Skip to content

Commit 9580e15

Browse files
Mark-SimulacrumMark-Simulacrum
authored
Merge pull request #1834 from Mark-Simulacrum/triage
Add triage for this week
2 parents 1e43daf + 35c0398 commit 9580e15

File tree

1 file changed

+191
-0
lines changed

1 file changed

+191
-0
lines changed

‎triage/2024-03-05.md

+191
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,191 @@
1+
# 2024-03-05 Triage Log
2+
3+
A bunch of noise this week which has been dropped from the report (but may be
4+
present in the summary figures). As a result, the week is pretty busy in amount
5+
of changes, but the net effect is nearly neutral to a slight regression for
6+
most workloads.
7+
8+
Triage done by **@simulacrum**.
9+
Revision range: [71ffdf7ff7ac6df5f9f64de7e780b8345797e8a0..41d97c8a5dea2731b0e56fe97cd7cb79e21cff79](https://perf.rust-lang.org/?start=71ffdf7ff7ac6df5f9f64de7e780b8345797e8a0&end=41d97c8a5dea2731b0e56fe97cd7cb79e21cff79&absolute=false&stat=instructions%3Au)
10+
11+
**Summary**:
12+
13+
| (instructions:u) | mean | range | count |
14+
|:----------------------------------:|:-----:|:--------------:|:-----:|
15+
| Regressions ❌ <br /> (primary) | 0.6% | [0.1%, 2.0%] | 136 |
16+
| Regressions ❌ <br /> (secondary) | 0.8% | [0.2%, 2.6%] | 78 |
17+
| Improvements ✅ <br /> (primary) | -0.6% | [-1.2%, -0.3%] | 9 |
18+
| Improvements ✅ <br /> (secondary) | -0.6% | [-1.0%, -0.2%] | 14 |
19+
| All ❌✅ (primary) | 0.6% | [-1.2%, 2.0%] | 145 |
20+
21+
22+
2 Regressions, 0 Improvements, 10 Mixed; 4 of them in rollups
23+
51 artifact comparisons made in total
24+
25+
#### Regressions
26+
27+
Weekly `cargo update` [#112865](https://github.com/rust-lang/rust/pull/112865) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=0decdac390cfeedcd7f2f44c45f72c59c70d8143&end=da02fff3b6e4e27156054dcdda6675fe2a2591a6&stat=instructions:u)
28+
29+
| (instructions:u) | mean | range | count |
30+
|:----------------------------------:|:----:|:------------:|:-----:|
31+
| Regressions ❌ <br /> (primary) | 1.1% | [0.2%, 2.9%] | 3 |
32+
| Regressions ❌ <br /> (secondary) | 1.1% | [0.2%, 1.9%] | 7 |
33+
| Improvements ✅ <br /> (primary) | - | - | 0 |
34+
| Improvements ✅ <br /> (secondary) | - | - | 0 |
35+
| All ❌✅ (primary) | 1.1% | [0.2%, 2.9%] | 3 |
36+
37+
Doesn't appear to be entirely noise (though some of the delta is likely due to
38+
bimodality). However marking this regression as triaged since investigating is
39+
likely to be painful and the regressions are predominantly in secondary
40+
benchmarks.
41+
42+
Add a scheme for moving away from `extern "rust-intrinsic"` entirely [#120675](https://github.com/rust-lang/rust/pull/120675) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=50e77f133f8eb1f745e05681163a0143d6c4dd7d&end=2eeff462b762ed00f9d557d8c6ad7a3d562f692a&stat=instructions:u)
43+
44+
| (instructions:u) | mean | range | count |
45+
|:----------------------------------:|:----:|:------------:|:-----:|
46+
| Regressions ❌ <br /> (primary) | 0.6% | [0.2%, 1.0%] | 93 |
47+
| Regressions ❌ <br /> (secondary) | 0.8% | [0.2%, 2.4%] | 28 |
48+
