization in {c:func}`CeedOperatorApply`. 323 324### Interface changes 325 326- Replace {code}`CeedElemRestrictionCreateIdentity` and 327 {code}`CeedElemRestrictionCreateBlocked` with more flexible 328 {c:func}`CeedElemRestrictionCreateStrided` and 329 {c:func}`CeedElemRestrictionCreateBlockedStrided`. 330- Add arguments to {c:func}`CeedQFunctionCreateIdentity`. 331- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification 332 with {cpp:enum}`CeedInterlaceMode`. This is now an attribute of the 333 {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no 334 longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and 335 {c:func}`CeedElemRestrictionApply`. 336 337### Examples 338 339libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`. 340Notable additions include: 341 342- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of 343 a domain in 1, 2, and 3 dimensions by applying a Laplacian. 344 345- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area 346 of domains (like the cube and sphere) by direct integration on a surface mesh; 347 demonstrates geometric dimension different from topological dimension. 348 349- PETSc {ref}`example-petsc-bps`: 350 351 - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support. 352 - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements 353 and transparent CUDA support. 354 - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`): 355 generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric 356 dimension different from topological dimension. 357 358- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid 359 solver with algebraic multigrid coarse solve. 360 361- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly 362 `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`), 363 implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and 364 quasi-2D computational domain support. 365 366- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for 367 linear elasticity, small-strain hyperelasticity, and globalized finite-strain 368 hyperelasticity using p-multigrid with algebraic multigrid coarse solve. 369 370(v0-5)= 371 372## v0.5 (Sep 18, 2019) 373 374For this release, several improvements were made. Two new CUDA backends were added to 375the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art 376performance using single-source {ref}`CeedQFunction`. From this release, users 377can define Q-Functions in a single source code independently of the targeted backend 378with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU 379compilation of the user provided {ref}`CeedQFunction` code. To allow a unified 380declaration, the {ref}`CeedQFunction` API has undergone a slight change: 381the `QFunctionField` parameter `ncomp` has been changed to `size`. This change 382requires setting the previous value of `ncomp` to `ncomp*dim` when adding a 383`QFunctionField` with eval mode `CEED EVAL GRAD`. 384 385Additionally, new CPU backends 386were included in this release, such as the `/cpu/self/opt/*` backends (which are 387written in pure C and use partial **E-vectors** to improve performance) and the 388`/cpu/self/ref/memcheck` backend (which relies upon the 389[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user 390{ref}`CeedQFunction` have no undefined values). 391This release also included various performance improvements, bug fixes, new examples, 392and improved tests. Among these improvements, vectorized instructions for 393{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD` 394instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and 395identity Q-Functions were introduced, and the PETSc benchmark problems were expanded 396to include unstructured meshes handling were. For this expansion, the prior version of 397the PETSc BPs, which only included data associated with structured geometries, were 398renamed `bpsraw`, and the new version of the BPs, which can handle data associated 399with any unstructured geometry, were called `bps`. Additionally, other benchmark 400problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively), 401and BP5 and BP6 (the collocated versions---for which the quadrature points are the same 402as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc 403examples. Furthermoew, another standalone libCEED example, called `ex2`, which 404computes the surface area of a given mesh was added to this release. 405 406Backends available in this release: 407 408| CEED resource (`-ceed`) | Backend | 409|--------------------------|-----------------------------------------------------| 410| `/cpu/self/ref/serial` | Serial reference implementation | 411| `/cpu/self/ref/blocked` | Blocked reference implementation | 412| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks | 413| `/cpu/self/opt/serial` | Serial optimized C implementation | 414| `/cpu/self/opt/blocked` | Blocked optimized C implementation | 415| `/cpu/self/avx/serial` | Serial AVX implementation | 416| `/cpu/self/avx/blocked` | Blocked AVX implementation | 417| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 418| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 419| `/cpu/occa` | Serial OCCA kernels | 420| `/gpu/occa` | CUDA OCCA kernels | 421| `/omp/occa` | OpenMP OCCA kernels | 422| `/ocl/occa` | OpenCL OCCA kernels | 423| `/gpu/cuda/ref` | Reference pure CUDA kernels | 424| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 425| `/gpu/cuda/shared` | Optimized pure CUDA kernels using shared memory | 426| `/gpu/cuda/gen` | Optimized pure CUDA kernels using code generation | 427| `/gpu/magma` | CUDA MAGMA kernels | 428 429Examples available in this release: 430 431:::{list-table} 432:header-rows: 