Overall, this week had very few regressions and a moderate amount of improvements. The two biggest improvements came in how metadata was being encoded including a change to only store StableCrateId once in DefPathTable which yielded a 0.3% average improvement across 79 different benchmarks.
Triage done by @rylev. Revision range: 1ab783112..67b6975
Summary:
| (instructions:u) | mean | range | count |
|---|---|---|---|
| Regressions ❌ (primary) |
0.7% | [0.3%, 1.5%] | 8 |
| Regressions ❌ (secondary) |
0.8% | [0.2%, 1.3%] | 23 |
| Improvements ✅ (primary) |
-0.6% | [-2.6%, -0.2%] | 121 |
| Improvements ✅ (secondary) |
-5.2% | [-62.5%, -0.2%] | 53 |
| All ❌✅ (primary) | -0.5% | [-2.6%, 1.5%] | 129 |
2 Regressions, 3 Improvements, 1 Mixed; 0 of them in rollups 46 artifact comparisons made in total
Emit better suggestions for &T == T and T == &T #118431 (Comparison Link)
| (instructions:u) | mean | range | count |
|---|---|---|---|
| Regressions ❌ (primary) |
24.0% | [18.1%, 29.5%] | 3 |
| Regressions ❌ (secondary) |
- | - | 0 |
| Improvements ✅ (primary) |
- | - | 0 |
| Improvements ✅ (secondary) |
- | - | 0 |
| All ❌✅ (primary) | 24.0% | [18.1%, 29.5%] | 3 |
- Blip in the perf infrastructure that was reversed in the next run
rustc_lint: Enforce rustc::potential_query_instability lint #119251 (Comparison Link)
| (instructions:u) | mean | range | count |
|---|---|---|---|
| Regressions ❌ (primary) |
0.2% | [0.2%, 0.3%] | 2 |
| Regressions ❌ (secondary) |
0.3% | [0.2%, 0.4%] | 8 |
| Improvements ✅ (primary) |
- | - | 0 |
| Improvements ✅ (secondary) |
- | - | 0 |
| All ❌✅ (primary) | 0.2% | [0.2%, 0.3%] | 2 |
- Small regression that might just be worth it. There is discussion in #119488 on whether to do a partial revert, but so far the argument is being made to just accept the small regression.
Suggest = to == in more cases, even in the face of reference mismatch #119328 (Comparison Link)
| (instructions:u) | mean | range | count |
|---|---|---|---|
| Regressions ❌ (primary) |
- | - | 0 |
| Regressions ❌ (secondary) |
- | - | 0 |
| Improvements ✅ (primary) |
-19.2% | [-22.7%, -15.3%] | 3 |
| Improvements ✅ (secondary) |
- | - | 0 |
| All ❌✅ (primary) | -19.2% | [-22.7%, -15.3%] | 3 |
- Reversal of perf infrastructure blip in previous run
Only store StableCrateId once in DefPathTable. #119259 (Comparison Link)
| (instructions:u) | mean | range | count |
|---|---|---|---|
| Regressions ❌ (primary) |
0.8% | [0.8%, 0.8%] | 1 |
| Regressions ❌ (secondary) |
- | - | 0 |
| Improvements ✅ (primary) |
-0.3% | [-0.6%, -0.2%] | 78 |
| Improvements ✅ (secondary) |
-0.4% | [-0.6%, -0.3%] | 16 |
| All ❌✅ (primary) | -0.3% | [-0.6%, 0.8%] | 79 |
Shrink span encoding further #119367 (Comparison Link)
| (instructions:u) | mean | range | count |
|---|---|---|---|
| Regressions ❌ (primary) |
- | - | 0 |
| Regressions ❌ (secondary) |
- | - | 0 |
| Improvements ✅ (primary) |
-0.3% | [-0.5%, -0.2%] | 27 |
| Improvements ✅ (secondary) |
-0.5% | [-0.6%, -0.3%] | 11 |
| All ❌✅ (primary) | -0.3% | [-0.5%, -0.2%] | 27 |
Implement constant propagation on top of MIR SSA analysis #116012 (Comparison Link)
| (instructions:u) | mean | range | count |
|---|---|---|---|
| Regressions ❌ (primary) |
0.7% | [0.2%, 1.9%] | 10 |
| Regressions ❌ (secondary) |
0.6% | [0.2%, 1.1%] | 15 |
| Improvements ✅ (primary) |
-0.6% | [-2.6%, -0.2%] | 26 |
| Improvements ✅ (secondary) |
-31.2% | [-62.1%, -0.2%] | 8 |
| All ❌✅ (primary) | -0.2% | [-2.6%, 1.9%] | 36 |
- Change has positive impacts on runtime performance and the improvements outweigh the regressions