♻️ Define target compilation with payload specifications by burgholzer · Pull Request #2219 · munich-quantum-toolkit/core

burgholzer · 2026-08-23T19:09:34Z

🤖 AI text below 🤖

Lead: @simon1hofmann. Cross-repository coordination: @burgholzer.

Description

Keep the compiler-only selected-program capability prototype independent of QDMI runtime adoption. Typed metadata and a cached target-environment analysis connect target selection to compilation without requiring unreleased QDMI APIs.

Compiler workstream: main → #2219 → #2162. QDMI adaptation remains the separate integration PR #2227. This compiler-only work targets Core 4.0, subject to Simon's compiler contract review and current CI. It does not require QDMI 1.4 or approval of the later QDMI metadata contract. #2131 tracks compiler requirements; #2365 coordinates the separate Core 4.1/QDMI 1.4 adaptation.

Retain the compiler behavior already released onto main through #2218/#2323/#2215: unknown topology or gate sets fail during inference, variadic controlled DDSIM gates and zero-arity global phase remain supported, and placement/decomposition retain their validated target. No driver replacement, batching or metadata cleanup is included.

Local validation

Independent release build and CTest: 3,879 passed, one existing optional-device skip.
Targeted Python compiler/QDMI tests: 558 passed.
Stub generation, repository lint and full changed-file C++ lint passed.
Hosted CI must be checked separately on this final head.

AI-assisted restructuring; human design review remains required.

Compiler contract review

Model the capabilities of the selected program/target combination rather than assuming a format's full language is executable. Keep the initial feature vocabulary bounded and informed by OpenQASM and QIR. Validate representative benchmark targets without requiring a QDMI 1.4 device. The later QDMI adapter may expose only a subset until its independent contract is ready.

Checklist

The pull request only contains commits that are focused and relevant to this change.
I have added appropriate tests that cover the new/changed functionality.
I have updated the documentation to reflect these changes.
I have added entries to the changelog for any noteworthy additions, changes, fixes, or removals.
I have added migration instructions to the upgrade guide (if needed).
The changes follow the project's style guidelines and introduce no new warnings.
The changes are fully tested and pass the CI checks.
I have reviewed my own code changes.

If PR contains AI-assisted content:

Any agent that created, edited, or submitted GitHub content was explicitly authorized for that scope, as required by our AI Usage Guidelines.
Every agent-authored or agent-edited public text body begins with the visible disclosure 🤖 *AI text below* 🤖 (titles are exempt).
I have disclosed AI assistance in the PR description.
I confirm that I have personally reviewed and understood all AI-generated content, and accept full responsibility for it.

codecov · 2026-08-23T19:16:24Z

Codecov Report

❌ Patch coverage is 93.11295% with 25 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
mlir/tools/mqt-cc/mqt-cc.cpp	76.7%	13 Missing ⚠️
mlir/lib/Compiler/TargetEnvironment.cpp	93.4%	12 Missing ⚠️

📢 Thoughts on this report? Let us know!

burgholzer

This one will likely still need a bit of work to iron out the details.

coderabbitai · 2026-08-24T13:33:25Z

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

burgholzer · 2026-08-24T18:28:28Z

This looks much cleaner now. And should also be ready.

burgholzer · 2026-08-25T09:46:24Z

@simon1hofmann @denialhaag any feedback on this one from your side?
I looked through this yesterday night already and it looked pretty clean to me.

simon1hofmann · 2026-08-25T10:21:30Z

@simon1hofmann @denialhaag any feedback on this one from your side? I looked through this yesterday night already and it looked pretty clean to me.

Going through this now and will provide feedback in around 15min.

simon1hofmann

Also went through the changes here and they look really clean 👍

denialhaag

Thanks for the continued improvements, @simon1hofmann! 🙂 Overall, this looks pretty good to me. I only have a few remarks:

Record complete compiler-target facts and one exact payload execution contract as typed module metadata. Assisted-by: GPT-5.6 Sol via Codex

Represent the selected hardware target and payload specification as one validated target environment. Cache that value through the MLIR analysis manager for mapping, native synthesis, and conformance. Derive targeted compiler output from the selected payload format in the C++, Python, and mqt-cc APIs. Keep untargeted output selection independent. Assisted-by: GPT-5.6 Sol via Codex

Declare the dialect used by runtime-angle synthesis and cover a fresh context that does not preload it. Assisted-by: GPT-6 via Codex

Zero-fill omitted numeric components at the payload snapshot boundary. Accept the same syntax in typed MLIR and keep compiler output selection exact. Assisted-by: GPT-6 via Codex

Use DDSIM for executable QIR examples and state the environment required by the C++ pipeline builder. Assisted-by: GPT-6 via Codex

Co-authored-by: Daniel Haag <121057143+denialhaag@users.noreply.github.com> Signed-off-by: simon1hofmann <119581649+simon1hofmann@users.noreply.github.com>

