llvm/llvm-project

getNegatedExpression should handle FP16_TO_FP and FP_TO_FP16

Open

#177448 opened on Jan 22, 2026

View on GitHub
 (11 comments) (0 reactions) (0 assignees)C++ (26,378 stars) (10,782 forks)batch import
floating-pointgood first issuellvm:SelectionDAGmissed-optimization

Description

getNegatedExpression does not handle the opcodes FP16_TO_FP and FP_TO_FP16, but it could. This could improve use of source modifiers on AMDGPU with illegal f16.

I started implementing it in this patch, but it requires additional work to avoid regressions.

0001-DAG-Implement-getNegatedExpression-for-FP16_TO_FP.patch

Contributor guide

Tech stack
cpp
Domain
backend
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
3-5 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
C++ programmingLLVM compiler infrastructureSelectionDAGAMDGPU architecture
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
25
Research direction
Review the getNegatedExpression function in llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp (line 7556) to understand the structure for handling opcodes. Analyze the attached patch (0001 DAG Implement getNegatedExpression for FP16 TO FP.patch) to see what has been implemented and what remains to avoid regressions. Investigate how FP16 TO FP and FP TO FP16 are used in the AMDGPU target to ensure the negation is correct. Consider writing tests to validate the changes.