Add scalar support for offload#150288

Sa4dUs · 2025-12-22T20:20:23Z

This PR adds scalar support to the offload feature. The scalar management has two main parts:

On the host side, each scalar arg is casted to ix type, zero extended to i64 and passed to the kernel like that.
On the device, the each scalar arg (i64 at that point), is truncated to ix and then casted to the original type.

r? @ZuseZ4

ZuseZ4 · 2025-12-30T13:46:56Z

@Sa4dUs, what's the current state of this? Given the omp/ol expectation of scalars being 64 bit integers I guess this code only works for i64 and usize and everything else will need the casts (or only works by chance)?
I'm fine with having casts in a follow-up PR, but I guess we should start having more test files in which we check that all the different argument types are being handled correctly, so at least an i64 test for this one?

Sa4dUs · 2025-12-30T19:45:15Z

it works for all 64-bit scalars. for f64 it's not what omp expects but somehow it just works, not sure it it will break in some weird edge case

compiler/rustc_codegen_llvm/src/builder/gpu_offload.rs

rust-bors · 2026-01-02T23:05:14Z

☔ The latest upstream changes made this pull request unmergeable. Please resolve the merge conflicts.

bors · 2026-01-03T00:03:49Z

☔ The latest upstream changes (presumably #150606) made this pull request unmergeable. Please resolve the merge conflicts.

rustbot · 2026-01-14T17:56:28Z

r? @jdonszelmann

rustbot has assigned @jdonszelmann.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

compiler/rustc_codegen_llvm/src/back/write.rs

compiler/rustc_codegen_llvm/src/intrinsic.rs

compiler/rustc_llvm/llvm-wrapper/RustWrapper.cpp

ZuseZ4 · 2026-01-19T20:21:14Z

compiler/rustc_codegen_llvm/src/back/write.rs

+        let mut old_args_rebuilt = Vec::with_capacity(old_param_types.len());
+
+        for (i, &old_ty) in old_param_types.iter().enumerate() {
+            let new_arg = unsafe { llvm::LLVMGetParam(new_fn, (i + 1) as u32) };


I had missed this one. Otherwise lgtm.
Use compiler/rustc_codegen_llvm/src/llvm/mod.rs

287 /// Safe wrapper around `LLVMGetParam`, because segfaults are no fun. 1 pub(crate) fn get_param(llfn: &Value, index: c_uint) -> &Value {

ZuseZ4 · 2026-01-19T20:29:57Z

tests/codegen-llvm/gpu_offload/scalar_host.rs

+// CHECK: store double 4.200000e+01, ptr %0, align 8
+// CHECK: %_0.i = load double, ptr %0, align 8
+// CHECK: store double %_0.i, ptr %addr, align 8
+// CHECK-NEXT: call void @__tgt_register_lib(ptr nonnull %EmptyDesc)


While at it, can you please remove the check for : call void @__tgt_register_lib(ptr nonnull %EmptyDesc)? My other open PR moves it into globals, so that would break this test.

ZuseZ4 · 2026-01-19T20:53:02Z

compiler/rustc_codegen_llvm/src/builder/gpu_offload.rs

+        let ty_kind = cx.type_kind(ty);
+        let (base_val, gep_base) = match ty_kind {
+            TypeKind::Pointer => (v, v),
+            TypeKind::Half | TypeKind::Float | TypeKind::Double | TypeKind::Integer => {


Add a FIXME that we should later check for f128 support. At least newer NVIDIA cards should support it.

ZuseZ4 · 2026-01-19T21:08:33Z

tests/codegen-llvm/gpu_offload/scalar_host.rs

+
+// CHECK: define{{( dso_local)?}} void @main()
+// CHECK-NOT: define
+// CHECK: %EmptyDesc = alloca %struct.__tgt_bin_desc, align 8


Oh and also remove this

ZuseZ4 · 2026-01-19T21:44:01Z

@bors delegate
r=me when CI passes.

rust-bors · 2026-01-19T21:44:04Z

✌️ @Sa4dUs, you can now approve this pull request!

If @ZuseZ4 told you to "r=me" after making some further change, then please make that change and post @bors r=ZuseZ4.

Sa4dUs · 2026-01-19T23:08:40Z

@bors r=ZuseZ4

rust-bors · 2026-01-19T23:08:43Z

📌 Commit 307a4fc has been approved by ZuseZ4

It is now in the queue for this repository.

@ZuseZ4

Add scalar support for offload This PR adds scalar support to the offload feature. The scalar management has two main parts: On the host side, each scalar arg is casted to `ix` type, zero extended to `i64` and passed to the kernel like that. On the device, the each scalar arg (`i64` at that point), is truncated to `ix` and then casted to the original type. r? @ZuseZ4

@ghost

Rollup of 8 pull requests Successful merges: - #149587 (coverage: Sort the expansion tree to help choose a single BCB for child expansions) - #150071 (Add dist step for Enzyme) - #150288 (Add scalar support for offload) - #151091 (Add new "hide deprecated items" setting in rustdoc) - #151255 (rustdoc: Fix ICE when deprecated note is not resolved on the correct `DefId`) - #151375 (Fix terminal width dependent tests) - #151384 (add basic `TokenStream` api tests) - #151391 (rustc-dev-guide subtree update) r? @ghost

