[llvm] [KeyInstr] MDNodeKeyImpl<DILocation> skip zero values for hash (PR #143357)

[Orlando Cazalet-Hyams via llvm-commits]

https://github.com/OCHyams created https://github.com/llvm/llvm-project/pull/143357

Hashing AtomGroup and AtomRank substantially impacts performance whether Key Instructions is enabled or not. We can't detect whether it's enabled here cheaply; avoiding hashing zero values is a good approximation. This affects Key Instruction builds too, but any potential costs incurred by messing with the hash distribution (hash_combine(x) != hash_combine(x, 0)) appear to still be massively outweighed by the overall compile time savings by performing this check.

---

>From compile-time-tracker:

1. this patch
2. set LLVM_EXPERIMENTAL_KEY_INSTRUCTIONS=ON by default (enabling support in LLVM, not enabling the feature)
3. base

```   
    Commit  stage1-O3  stage1-ReleaseThinLTO  stage1-ReleaseLTO-g  stage1-O0-g  stage1-aarch64-O3  stage1-aarch64-O0-g  stage2-O3  stage2-O0-g  stage2-clang
1.  3fa3a147a0  61213M (+0.01%)  77400M (+0.01%)  89460M (-0.16%)  18896M (-0.42%)  68673M (+0.00%)  23128M (-0.31%)  53427M (+0.05%)  16547M (-1.53%)  34059533M (+0.00%)
2.  882faf0ac5  61209M (+0.01%)  77393M (+0.00%)  89608M (+0.22%)  18975M (+0.53%)  68672M (+0.02%)  23201M (+0.42%)  53398M (-0.05%)  16805M (+1.75%)  34059472M (+0.01%)
3.  54d544b831  61205M           77393M           89410M           18874M           68660M           23104M           53426M           16516M           34055496M
```

Compare 2-1: https://llvm-compile-time-tracker.com/compare.php?from=882faf0ac573476edf0f4026a69f6f86f316c821&to=3fa3a147a0d7a579f92f5ca6db1bfcdd485f8ffa&stat=instructions%3Au
Compare 3-2: https://llvm-compile-time-tracker.com/compare.php?from=54d544b83141dc0b20727673f68793728ed54793&to=882faf0ac573476edf0f4026a69f6f86f316c821&stat=instructions%3Au


>From 85bea6e0f7971b942878a5dec38288a85b26e83c Mon Sep 17 00:00:00 2001
From: Orlando Cazalet-Hyams <orlando.hyams at sony.com>
Date: Mon, 9 Jun 2025 09:24:07 +0100
Subject: [PATCH] [KeyInstr] MDNodeKeyImpl<DILocation> skip zero values for
 hash

Hashing AtomGroup and AtomRank substantially impacts performance whether Key
Instructions is enabled or not. We can't detect whether it's enabled here
cheaply; avoiding hashing zero values is a good approximation. This affects
Key Instruction builds too, but any potential costs incurred by messing with
the hash distribution (hash_combine(x) != hash_combine(x, 0)) appear to still
be massively outweighed by the overall compile time savings by performing this
check.
---
 llvm/lib/IR/LLVMContextImpl.h | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/llvm/lib/IR/LLVMContextImpl.h b/llvm/lib/IR/LLVMContextImpl.h
index 21f5c06ea24f3..7b6083a7a3496 100644
--- a/llvm/lib/IR/LLVMContextImpl.h
+++ b/llvm/lib/IR/LLVMContextImpl.h
@@ -355,13 +355,19 @@ template <> struct MDNodeKeyImpl<DILocation> {
   }
 
   unsigned getHashValue() const {
-    return hash_combine(Line, Column, Scope, InlinedAt, ImplicitCode
 #ifdef EXPERIMENTAL_KEY_INSTRUCTIONS
-                        ,
-                        AtomGroup, (uint8_t)AtomRank);
-#else
-    );
+    // Hashing AtomGroup and AtomRank substantially impacts performance whether
+    // Key Instructions is enabled or not. We can't detect whether it's enabled
+    // here cheaply; avoiding hashing zero values is a good approximation. This
+    // affects Key Instruction builds too, but any potential costs incurred by
+    // messing with the hash distribution* appear to still be massively
+    // outweighed by the overall compile time savings by performing this check.
+    // * (hash_combine(x) != hash_combine(x, 0))
+    if (AtomGroup || AtomRank)
+      return hash_combine(Line, Column, Scope, InlinedAt, ImplicitCode,
+                          AtomGroup, (uint8_t)AtomRank);
 #endif
+    return hash_combine(Line, Column, Scope, InlinedAt, ImplicitCode);
   }
 };