Add mul_add_relaxed methods for floating-point types by landsharkiest · Pull Request #151793 · rust-lang/rust

landsharkiest · 2026-01-28T20:03:28Z

Implements mul_add_relaxed for f16, f32, f64, and f128, which computes (self * a) + b with relaxed precision semantics. Unlike mul_add which guarantees a fused operation, this variant allows the compiler to choose between fused or separate operations based on target performance.

This fills the gap between the precision-guaranteed mul_add and the fully-optimizable algebraic operators, providing target-specific optimization while maintaining reasonable floating-point semantics.

Tracking issue: #151770

try-job: dist-i586-gnu-i586-i686-musl

rustbot · 2026-01-28T20:03:33Z

r? @Mark-Simulacrum

rustbot has assigned @Mark-Simulacrum.
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

RalfJung · 2026-07-16T18:39:19Z

@landsharkiest friendly reminder that this PR is waiting for you to react to the reviewer's feedback and to fix CI. :)
if you no longer have time to work on this, that's fine -- just let us know so the task can be picked up by someone else.

folkertdev · 2026-07-16T18:50:48Z

Another thing, please update the doc comment of the intrinsics to point to the public methods, like here

/// The stabilized version of this intrinsic is 
/// [`f128::mul_add`](../../std/primitive.f128.html#method.mul_add)

rust/library/core/src/intrinsics/mod.rs

Lines 1459 to 1480 in a6050b7

    
           /// Returns `a * b + c` for `f128` values. 
        
           /// 
        
           /// The stabilized version of this intrinsic is 
        
           /// [`f128::mul_add`](../../std/primitive.f128.html#method.mul_add) 
        
           #[rustc_intrinsic_const_stable_indirect] 
        
           #[rustc_intrinsic] 
        
           #[rustc_nounwind] 
        
           pub const fn fmaf128(a: f128, b: f128, c: f128) -> f128; 
        
           /// Returns `a * b + c` for `f16` values, non-deterministically executing 
        
           /// either a fused multiply-add or two operations with rounding of the 
        
           /// intermediate result. 
        
           /// 
        
           /// The operation is fused if the code generator determines that target 
        
           /// instruction set has support for a fused operation, and that the fused 
        
           /// operation is more efficient than the equivalent, separate pair of mul 
        
           /// and add instructions. It is unspecified whether or not a fused operation 
        
           /// is selected, and that may depend on optimization level and context, for 
        
           /// example. 
        
           #[rustc_intrinsic] 
        
           #[rustc_nounwind] 
        
           pub const fn fmuladdf16(a: f16, b: f16, c: f16) -> f16;

We use the prasing of "stabilized" here even if the methods are not actually stable.

rustbot · 2026-07-18T01:25:49Z

Some changes occurred to the intrinsics. Make sure the CTFE / Miri interpreter
gets adapted for the changes, if necessary.

cc @rust-lang/miri, @RalfJung, @oli-obk, @lcnr

landsharkiest · 2026-07-18T01:27:17Z

@landsharkiest friendly reminder that this PR is waiting for you to react to the reviewer's feedback and to fix CI. :) if you no longer have time to work on this, that's fine -- just let us know so the task can be picked up by someone else.

Sorry got caught up in school and job searching, will finally put this away! :)

scottmcm · 2026-07-26T00:16:10Z

i586 is known sketchy for floats always, right? Could potentially just skip them for that...

folkertdev · 2026-07-26T00:27:32Z

Yeah some tests were already skipped since the last CI failure, seems like a bit more ignoring is needed.

https://triagebot.infra.rust-lang.org/gh-changes-since/rust-lang/rust/151793/234c31cd674e11703f15d290cba7ff81dfe8b4b8..3b7025f3a35e74943089a7e2dfe139bf2f821a89

RalfJung · 2026-07-26T07:30:58Z

Yeah if you check precise results then it's not surprising to see failures on that target.

tgross35 · 2026-08-04T08:00:42Z

+    /// # #[cfg(target_has_reliable_f16)] {
+    ///
+    /// let result = 1.0f16.mul_add_relaxed(2.0, 3.0);
+    /// assert_eq!(result, 5.0);


Could you add the (0.1, 0.1, -0.01) example for f16 and f128 too? Answers should be either -7.6e-6 or -7.03e-6 for f16, 1.5046327690525280101999827676444745E-36 or 7.824090399073145653039910391751267E-37 for f128.

View changes since the review

On 32-bit x86 without SSE2, x87 rounds the f64 product to a 64-bit significand rather than keeping it exact or rounding it to 53 bits, so the doc example can produce a value (9.0209e-19) that is neither the fused nor the unfused result. Skip that check there, like other float doctests do for i586. The f32 example is unaffected: its intermediate values fit in the x87 significand, so it always yields one of the two allowed results. Also add the fused-or-unfused example to f16 and f128, as requested in review.

rustbot · 2026-09-26T16:11:16Z

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

RalfJung · 2026-09-26T16:25:57Z

r? @folkertdev

@landsharkiest remember to post @rustbot ready when this PR is ready for review again.

RalfJung · 2026-09-26T16:33:00Z

@bors try

…r=<try> Add mul_add_relaxed methods for floating-point types

Say that even two invocations with the same inputs may produce different results, and remove the `1.0.mul_add_relaxed(2.0, 3.0)` example: it is atypical, since the fused and unfused results are the same there. The fused-or-unfused example remains.

