dotnet/runtime

JIT: A case of needing `JitOptRepeat=3` for constant folding and conditional move

Open

#124700 opened on Feb 21, 2026

View on GitHub
 (7 comments) (0 reactions) (0 assignees)C# (17,886 stars) (5,445 forks)batch import
area-CodeGen-coreclrhelp wantedoptimization

Description

Consider this (https://godbolt.org/z/7b4qsdKTa):

static int Unoptimized(int a, int b)
{
    int num = 0;

    if (a == b)
    {
        num += 3;
    }
    if (a == b)
    {
        num++;
    }
    if (a == b)
    {
        num++;
    }

    return num;
}

which compiles into:

G_M13546_IG02:  ;; offset=0x0000
       xor      eax, eax
       cmp      ecx, edx
       jne      SHORT G_M13546_IG04
						;; size=6 bbWeight=1 PerfScore 1.50

G_M13546_IG03:  ;; offset=0x0006
       mov      eax, 3
       inc      eax
       inc      eax
						;; size=9 bbWeight=0.50 PerfScore 0.38

As can be seen the mov + inc + inc chain is not constant folded. Having JitOptRepeat=2 fixes that:

G_M13546_IG02:  ;; offset=0x0000
       xor      eax, eax
       cmp      ecx, edx
       jne      SHORT G_M13546_IG04
						;; size=6 bbWeight=1 PerfScore 1.50

G_M13546_IG03:  ;; offset=0x0006
       mov      eax, 5
						;; size=5 bbWeight=0.50 PerfScore 0.12

However, there is still a jump. To get a conditional move we need JitOptRepeat=3:

G_M13546_IG02:  ;; offset=0x0000
       xor      eax, eax
       mov      r8d, 5
       cmp      ecx, edx
       cmove    eax, r8d
						;; size=14 bbWeight=1 PerfScore 1.00

The codegen is now equivalent to return (a == b) ? 5 : 0;

Contributor guide

Tech stack
csharp
Domain
performance
Issue type
performance
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.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
fresh
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
JIT compiler basicsC#Assembly
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
15
Research direction
The issue describes a case where the JIT fails to constant fold and use conditional move optimization unless JitOptRepeat is increased. The expected behavior is that the JIT should apply these optimizations with default settings. To reproduce, use the provided code snippet on Godbolt. Investigate the JIT's optimization pipeline, specifically the constant folding and conditional move phases, and understand why multiple optimization iterations are required. Look at the JIT source code in the dotnet/runtime repository, particularly the optimization passes (e.g., in src/coreclr/jit/). Identify the condition that prevents early optimization and propose a fix that allows the JIT to fold the constants and use cmov without needing JitOptRepeat=3.