| Improvements ✅ <br /> (primary) | - | - | 0 |
49+
| Improvements ✅ <br /> (secondary) | - | - | 0 |
50+
| All ❌✅ (primary) | 0.6% | [0.2%, 1.0%] | 93 |
51+
52+
This regression is being addressed in a followup PR:
53+
https://github.com/rust-lang/rust/pull/122010
54+
55+
#### Mixed
56+
57+
syms for legacy numeric constants diag items [#121667](https://github.com/rust-lang/rust/pull/121667) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=91cae1dcdcf1a31bd8a92e4a63793d65cfe289bb&end=1c28a2c1b0b82b525262e6ccc7675cab61ed040c&stat=instructions:u)
58+
59+
| (instructions:u) | mean | range | count |
60+
|:----------------------------------:|:-----:|:--------------:|:-----:|
61+
| Regressions ❌ <br /> (primary) | - | - | 0 |
62+
| Regressions ❌ <br /> (secondary) | 0.5% | [0.3%, 1.1%] | 3 |
63+
| Improvements ✅ <br /> (primary) | -0.3% | [-0.3%, -0.2%] | 2 |
64+
| Improvements ✅ <br /> (secondary) | -0.8% | [-1.5%, -0.2%] | 3 |
65+
| All ❌✅ (primary) | -0.3% | [-0.3%, -0.2%] | 2 |
66+
67+
Regression seems likely to be noise (bimodality) or not significant enough to
68+
investigate further. Marking as triaged. Some of the improvements seem genuine.
69+
70+
Diagnostic renaming [#121489](https://github.com/rust-lang/rust/pull/121489) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=bf9c7a64ad222b85397573668b39e6d1ab9f4a72&end=c475e2303b551d726307c646181e0677af1e0069&stat=instructions:u)
71+
72+
| (instructions:u) | mean | range | count |
73+
|:----------------------------------:|:-----:|:--------------:|:-----:|
74+
| Regressions ❌ <br /> (primary) | - | - | 0 |
75+
| Regressions ❌ <br /> (secondary) | 1.2% | [1.2%, 1.2%] | 1 |
76+
| Improvements ✅ <br /> (primary) | -0.9% | [-0.9%, -0.8%] | 4 |
77+
| Improvements ✅ <br /> (secondary) | -0.5% | [-0.8%, -0.4%] | 9 |
78+
| All ❌✅ (primary) | -0.9% | [-0.9%, -0.8%] | 4 |
79+
80+
Regression is essentially guaranteed to be bimodality, not a real change. Marking as triaged.
81+
82+
Rollup of 10 pull requests [#121790](https://github.com/rust-lang/rust/pull/121790) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=d3d145ea1cae47ad392173f890577788117da3d9&end=71a7b66f20c551f640f2f382bc7e7923ba0a5dab&stat=instructions:u)
83+
84+
| (instructions:u) | mean | range | count |
85+
|:----------------------------------:|:-----:|:--------------:|:-----:|
86+
| Regressions ❌ <br /> (primary) | 0.6% | [0.6%, 0.6%] | 1 |
87+
| Regressions ❌ <br /> (secondary) | - | - | 0 |
88+
| Improvements ✅ <br /> (primary) | -0.4% | [-0.6%, -0.3%] | 2 |
89+
| Improvements ✅ <br /> (secondary) | -1.9% | [-2.4%, -1.5%] | 2 |
90+
| All ❌✅ (primary) | -0.1% | [-0.6%, 0.6%] | 3 |
91+
92+
The single primary benchmark regression (Cargo) change looks to be both above
93+
any noise floor and does not show signs of reverting. However, it's not
94+
significant enough (given limit to just Cargo) to investigate further.
95+
96+
Rollup of 7 pull requests [#121804](https://github.com/rust-lang/rust/pull/121804) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=384d26fc7e3bdd7687cc17b2662b091f6017ec2a&end=1a1876c9790f168fb51afa335a7ba3e6fc267d75&stat=instructions:u)