Assisted-by: OpenAI via Codex

Preserve upstream regressions after rebasing onto main by using the existing target-environment helpers. Assisted-by: OpenAI via Codex

denialhaag

Thanks for addressing all of my comments, @simon1hofmann! I'll leave the final comment open for now, but I'll not die on that hill; you can resolve it whenever we have come to any conclusion. I'm approving for now. Feel free to request another review if there are new changes I can have a look at! 🙂

simon1hofmann · 2026-09-08T14:24:50Z

Also looks good to me, will leave the final check to @burgholzer before merging.

burgholzer · 2026-09-08T14:31:16Z

I'll tackle this one next

Make the pipeline take one environment and reuse its prepared target in the cached analysis. Standalone passes decode the typed pair on demand. Remove the unused DLTI extension and query layer. Assisted-by: GPT-6 via Codex

burgholzer

I ran this through another iteration with the goal of simplifying the implementation, which was pretty successful I believe.
I'd vote for getting this in now and looking at the fallout in subsequent audits until the release.

Move the #2162 reference beside #2219 and describe payload-aware control-flow legalization in the existing target-compilation entry. Assisted-by: OpenAI Codex

* ✨ Legalize control flow for selected payloads Use stock MLIR normalization, bounded static loop unrolling, and dialect conversion to match residual structured control flow to the selected payload specification. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-5.6 Sol via Codex * ♻️ Remove redundant modifier legality exemptions QCO modifiers do not implement branch interfaces, so the existing fallback already accepts them. Assisted-by: GPT-6 via Codex * 📝 Align control flow release scope Preserve the capability snapshot while distinguishing Core 4.0 compiler review from the separate QDMI 1.4 adaptation. Assisted-by: GPT-6 via Codex * ♻️ Simplify structured payload legalization Use native static trip counts while retaining literal-bound proofs and full unroll safety limits. Accept single-case switches with multiway support. Require structured input at pipeline entry and remove CFG lifting and the redundant cleanup after control legalization. Assisted-by: GPT-6 via Codex * ♻️ Validate loop inputs before legalization Check the supported SCF loop input form at both pass entry points, so loop legalization can rely on explicit iteration arguments. Keep branch checks after unrolling and leave the QCO linearity contract unchanged. Assisted-by: OpenAI Codex * ♻️ Clarify control-flow pass names Keep payload-loop unrolling separate from residual control-flow checks so constant propagation and QCO cleanup can run between them. Leave the factor-based quantum-loop optimization unchanged. Assisted-by: OpenAI Codex * ♻️ Centralize control-flow capability IDs Keep the compiler snapshot identifiers beside ProgramCapability and ProgramConstraint so producers and legalization share their spellings. Do not depend on the draft QDMI capability API. Assisted-by: OpenAI Codex * 🧪 Keep control fixtures allocation-independent Exercise control-flow rejection with valid quantum inputs and remove unrelated quantum operations from classical fixtures. Assisted-by: OpenAI Codex * 🐛 Check the width of fully unrolled loop steps Reject signed step truncation before native unrolling mutates the loop. Preserve safe signed and unsigned narrow-integer loops, and check their observed induction values. Assisted-by: OpenAI Codex * 🐛 Bound switch expansion before rewriting Check payload depth and a compiler nesting ceiling before moving case regions. Build accepted chains iteratively and retain native multiway switches. Cover expansion boundaries, moved control, state selection, and cumulative loop cloning. Assisted-by: OpenAI Codex * ♻️ Reuse native trivial-loop promotion Reuse the native zero/one-trip utility, apply required C++ initializer style, and replace stale local handoff notes with scoped validation. Keep the switch safety comment purely technical. Assisted-by: OpenAI Codex * 🧪 Exercise payload limits through frontends Replace three handwritten positive fixtures with OpenQASM imports and test a live Qiskit classical switch against selected capabilities. Keep the focused raw-MLIR safety regressions and production contract unchanged. Assisted-by: OpenAI Codex * 📝 Group payload legalization with target compilation Move the #2162 reference beside #2219 and describe payload-aware control-flow legalization in the existing target-compilation entry. Assisted-by: OpenAI Codex * 🐛 Legalize empty and unsigned counted loops Reuse LLVM's state and induction-value remapping for terminator-only loops within the existing clone budget. Zero-extend unsigned bounds before checking whether full unrolling is safe. Add frontend state-permutation and direct induction-value regressions, and cover unsigned bounds above the signed range of their type. Assisted-by: GPT-6 via Codex --------- Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

burgholzer added feature New feature or request c++ Anything related to C++ code python Anything related to Python code MLIR Anything related to MLIR QDMI Anything related to QDMI labels Aug 23, 2026

burgholzer self-assigned this Aug 23, 2026

burgholzer force-pushed the codex/selected-payload-environment branch from b53e303 to c62e088 Compare August 23, 2026 19:10

burgholzer force-pushed the codex/selected-payload-environment branch from c62e088 to ae5b72f Compare August 23, 2026 20:38