@ZuseZ4

Rollup merge of #150288 - offload-bench-fix, r=ZuseZ4 Add scalar support for offload This PR adds scalar support to the offload feature. The scalar management has two main parts: On the host side, each scalar arg is casted to `ix` type, zero extended to `i64` and passed to the kernel like that. On the device, the each scalar arg (`i64` at that point), is truncated to `ix` and then casted to the original type. r? @ZuseZ4

@ghost

Rollup of 8 pull requests Successful merges: - rust-lang/rust#149587 (coverage: Sort the expansion tree to help choose a single BCB for child expansions) - rust-lang/rust#150071 (Add dist step for Enzyme) - rust-lang/rust#150288 (Add scalar support for offload) - rust-lang/rust#151091 (Add new "hide deprecated items" setting in rustdoc) - rust-lang/rust#151255 (rustdoc: Fix ICE when deprecated note is not resolved on the correct `DefId`) - rust-lang/rust#151375 (Fix terminal width dependent tests) - rust-lang/rust#151384 (add basic `TokenStream` api tests) - rust-lang/rust#151391 (rustc-dev-guide subtree update) r? @ghost

Sa4dUs force-pushed the offload-bench-fix branch from d99cf66 to 0330a0e Compare December 26, 2025 16:49

ZuseZ4 reviewed Dec 31, 2025

View reviewed changes

compiler/rustc_codegen_llvm/src/builder/gpu_offload.rs Outdated Show resolved Hide resolved

compiler/rustc_codegen_llvm/src/builder/gpu_offload.rs Show resolved Hide resolved

Sa4dUs force-pushed the offload-bench-fix branch from 0330a0e to ea71b7b Compare January 3, 2026 17:03

ZuseZ4 mentioned this pull request Jan 6, 2026

Tracking Issue for GPU-offload #131513

Open

5 tasks

Sa4dUs force-pushed the offload-bench-fix branch from ea71b7b to 8b7e6b6 Compare January 7, 2026 13:28

Sa4dUs added 2 commits January 14, 2026 11:37

Minor doc and ty fixes

bc751ad

Allow bounded types

2c9c5d1

Sa4dUs force-pushed the offload-bench-fix branch 2 times, most recently from 5189bbb to 45ccec5 Compare January 14, 2026 15:30

Sa4dUs changed the title ~~WIP Offload Benchmarks Fixes~~ Add scalar support for offload Jan 14, 2026

Sa4dUs force-pushed the offload-bench-fix branch from 45ccec5 to 38f7fc7 Compare January 14, 2026 15:38

Sa4dUs marked this pull request as ready for review January 14, 2026 17:56

rustbot added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jan 14, 2026

rustbot assigned jdonszelmann Jan 14, 2026

rustbot removed the S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. label Jan 14, 2026

rustbot assigned ZuseZ4 and unassigned jdonszelmann Jan 14, 2026

ZuseZ4 reviewed Jan 14, 2026

View reviewed changes

Sa4dUs force-pushed the offload-bench-fix branch from 38f7fc7 to 5e980e1 Compare January 14, 2026 19:42

This comment has been minimized.

Sign in to view

Sa4dUs force-pushed the offload-bench-fix branch from 5e980e1 to 272a1a6 Compare January 15, 2026 00:06

ZuseZ4 mentioned this pull request Jan 16, 2026

Finish the std::offload module rust-lang/rust-project-goals#109

Open

8 tasks

ZuseZ4 reviewed Jan 19, 2026

View reviewed changes

Add scalar support for both host and device

307a4fc

Sa4dUs force-pushed the offload-bench-fix branch from 272a1a6 to 307a4fc Compare January 19, 2026 21:31

rust-bors bot added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Jan 19, 2026

This was referenced Jan 20, 2026

Rollup of 9 pull requests #151392

Closed

Rollup of 8 pull requests #151395

Merged

rust-bors bot merged commit 1262ff9 into rust-lang:main Jan 20, 2026
11 checks passed

rustbot added this to the 1.95.0 milestone Jan 20, 2026

Uh oh!

Conversation

Sa4dUs commented Dec 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ZuseZ4 commented Dec 30, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Sa4dUs commented Dec 30, 2025

Uh oh!

Uh oh!

Uh oh!

rust-bors bot commented Jan 2, 2026

Uh oh!

bors commented Jan 3, 2026

Uh oh!

rustbot commented Jan 14, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

This comment has been minimized.

ZuseZ4 Jan 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ZuseZ4 Jan 19, 2026

Choose a reason for hiding this comment

Uh oh!

ZuseZ4 Jan 19, 2026

Choose a reason for hiding this comment

Uh oh!

ZuseZ4 Jan 19, 2026

Choose a reason for hiding this comment

Uh oh!

ZuseZ4 commented Jan 19, 2026

Uh oh!

rust-bors bot commented Jan 19, 2026

Uh oh!

Sa4dUs commented Jan 19, 2026

Uh oh!

rust-bors bot commented Jan 19, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Sa4dUs commented Dec 22, 2025 •

edited

Loading

ZuseZ4 commented Dec 30, 2025 •

edited

Loading

ZuseZ4 Jan 19, 2026 •

edited

Loading