97+
98+
| (instructions:u) | mean | range | count |
99+
|:----------------------------------:|:-----:|:--------------:|:-----:|
100+
| Regressions ❌ <br /> (primary) | 0.3% | [0.2%, 0.6%] | 7 |
101+
| Regressions ❌ <br /> (secondary) | 1.8% | [1.7%, 1.9%] | 6 |
102+
| Improvements ✅ <br /> (primary) | -0.8% | [-0.9%, -0.8%] | 4 |
103+
| Improvements ✅ <br /> (secondary) | -0.5% | [-0.8%, -0.2%] | 8 |
104+
| All ❌✅ (primary) | -0.1% | [-0.9%, 0.6%] | 11 |
105+
106+
Trying to track down the cause of the regression. Suspecting #121000 based on
107+
the regressed benchmarks.
108+
109+
Combine `Sub` and `Equate` [#121462](https://github.com/rust-lang/rust/pull/121462) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=6cbf0926d54c80ea6d15df333be9281f65bbeb36&end=b0696a5160711c068cb1f01b7437db7990d15750&stat=instructions:u)
110+
111+
| (instructions:u) | mean | range | count |
112+
|:----------------------------------:|:-----:|:--------------:|:-----:|
113+
| Regressions ❌ <br /> (primary) | 0.5% | [0.2%, 0.9%] | 13 |
114+
| Regressions ❌ <br /> (secondary) | 0.5% | [0.3%, 0.8%] | 17 |
115+
| Improvements ✅ <br /> (primary) | - | - | 0 |
116+
| Improvements ✅ <br /> (secondary) | -0.5% | [-0.9%, -0.4%] | 7 |
117+
| All ❌✅ (primary) | 0.5% | [0.2%, 0.9%] | 13 |
118+
119+
Future PRs may improve perf here, but this is targeting clean up and preparing for other refactoring first:
120+
121+
> Some ideas how to reduce the perf impact. Even if they don't completely
122+
> remove it, this will allow significant improvements and cleanups going
123+
> forward and generally simplifies the core type system. I would merge this
124+
> even with these regressions.
125+
126+
(https://github.com/rust-lang/rust/pull/121462#issuecomment-1973282815)
127+
128+
Rollup of 12 pull requests [#121859](https://github.com/rust-lang/rust/pull/121859) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=17edacef07e8afc3b580ed8feead6c5e90d24a56&end=2dceda4f32b97f60b122f2b32491e0267ef5cc0c&stat=instructions:u)
129+
130+
| (instructions:u) | mean | range | count |
131+
|:----------------------------------:|:-----:|:--------------:|:-----:|
132+
| Regressions ❌ <br /> (primary) | 1.9% | [0.5%, 3.3%] | 2 |
133+
| Regressions ❌ <br /> (secondary) | - | - | 0 |
134+
| Improvements ✅ <br /> (primary) | -0.8% | [-1.1%, -0.5%] | 2 |
135+
| Improvements ✅ <br /> (secondary) | - | - | 0 |
136+
| All ❌✅ (primary) | 0.6% | [-1.1%, 3.3%] | 4 |
137+
138+
Cargo regression looks real but otherwise this looks like noise to me. I'm
139+
going to mark as triaged, I don't think it merits digging through individual
140+
PRs to try and isolate it.
141+
142+
Don't grab variances in `TypeRelating` relation if we're invariant [#121864](https://github.com/rust-lang/rust/pull/121864) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=4cdd20584ccb75890d7d9bfae266054abfae5d46&end=2e3581bca93fdcce474e17cd43430b594a7955a0&stat=instructions:u)
143+
144+
| (instructions:u) | mean | range | count |
145+
|:----------------------------------:|:-----:|:--------------:|:-----:|
