feat(hipblas): add the generated Fortran binding in bindings/fortran/hipblas by amontoison · Pull Request #12370 · ROCm/rocm-libraries

amontoison · 2026-09-21T15:35:39Z

Implements RFC0015, Fortran Bindings Split (update) for hipBLAS: bindings/fortran/hipblas/ is a standalone CMake project that builds the generated Fortran binding against an installed hipBLAS. The library is untouched: no option, no add_subdirectory, no change to projects/hipblas/.

Draft until ROCm/rocm-systems#11923 (HIP Fortran binding, which the tests use) merges.

What lands

bindings/fortran/hipblas/hipblas.F90 (163 733 lines, generated with ROCm/rocm-fortran-generator 9c8b3e4 from the hipBLAS headers, marked linguist-generated): one module hipblas with the bind(C) interfaces, the enum constants, the hipblasCheck helper and array overloads for the routines the generator supports.
metadata/hipblas.toml: the generator metadata it was generated with.
test/: 209 CTest tests, 126 .f03 tests of the C interfaces and 83 .f08 tests of the array overloads (registered as <name>_f2008). They also install: the executables in bin/, and a CTestTestfile.cmake with the same tests and properties in bin/hipblas-fortran/, for ctest --test-dir.
The tests are ported from hipFORT's hipBLAS tests.
CMakeLists.txt, the hipblas-fortran config template, cmake/ROCmFortranCompiler.cmake (the same in every binding: ROCm's amdflang is preferred over the first compiler on PATH unless CMAKE_Fortran_COMPILER or FC is set), cmake/toolchains/ and the binding's CHANGELOG.md.
.github/CODEOWNERS: /bindings/fortran/hipblas/ @ROCm/rocm-fortran-maintainers @ROCm/hipblas-reviewers.
.github/repos-config.json: a bindings/fortran entry, so CI detects changes to the binding. Each binding PR appends its entry at the end of the list, so all but the first to merge need a trivial rebase.

A consumer uses find_package(hipblas-fortran) and links roc::hipblas_fortran. The archive and .mod files install per compiler under lib/fortran/<compiler>/ and include/fortran/<compiler>/; HIPBLAS_FORTRAN_COMPILER_DIR names that directory. The archive has one section per procedure, so a program linked with -Wl,--gc-sections keeps only the wrappers it calls.

Options

Option	Default	Effect
`BUILD_FORTRAN_BINDINGS`	`${UNIX}`	builds the binding; skipped without a Fortran compiler
`BUILD_FORTRAN_CLIENTS`	`ON`	builds and installs the tests in `test/`; off under `BUILD_ADDRESS_SANITIZER`
`FORTRAN_ARRAY_INTERFACES`	`assumed-shape`	`none` / `assumed-shape` / `assumed-rank` array overloads

In TheRock

ROCm/TheRock#9446 builds bindings/fortran/hipblas as the hipblas-fortran subproject against the hipBLAS it built, with -DBUILD_FORTRAN_BINDINGS=ON -DBUILD_FORTRAN_CLIENTS=${THEROCK_BUILD_TESTING}, ships it in the blas-fortran artifact and runs the installed tests in its hipblas-fortran GPU test job.

Follow-ups

This binding replaces the hand-written library/src/hipblas_module.f90, which stays for now. Deprecating it, then removing it and moving the library's Fortran clients to this binding, come in separate pull requests.

Build and test

From the repository root, with $PREFIX the install prefix:

lib=hipblas
cmake -S bindings/fortran/$lib -B build/$lib-fortran \
  -DCMAKE_PREFIX_PATH="/opt/rocm;$HIP_FORTRAN_PREFIX" -DCMAKE_INSTALL_PREFIX="$PREFIX"
cmake --build build/$lib-fortran
ctest --test-dir build/$lib-fortran
cmake --install build/$lib-fortran
ctest --test-dir "$PREFIX/bin/$lib-fortran"

$HIP_FORTRAN_PREFIX holds the HIP Fortran binding (hip-fortran) built with the same compiler; without it the tests are skipped.

Verification

Standalone against ROCm 10.1 with amdflang on one GPU: 209 tests, 209 pass.