This was referenced Aug 23, 2026

🐛 Reject nonlinear QCO programs at checked boundaries #2220

Merged

✨ Legalize control flow for selected payloads #2162

Merged

burgholzer added this to the MLIR Support milestone Aug 24, 2026

burgholzer commented Aug 24, 2026

View reviewed changes

mergify Bot added the conflict label Aug 24, 2026

burgholzer force-pushed the codex/selected-payload-environment branch from ae5b72f to b0dd192 Compare August 24, 2026 13:31

burgholzer changed the title ~~✨ Select exact payload environments for target compilation~~ ♻️ Define target compilation with payload specifications Aug 24, 2026

burgholzer removed QDMI Anything related to QDMI conflict labels Aug 24, 2026

burgholzer force-pushed the codex/selected-payload-environment branch from b0dd192 to bb0240b Compare August 24, 2026 14:15

This was referenced Aug 24, 2026

✨ Adopt exact QDMI 1.4 payload contracts across Core #2226

Draft

✨ Build target environments from QDMI payloads #2227

Draft

burgholzer force-pushed the codex/selected-payload-environment branch from bb0240b to 28284fa Compare August 24, 2026 23:47

simon1hofmann approved these changes Aug 25, 2026

View reviewed changes

mergify Bot removed the conflict label Sep 7, 2026

simon1hofmann requested a review from denialhaag September 7, 2026 11:28

simon1hofmann marked this pull request as ready for review September 7, 2026 13:13

denialhaag requested changes Sep 7, 2026

View reviewed changes

simon1hofmann mentioned this pull request Sep 7, 2026

♻️ Split the MQT dialect implementation #2445

Open

mergify Bot added the conflict label Sep 7, 2026

simon1hofmann force-pushed the codex/selected-payload-environment branch from 2a25b61 to 245a55f Compare September 8, 2026 07:34

mergify Bot added conflict and removed conflict labels Sep 8, 2026

burgholzer and others added 10 commits September 8, 2026 15:25

✨ Record typed compiler target environments in MQT IR

5b4c1a4

Record complete compiler-target facts and one exact payload execution contract as typed module metadata. Assisted-by: GPT-5.6 Sol via Codex

📝 Align payload model release target

8ccedc1

🐛 Load math for target native synthesis

20984b3

Declare the dialect used by runtime-angle synthesis and cover a fresh context that does not preload it. Assisted-by: GPT-6 via Codex

✨ Accept exact payload version shorthand

ff704c1

Zero-fill omitted numeric components at the payload snapshot boundary. Accept the same syntax in typed MLIR and keep compiler output selection exact. Assisted-by: GPT-6 via Codex

📝 Clarify target and payload prerequisites

321d8dd

Use DDSIM for executable QIR examples and state the environment required by the C++ pipeline builder. Assisted-by: GPT-6 via Codex

Apply batched suggestions from code review

4c05ee9

Co-authored-by: Daniel Haag <121057143+denialhaag@users.noreply.github.com> Signed-off-by: simon1hofmann <119581649+simon1hofmann@users.noreply.github.com>

🎨 Restore QCO namespace-closing comment

7ba9534

Assisted-by: OpenAI via Codex

📝 Add payload and target environment examples

95e5fec

Assisted-by: OpenAI via Codex

🧪 Adapt new target tests to payload environments

8a050a0

Preserve upstream regressions after rebasing onto main by using the existing target-environment helpers. Assisted-by: OpenAI via Codex

simon1hofmann force-pushed the codex/selected-payload-environment branch from 245a55f to 8a050a0 Compare September 8, 2026 13:34

mergify Bot removed the conflict label Sep 8, 2026

denialhaag approved these changes Sep 8, 2026

View reviewed changes

♻️ Share the selected target environment

0555f26

Make the pipeline take one environment and reuse its prepared target in the cached analysis. Standalone passes decode the typed pair on demand. Remove the unused DLTI extension and query layer. Assisted-by: GPT-6 via Codex

burgholzer commented Sep 8, 2026

View reviewed changes

Comment thread mlir/include/mlir/Compiler/TargetEnvironment.h

simon1hofmann merged commit 784519c into main Sep 8, 2026
26 checks passed

simon1hofmann deleted the codex/selected-payload-environment branch September 8, 2026 19:13

Uh oh!

Uh oh!

Conversation

burgholzer commented Aug 23, 2026 • edited by simon1hofmann Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Local validation

Compiler contract review

Checklist

Uh oh!

codecov Bot commented Aug 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

burgholzer left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

coderabbitai Bot commented Aug 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

burgholzer commented Aug 24, 2026

Uh oh!

burgholzer commented Aug 25, 2026

Uh oh!

simon1hofmann commented Aug 25, 2026

Uh oh!

simon1hofmann left a comment

Choose a reason for hiding this comment

Uh oh!

denialhaag left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

denialhaag left a comment

Choose a reason for hiding this comment

Uh oh!

simon1hofmann commented Sep 8, 2026

Uh oh!

burgholzer commented Sep 8, 2026

Uh oh!

burgholzer left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

burgholzer commented Aug 23, 2026 •

edited by simon1hofmann

Loading

codecov Bot commented Aug 23, 2026 •

edited

Loading

coderabbitai Bot commented Aug 24, 2026 •

edited

Loading