146+
| Regressions ❌ <br /> (primary) | - | - | 0 |
147+
| Regressions ❌ <br /> (secondary) | 0.2% | [0.2%, 0.2%] | 1 |
148+
| Improvements ✅ <br /> (primary) | -0.6% | [-0.8%, -0.2%] | 11 |
149+
| Improvements ✅ <br /> (secondary) | -0.5% | [-0.6%, -0.4%] | 8 |
150+
| All ❌✅ (primary) | -0.6% | [-0.8%, -0.2%] | 11 |
151+
152+
Regression is likely real, but is in a secondary benchmark and incr-unchanged
153+
scenario. Not worth further investigation, particularly given the improvements.
154+
155+
Always generate GEP i8 / ptradd for struct offsets [#121665](https://github.com/rust-lang/rust/pull/121665) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=516b6162a2ea8e66678c09e8243ebd83e4b8eeea&end=70aa0b86c066e721012852a9851fdf8586117823&stat=instructions:u)
156+
157+
| (instructions:u) | mean | range | count |
158+
|:----------------------------------:|:-----:|:--------------:|:-----:|
159+
| Regressions ❌ <br /> (primary) | 0.3% | [0.2%, 0.4%] | 3 |
160+
| Regressions ❌ <br /> (secondary) | 0.5% | [0.2%, 1.0%] | 12 |
161+
| Improvements ✅ <br /> (primary) | -0.7% | [-1.6%, -0.3%] | 14 |
162+
| Improvements ✅ <br /> (secondary) | -0.5% | [-1.0%, -0.3%] | 12 |
163+
| All ❌✅ (primary) | -0.5% | [-1.6%, 0.4%] | 17 |
164+
165+
Changes are expected since this changed a lot of our codegen. In practice
166+
though mostly not meaningful (i.e., some improvements, some regressions).
167+
168+
Rollup of 5 pull requests [#121955](https://github.com/rust-lang/rust/pull/121955) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=70aa0b86c066e721012852a9851fdf8586117823&end=89b78304e82dc5114e3b2faa0fbec747a28a2b37&stat=instructions:u)
169+
170+
| (instructions:u) | mean | range | count |
171+
|:----------------------------------:|:-----:|:--------------:|:-----:|
172+
| Regressions ❌ <br /> (primary) | 0.7% | [0.2%, 3.7%] | 17 |
173+
| Regressions ❌ <br /> (secondary) | - | - | 0 |
174+
| Improvements ✅ <br /> (primary) | -0.4% | [-0.8%, -0.2%] | 10 |
175+
| Improvements ✅ <br /> (secondary) | - | - | 0 |
176+
| All ❌✅ (primary) | 0.3% | [-0.8%, 3.7%] | 27 |
177+
178+
Likely to be caused by a correctness fix: https://github.com/rust-lang/rust/pull/121955#issuecomment-1975993842
179+
180+
perf: improve write_fmt to handle simple strings [#121001](https://github.com/rust-lang/rust/pull/121001) [(Comparison Link)](https://perf.rust-lang.org/compare.html?start=1547c076bfec8abb819d6a81e1e4095d267bd5b4&end=5a1e5449c8f4cb6b12b4f64238e3c058767ebf02&stat=instructions:u)
181+
182+
| (instructions:u) | mean | range | count |
183+
|:----------------------------------:|:-----:|:--------------:|:-----:|
184+
| Regressions ❌ <br /> (primary) | 1.7% | [0.5%, 3.6%] | 4 |
185+
| Regressions ❌ <br /> (secondary) | 0.3% | [0.3%, 0.3%] | 1 |
186+
| Improvements ✅ <br /> (primary) | - | - | 0 |
187+
| Improvements ✅ <br /> (secondary) | -1.3% | [-1.3%, -1.3%] | 1 |
188+
| All ❌✅ (primary) | 1.7% | [0.5%, 3.6%] | 4 |
189+
190+
Improvements to runtime (or at least assembly quality), at the cost of a bit
191+
more time in LLVM.

0 commit comments

Comments
 (0)