JIRA ID: ROCM-31390

therock-pr-bot · 2026-09-21T15:40:59Z

✅ All Checks Passed — Ready for Review

Check	Status	Details
📝 PR Description	✅ Pass	—
⛔ Forbidden Files	✅ Pass	—
🧪 Unit Test	✅ Pass	PR does not contain code files — Unit Test auto-passed
🔎 pre-commit	✅ Pass	—
🚫 Draft PR	🔜 To Be Enabled	—
🚩 Feature Flag	🔜 To Be Enabled	—
📊 Code Coverage	🔜 To Be Enabled	—
🤖 therock-pr-bot	✅ Pass	—

🎉 All checks passed! This PR is ready for review.

📖 Need help? See the Policy FAQ for details on every check and how to fix failures.

🙋 Wish to Override Policy?

therock-pr-bot · 2026-09-21T15:41:02Z

🎉 All checks passed! This PR is ready for review.

…gets A library that ships generated Fortran bindings installs them as their own config package, <package>-fortran-config.cmake, so that the binding can be built and installed independently of the C library -- which is the supported path for anyone whose Fortran compiler is not the one ROCm ships the bindings for. For find_package(<package>) to then expose <ns>::<package>_fortran, the C package has to pick that file up as a sibling. rocm_export_targets already has an INCLUDE argument, and it is not usable for this, for two independent reasons: - it emits a bare include(), so the file must always exist, whereas a Fortran binding is present only when BUILD_FORTRAN_BINDINGS was on and a Fortran compiler was available; and - it emits it BEFORE the targets file, so the Fortran config would run before the imported C targets exist, and the Fortran interface library could not link against them. So emit a separate include, after the targets and marked OPTIONAL. It is a no-op for every package that ships no such file, which is all of them today, and it lets a Fortran binding installed after the C library be found at find_package(<package>) time and ignored when absent. Guarded against the case where a caller already passed the same file through INCLUDE, so it is not included twice. rocm_write_package_template_function joins ARGN with spaces into NAME(...), so the trailing OPTIONAL keyword passes through verbatim; no new emission path is needed. Verified, the emitted line is include("@PACKAGE_CONFIG_PACKAGE_INSTALL_DIR@/rocblas-fortran-config.cmake" OPTIONAL) using the same @PACKAGE_...@ substitution the targets include already relies on, and include(... OPTIONAL) on a missing file is confirmed to be a clean no-op. Needed by the RFC0014 Fortran bindings rollout in rocm-libraries and rocm-systems (ROCm/rocm-libraries#12370, #12505, #12506, #12514, #12533).

codecov · 2026-09-28T23:43:49Z

Codecov Report

✅ All modified and coverable lines are covered by tests.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop   #12370   +/-   ##
========================================
  Coverage    75.95%   75.95%           
========================================
  Files         2880     2880           
  Lines       479025   479025           
  Branches     71958    71958           
========================================
  Hits        363828   363828           
  Misses       90338    90338           
  Partials     24859    24859

Flag	Coverage Δ	*Carryforward flag
TensileLite-CPP	`48.93% <ø> (ø)`	Carriedforward from 303be95
TensileLite-Unit	`83.68% <ø> (ø)`	Carriedforward from 303be95
hipBLAS	`90.62% <ø> (ø)`
hipBLASLt	`64.43% <ø> (ø)`	Carriedforward from 303be95
hipCUB	`82.68% <ø> (ø)`	Carriedforward from 303be95
hipDNN	`86.78% <ø> (ø)`	Carriedforward from 303be95
hipFFT	`42.00% <ø> (ø)`	Carriedforward from 303be95
hipRAND	`76.12% <ø> (ø)`	Carriedforward from 303be95
hipSOLVER	`69.67% <ø> (ø)`	Carriedforward from 303be95
hipSPARSE	`86.99% <ø> (ø)`	Carriedforward from 303be95
rocBLAS	`68.73% <ø> (ø)`	Carriedforward from 303be95
rocFFT	`48.10% <ø> (ø)`	Carriedforward from 303be95
rocRAND	`57.30% <ø> (ø)`	Carriedforward from 303be95
rocSOLVER	`76.74% <ø> (ø)`	Carriedforward from 303be95
rocSPARSE	`76.92% <ø> (ø)`	Carriedforward from 303be95
rocThrust	`93.15% <ø> (ø)`	Carriedforward from 303be95

*This pull request uses carry forward flags. Click here to find out more.
see 1 file with indirect coverage changes

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

Add fortran/hipblas.F90, the hipblas module generated from the C headers by the rocm-fortran generator under RFC0019 (ROCm/TheRock#8122). It is committed next to the headers, is not edited by hand, and .gitattributes marks it linguist-generated.