RalfJung · 2026-09-26T16:37:26Z

@bors try

…r=<try> Add mul_add_relaxed methods for floating-point types

RalfJung · 2026-09-26T16:45:38Z

It seems to ignore try-jobs in the PR description? Strange.
EDIT: Ah, it has to be "try-job". Without the s.

@bors try jobs=dist-i586-gnu-i586-i686-musl

…r=<try> Add mul_add_relaxed methods for floating-point types try-job: dist-i586-gnu-i586-i686-musl

rust-bors · 2026-09-26T17:15:49Z

☀️ Try build successful (CI)
Build commit: d7f43bc (d7f43bc5cd6a4b69dc4f074274243879f840c658)
Base parent: a22b02e (a22b02eaecd6ac937d752139c79d0159c725932d)

landsharkiest · 2026-09-26T23:31:33Z

@rustbot ready

tgross35

LGTM, but please squash

View changes since this review

rustbot assigned Mark-Simulacrum Jan 28, 2026

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Jan 28, 2026

This comment has been minimized.

Sign in to view

tgross35 reviewed Jan 29, 2026

View reviewed changes

Comment thread tests/ui/intrinsics/float-mul-add-relaxed.rs

This comment has been minimized.

Sign in to view

scottmcm mentioned this pull request Jan 29, 2026

Tracking Issue for float_mul_add_relaxed #151770

Open

5 tasks

scottmcm reviewed Jan 29, 2026

View reviewed changes

Comment thread library/core/src/num/f32.rs Outdated

This comment has been minimized.

Sign in to view

Mark-Simulacrum added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Feb 1, 2026

This comment has been minimized.

Sign in to view

folkertdev mentioned this pull request Jul 16, 2026

add a fallback for fmuladdf* #159386

Merged

landsharkiest force-pushed the float_mul_add_relaxed-#151770 branch from 0ca5122 to 174bea7 Compare July 18, 2026 01:25

This comment has been minimized.

Sign in to view

landsharkiest force-pushed the float_mul_add_relaxed-#151770 branch from 174bea7 to 67e0947 Compare July 18, 2026 02:44

This comment has been minimized.

Sign in to view

tgross35 reviewed Aug 4, 2026

View reviewed changes

This comment has been minimized.

Sign in to view

landsharkiest added 2 commits September 26, 2026 11:46

Add mul_add_relaxed methods for floating-point types

ee249af

landsharkiest force-pushed the float_mul_add_relaxed-#151770 branch from ace8a46 to 335e78f Compare September 26, 2026 16:11

rustbot assigned folkertdev and unassigned Mark-Simulacrum Sep 26, 2026

RalfJung reviewed Sep 26, 2026

View reviewed changes

Comment thread library/core/src/num/f128.rs Outdated

RalfJung reviewed Sep 26, 2026

View reviewed changes

Comment thread library/core/src/num/f128.rs Outdated

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Sep 26, 2026

Auto merge of #151793 - landsharkiest:float_mul_add_relaxed-#151770, …

756ff24

…r=<try> Add mul_add_relaxed methods for floating-point types

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Sep 26, 2026

Auto merge of #151793 - landsharkiest:float_mul_add_relaxed-#151770, …

dcbc407

…r=<try> Add mul_add_relaxed methods for floating-point types

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Sep 26, 2026

Auto merge of #151793 - landsharkiest:float_mul_add_relaxed-#151770, …

d7f43bc

…r=<try> Add mul_add_relaxed methods for floating-point types try-job: dist-i586-gnu-i586-i686-musl

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Sep 26, 2026

tgross35 approved these changes Sep 28, 2026

View reviewed changes

Uh oh!

Conversation

landsharkiest commented Jan 28, 2026 • edited by RalfJung Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Jan 28, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

Uh oh!

This comment has been minimized.

This comment has been minimized.

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

RalfJung commented Jul 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

folkertdev commented Jul 16, 2026

Uh oh!

rustbot commented Jul 18, 2026

Uh oh!

This comment has been minimized.

landsharkiest commented Jul 18, 2026

Uh oh!

This comment has been minimized.

scottmcm commented Jul 26, 2026

Uh oh!

folkertdev commented Jul 26, 2026

Uh oh!

This comment has been minimized.

RalfJung commented Jul 26, 2026

Uh oh!

tgross35 Aug 4, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

This comment has been minimized.

rustbot commented Sep 26, 2026

Uh oh!

RalfJung commented Sep 26, 2026

Uh oh!

Uh oh!

Uh oh!

RalfJung commented Sep 26, 2026

Uh oh!

This comment has been minimized.

RalfJung commented Sep 26, 2026

Uh oh!

This comment has been minimized.

RalfJung commented Sep 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment has been minimized.

rust-bors Bot commented Sep 26, 2026

Uh oh!

landsharkiest commented Sep 26, 2026

Uh oh!

tgross35 left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

This branch has not been deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

9 participants

landsharkiest commented Jan 28, 2026 •

edited by RalfJung

Loading

RalfJung commented Jul 16, 2026 •

edited

Loading

tgross35 Aug 4, 2026 •

edited by rustbot

Loading

RalfJung commented Sep 26, 2026 •

edited

Loading

tgross35 left a comment •

edited by rustbot

Loading