1 433:widths: auto 434* - User code 435 - Example 436* - `ceed` 437 - * ex1 (volume) 438 * ex2 (surface) 439* - `mfem` 440 - * BP1 (scalar mass operator) 441 * BP3 (scalar Laplace operator) 442* - `petsc` 443 - * BP1 (scalar mass operator) 444 * BP2 (vector mass operator) 445 * BP3 (scalar Laplace operator) 446 * BP4 (vector Laplace operator) 447 * BP5 (collocated scalar Laplace operator) 448 * BP6 (collocated vector Laplace operator) 449 * Navier-Stokes 450* - `nek5000` 451 - * BP1 (scalar mass operator) 452 * BP3 (scalar Laplace operator) 453::: 454 455(v0-4)= 456 457## v0.4 (Apr 1, 2019) 458 459libCEED v0.4 was made again publicly available in the second full CEED software 460distribution, release CEED 2.0. This release contained notable features, such as 461four new CPU backends, two new GPU backends, CPU backend optimizations, initial 462support for operator composition, performance benchmarking, and a Navier-Stokes demo. 463The new CPU backends in this release came in two families. The `/cpu/self/*/serial` 464backends process one element at a time and are intended for meshes with a smaller number 465of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of 466eight interlaced elements and are intended for meshes with higher numbers of elements. 467The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU 468performance. The `/cpu/self/xsmm/*` backends rely upon the 469[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU 470performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA. 471The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable 472performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple 473parallelization approach, where each thread treats a finite element. Using just in time 474compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this 475backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend 476achieve good peak performance for 1D, 2D, and low order 3D problems, but performance 477deteriorates very quickly when threads run out of registers. 478 479A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED 480examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`). 481This example solves the time-dependent Navier-Stokes equations of compressible gas 482dynamics in a static Eulerian three-dimensional frame, using structured high-order 483finite/spectral element spatial discretizations and explicit high-order time-stepping 484(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc, 485so that the pointwise physics (defined at quadrature points) is separated from the 486parallelization and meshing concerns. 487 488Backends available in this release: 489 490| CEED resource (`-ceed`) | Backend | 491|--------------------------|-----------------------------------------------------| 492| `/cpu/self/ref/serial` | Serial reference implementation | 493| `/cpu/self/ref/blocked` | Blocked reference implementation | 494| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 495| `/cpu/self/avx/serial` | Serial AVX implementation | 496| `/cpu/self/avx/blocked` | Blocked AVX implementation | 497| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 498| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 499| `/cpu/occa` | Serial OCCA kernels | 500| `/gpu/occa` | CUDA OCCA kernels | 501| `/omp/occa` | OpenMP OCCA kernels | 502| `/ocl/occa` | OpenCL OCCA kernels | 503| `/gpu/cuda/ref` | Reference pure CUDA kernels | 504| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 505| `/gpu/magma` | CUDA MAGMA kernels | 506 507Examples available in this release: 508 509:::{list-table} 510:header-rows: 1 511:widths: auto 512* - User code 513 - Example 514* - `ceed` 515 - * ex1 (volume) 516* - `mfem` 517 - * BP1 (scalar mass operator) 518 * BP3 (scalar Laplace operator) 519* - `petsc` 520 - * BP1 (scalar mass operator) 521 * BP3 (scalar Laplace operator) 522 * Navier-Stokes 523* - `nek5000` 524 - * BP1 (scalar mass operator) 525 * BP3 (scalar Laplace operator) 526::: 527 528(v0-3)= 529 530## v0.3 (Sep 30, 2018) 531 532Notable features in this release include active/passive field interface, support for 533non-tensor bases, backend optimization, and improved Fortran interface. This release 534also focused on providing improved continuous integration, and many new tests with code 535coverage reports of about 90%. This release also provided a significant change to the 536public interface: a {ref}`CeedQFunction` can take any number of named input and output 537arguments while {ref}`CeedOperator` connects them to the actual data, which may be 538supplied explicitly to `CeedOperatorApply()` (active) or separately via 539`CeedOperatorSetField()` (passive). This interface change enables reusable libraries 540of CeedQFunctions and composition of block solvers constructed using 541{ref}`CeedOperator`. A concept of blocked restriction was added to this release and 542used in an optimized CPU backend. Although this is typically not visible to the user, 543it enables effective use of arbitrary-length SIMD while maintaining cache locality. 544This CPU backend also implements an algebraic factorization of tensor product gradients 545to perform fewer operations than standard application of interpolation and 546differentiation from nodes to quadrature points. This algebraic formulation 547automatically supports non-polynomial and non-interpolatory bases, thus is more general 548than the more common derivation in terms of Lagrange polynomials on the quadrature points. 