…ted one Build fortran/hipblas.F90 as roc::hipblas_fortran from fortran/, a CMake project of its own that the root adds. BUILD_FORTRAN_BINDINGS (default ${UNIX}, OFF on the CUDA backend) and FORTRAN_ARRAY_INTERFACES (default assumed-shape) control it; the existing BUILD_FORTRAN_CLIENTS is forced OFF when no binding is built. HIPBLAS_BUILD_FORTRAN_* override both per library. The .mod and the archive install per compiler under <includedir>/fortran/<compiler> and <libdir>/fortran/<compiler>, beside a hipblas-fortran config package. Remove library/src/hipblas_module.f90. Its hipblas_enums module is folded into hipblas, so use hipblas_enums becomes use hipblas, and hipblasCreate and the getters take their output by reference. The clients shims and samples use the generated module; a client-only hip_datatypes_stub.F90 stands in for the HIP binding until ROCm/rocm-systems#11923 lands. There is no fortran/test suite. Refs: ROCm/TheRock#8122 (RFC0019)

The hipBLAS Fortran binding consumes hipBLAS's public API and builds standalone against an installed hipBLAS, so it moves out of the library into bindings/fortran/hipblas, the tree every monorepo and TheRock use for the Fortran bindings (RFC0015). Nothing is wired into the library build any more: the library is back to its state on develop. Its hand-written Fortran module stays for now: its deprecation and removal, with the Fortran clients that use it, follow in separate pull requests. - bindings/fortran/hipblas reads hipBLAS's version from projects/hipblas - the binding has its own .gitattributes and CHANGELOG.md - CODEOWNERS: /bindings/fortran/hipblas/, same owners - repos-config.json: register bindings/fortran/hipblas so CI detects changes

…tests - regenerate bindings/fortran/hipblas with rocm-fortran-generator 9c8b3e4 - commit its metadata in bindings/fortran/hipblas/metadata/, where the generator reads it from a checkout - add the binding's tests in test/, under BUILD_FORTRAN_CLIENTS, ported from hipFORT's hipBLAS tests: the library's own clients still test the hand-written module while it is deprecated, so they did not test the binding - drop hip_datatypes_stub.F90, a stand-in for the HIP module that the library's clients used and nothing builds any more

Install the test executables in bin/, and a CTestTestfile.cmake in bin/hipblas-fortran/ with the same tests and properties as the build tree, so that ctest --test-dir runs them from an install, as TheRock's test jobs do for the libraries.

The binding is built against an installed hipBLAS and is never added to the library's build, so drop what only served that: the per-library BUILD_FORTRAN_* overrides, the PARENT_SCOPE results, the ROCM_HAVE_FORTRAN shortcut, the gating on the library's client and test options, and the comments about the root. Find the Fortran compiler through the cmake/ROCmFortranCompiler.cmake shared by all the bindings, and require CMake 3.19. The binding targets the ROCm backend only, so BUILD_FORTRAN_BINDINGS no longer defaults to OFF on the CUDA backend.

…-fortran The package files went to lib/cmake/hipblas, the C library's directory, with a forwarder in lib/cmake/hipblas-fortran. rocRAND and hipRAND install their own legacy <lib>-fortran-config.cmake there, so the files overlapped. Install the config, targets and version files directly where find_package(<package>) looks, as the HIP and ROCTx bindings do, and drop the forwarder.

…ut recursion rocm-cmake's generated hipblas-config.cmake is to include lib/cmake/hipblas-fortran/hipblas-fortran-config.cmake after its targets (ROCm/rocm-cmake#303). Skip find_dependency(hipblas) when hipblas is the package being found, not only when roc::hipblas is already imported, so that include never re-enters hipblas-config.cmake.

github-actions Bot added the project: hipblas label Sep 21, 2026

assistant-librarian Bot added the organization: ROCm label Sep 21, 2026

amontoison force-pushed the users/amontoison/hipblas-fortran-bindings branch from 0f51aca to 1affc44 Compare September 21, 2026 15:40

amontoison force-pushed the users/amontoison/hipblas-fortran-bindings branch from 1affc44 to cbe1821 Compare September 24, 2026 16:49

github-actions Bot added the documentation label Sep 24, 2026

amontoison mentioned this pull request Sep 24, 2026

feat: include <package>-fortran/<package>-fortran-config.cmake, OPTIONAL, after the targets ROCm/rocm-cmake#303

Open

amontoison force-pushed the users/amontoison/hipblas-fortran-bindings branch 3 times, most recently from 02b3d7f to e601d68 Compare September 25, 2026 02:04

This was referenced Sep 25, 2026

feat(hipsolver): add the generated Fortran binding in bindings/fortran/hipsolver #12581

Draft

feat(rocsolver): add the generated Fortran binding in bindings/fortran/rocsolver #12582

Draft

amontoison changed the title ~~feat(hipblas): add generated Fortran bindings (RFC0014)~~ feat(hipblas): replace the hand-written Fortran binding with a generated one (RFC0014) Sep 25, 2026

amontoison mentioned this pull request Sep 26, 2026

feat(hipsparse): add the generated Fortran binding in bindings/fortran/hipsparse #12368

Draft

amontoison force-pushed the users/amontoison/hipblas-fortran-bindings branch from e601d68 to 63ed4d6 Compare September 28, 2026 03:47

amontoison mentioned this pull request Sep 28, 2026

[Sandbox] Move the Fortran bindings to per-library modules, with hipfftxt beside hipfft (ROCm only) ROCm/hipfort#540

Open

amontoison force-pushed the users/amontoison/hipblas-fortran-bindings branch 2 times, most recently from 48eea04 to b0dffaa Compare September 28, 2026 19:42

amontoison force-pushed the users/amontoison/hipblas-fortran-bindings branch 2 times, most recently from e491e5b to 08c9e40 Compare October 4, 2026 03:45

amontoison mentioned this pull request Oct 4, 2026

Add migration guides for the ROCm-packaged Fortran bindings and from CUDA Fortran, and deprecate hipFORT ROCm/hipfort#548

Draft

amontoison changed the title ~~feat(hipblas): replace the hand-written Fortran binding with a generated one (RFC0014)~~ feat(hipblas): replace the hand-written Fortran binding with a generated one Oct 4, 2026

amontoison force-pushed the users/amontoison/hipblas-fortran-bindings branch from 08c9e40 to 8a970ae Compare October 5, 2026 05:12

amontoison mentioned this pull request Oct 5, 2026

feat(fortran): add generated Fortran bindings for HIP ROCm/rocm-systems#11923

Open

amontoison mentioned this pull request Oct 21, 2026

docs(rfcs): update RFC0015 ROCm/TheRock#9433

Open

amontoison force-pushed the users/amontoison/hipblas-fortran-bindings branch 2 times, most recently from 23daf28 to 81e7888 Compare October 22, 2026 15:30

amontoison force-pushed the users/amontoison/hipblas-fortran-bindings branch from 81e7888 to 0940a13 Compare October 22, 2026 18:13

github-actions Bot added github actions project: none Does not target any component labels Nov 3, 2026

amontoison changed the title ~~feat(hipblas): replace the hand-written Fortran binding with a generated one~~ feat(hipblas): add the generated Fortran binding in bindings/fortran/hipblas Nov 3, 2026

amontoison mentioned this pull request Nov 3, 2026

test(ci): build the Fortran bindings with ROCm/TheRock#9446 (do not merge) #15008

Draft

amontoison added 5 commits November 3, 2026 16:58

Navigation Menu

Repository navigation

feat(hipblas): add the generated Fortran binding in bindings/fortran/hipblas - #12370

feat(hipblas): add the generated Fortran binding in bindings/fortran/hipblas#12370
amontoison wants to merge 8 commits into
ROCm:developfrom
amontoison:users/amontoison/hipblas-fortran-bindings

amontoison commented Sep 21, 2026 •

edited

Loading

Uh oh!

therock-pr-bot Bot commented Sep 21, 2026 •

edited

Loading

Uh oh!

therock-pr-bot Bot commented Sep 21, 2026 •

edited

Loading

Uh oh!

codecov Bot commented Sep 28, 2026 •

edited

Loading

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

amontoison commented Sep 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What lands

Options

In TheRock

Follow-ups

Build and test

Verification

Uh oh!

therock-pr-bot Bot commented Sep 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ All Checks Passed — Ready for Review

Uh oh!

therock-pr-bot Bot commented Sep 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov Bot commented Sep 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

amontoison commented Sep 21, 2026 •

edited

Loading

therock-pr-bot Bot commented Sep 21, 2026 •

edited

Loading

therock-pr-bot Bot commented Sep 21, 2026 •

edited

Loading

codecov Bot commented Sep 28, 2026 •

edited

Loading