549 550Backends available in this release: 551 552| CEED resource (`-ceed`) | Backend | 553|-------------------------|-----------------------------------------------------| 554| `/cpu/self/blocked` | Blocked reference implementation | 555| `/cpu/self/ref` | Serial reference implementation | 556| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 557| `/cpu/occa` | Serial OCCA kernels | 558| `/gpu/occa` | CUDA OCCA kernels | 559| `/omp/occa` | OpenMP OCCA kernels | 560| `/ocl/occa` | OpenCL OCCA kernels | 561| `/gpu/magma` | CUDA MAGMA kernels | 562 563Examples available in this release: 564 565:::{list-table} 566:header-rows: 1 567:widths: auto 568* - User code 569 - Example 570* - `ceed` 571 - * ex1 (volume) 572* - `mfem` 573 - * BP1 (scalar mass operator) 574 * BP3 (scalar Laplace operator) 575* - `petsc` 576 - * BP1 (scalar mass operator) 577 * BP3 (scalar Laplace operator) 578* - `nek5000` 579 - * BP1 (scalar mass operator) 580 * BP3 (scalar Laplace operator) 581::: 582 583(v0-21)= 584 585## v0.21 (Sep 30, 2018) 586 587A MAGMA backend (which relies upon the 588[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this 589release. This initial integration set up the framework of using MAGMA and provided the 590libCEED functionality through MAGMA kernels as one of libCEED’s computational backends. 591As any other backend, the MAGMA backend provides extended basic data structures for 592{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements 593the fundamental CEED building blocks to work with the new data structures. 594In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data 595but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled 596internally, and thus seamlessly to the user, through the functions/methods that are 597provided to support them. 598 599Backends available in this release: 600 601| CEED resource (`-ceed`) | Backend | 602|-------------------------|---------------------------------| 603| `/cpu/self` | Serial reference implementation | 604| `/cpu/occa` | Serial OCCA kernels | 605| `/gpu/occa` | CUDA OCCA kernels | 606| `/omp/occa` | OpenMP OCCA kernels | 607| `/ocl/occa` | OpenCL OCCA kernels | 608| `/gpu/magma` | CUDA MAGMA kernels | 609 610Examples available in this release: 611 612:::{list-table} 613:header-rows: 1 614:widths: auto 615* - User code 616 - Example 617* - `ceed` 618 - * ex1 (volume) 619* - `mfem` 620 - * BP1 (scalar mass operator) 621 * BP3 (scalar Laplace operator) 622* - `petsc` 623 - * BP1 (scalar mass operator) 624* - `nek5000` 625 - * BP1 (scalar mass operator) 626::: 627 628(v0-2)= 629 630## v0.2 (Mar 30, 2018) 631 632libCEED was made publicly available the first full CEED software distribution, release 633CEED 1.0. The distribution was made available using the Spack package manager to provide 634a common, easy-to-use build environment, where the user can build the CEED distribution 635with all dependencies. This release included a new Fortran interface for the library. 636This release also contained major improvements in the OCCA backend (including a new 637`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to 638compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an 639`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application 640of the Laplace operator) was also added to this release. 641 642Backends available in this release: 643 644| CEED resource (`-ceed`) | Backend | 645|-------------------------|---------------------------------| 646| `/cpu/self` | Serial reference implementation | 647| `/cpu/occa` | Serial OCCA kernels | 648| `/gpu/occa` | CUDA OCCA kernels | 649| `/omp/occa` | OpenMP OCCA kernels | 650| `/ocl/occa` | OpenCL OCCA kernels | 651 652Examples available in this release: 653 654:::{list-table} 655:header-rows: 1 656:widths: auto 657* - User code 658 - Example 659* - `ceed` 660 - * ex1 (volume) 661* - `mfem` 662 - * BP1 (scalar mass operator) 663 * BP3 (scalar Laplace operator) 664* - `petsc` 665 - * BP1 (scalar mass operator) 666* - `nek5000` 667 - * BP1 (scalar mass operator) 668::: 669 670(v0-1)= 671 672## v0.1 (Jan 3, 2018) 673 674Initial low-level API of the CEED project. The low-level API provides a set of Finite 675Elements kernels and components for writing new low-level kernels. Examples include: 676vector and sparse linear algebra, element matrix assembly over a batch of elements, 677partial assembly and action for efficient high-order operators like mass, diffusion, 678advection, etc. The main goal of the low-level API is to establish the basis for the 679high-level API. Also, identifying such low-level kernels and providing a reference 680implementation for them serves as the basis for specialized backend implementations. 681This release contained several backends: `/cpu/self`, and backends which rely upon the 682[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`, 683`/gpu/occa`, and `/omp/occa`. 684It also included several examples, in the `examples` folder: 685A standalone code that shows the usage of libCEED (with no external 686dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1 687(with the application of the mass operator); and a `petsc` example to perform BP1 688(with the application of the mass operator). 689 690Backends available in this release: 691 692| CEED resource (`-ceed`) | Backend | 693|-------------------------|---------------------------------| 694| `/cpu/self` | Serial reference implementation | 695| `/cpu/occa` | Serial OCCA kernels | 696| `/gpu/occa` | CUDA OCCA kernels | 697| `/omp/occa` | OpenMP OCCA kernels | 698 699Examples available in this release: 700 701| User code | Example | 702|-----------------------|-----------------------------------| 703| `ceed` | ex1 (scalar Laplace operator) | 704| `mfem` | BP1 (scalar mass operator) | 705| `petsc` | BP1 (scalar mass operator) | 706``` 707