Add offload guard flags to typeck to prevent perf regressions

[Sa4dUs]

Fixes perf regression in #158693

r? @ZuseZ4

[rustbot]

r? @chenyukang

rustbot has assigned @chenyukang.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler
• compiler expanded to 75 candidates
• Random selection from 19 candidates

[ZuseZ4]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 48beb0f (48beb0fedb77a7ebf93b8a3b81bb05446f2c00d5)
Base parent: 7c329d6 (7c329d6c76e11ca40c5673818ab0439c1be8962c)

[rust-timer]

Finished benchmarking commit (48beb0f): comparison URL.

Overall result: ✅ improvements - no action needed

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

@bors rollup=never rustc-perf
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.4%	[-0.5%, -0.2%]	10
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results (primary -2.6%, secondary -0.1%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	2.7%	[1.0%, 4.8%]	3
Improvements ✅ (primary)	-2.6%	[-4.4%, -0.8%]	2
Improvements ✅ (secondary)	-1.7%	[-5.6%, -0.5%]	5
All ❌✅ (primary)	-2.6%	[-4.4%, -0.8%]	2

Cycles

Results (primary 0.1%, secondary 6.7%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	1.2%	[0.5%, 2.0%]	6
Regressions ❌ (secondary)	8.2%	[0.4%, 19.8%]	20
Improvements ✅ (primary)	-1.2%	[-2.0%, -0.5%]	5
Improvements ✅ (secondary)	-3.3%	[-5.0%, -1.2%]	3
All ❌✅ (primary)	0.1%	[-2.0%, 2.0%]	11

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 491.018s -> 491.074s (0.01%)
Artifact size: 390.29 MiB -> 391.01 MiB (0.18%)

[ZuseZ4]

One more location to try is #158693 (comment)

@panstromek GPU support unfortunately requires us to compile the same file 3 times, two of those for the same cpu target. So it's preferable to cache those checks where sensible.
Other than that, many-assoc-items cancels out precisely 1.0059 × (1−0.0054)=1.00046814, as well as derive `+/- 0.16%). Not sure about include-blob; maybe the last location I linked has an impact there? It also just had a huge improvement (33%), maybe that perturbed the significance numbers? There's also another 20% improvement coming up: #160453

[panstromek]

Yea, this looks great, thanks for looking into this so quickly. Detailed results show that the query hits are exactly inverse, too: before, after, same with dep graph sizes.

include-blob is probably noise, that one is bimodal, and it flips flops between two states pretty often, I think we can ignore that one. I wasn't sure whether the derive is relevant in the original, too (it's 300 structs), but good to see it improved (and I see now that it has a bunch of new query hits in the original, so it was probably relevant as well).

[ZuseZ4]

perfect, then I guess all relevant locations are covered
@bors r+ rollup=iffy Perf positive, but long queue

[rust-bors]

📌 Commit e04a81f has been approved by ZuseZ4

It is now in the queue for this repository.

🌲 The tree is currently closed for pull requests below priority 10. This pull request will be tested once the tree is reopened.

Reason for tree closure: manually handling queue due to backlog

[JonathanBrouwer]

Verifying that actual perf results after merge match expected results
@rust-timer build d41deaa

[rust-timer]

Finished benchmarking commit (d41deaa): comparison URL.

Overall result: ✅ improvements - no action needed

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

@bors rollup=never rustc-perf
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.4%	[-0.5%, -0.2%]	10
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results (primary 0.5%, secondary 0.8%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	0.5%	[0.4%, 0.6%]	4
Regressions ❌ (secondary)	2.2%	[0.6%, 5.0%]	4
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.0%	[-1.2%, -0.8%]	3
All ❌✅ (primary)	0.5%	[0.4%, 0.6%]	4

Cycles

Results (primary 0.4%, secondary 0.1%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	1.3%	[0.5%, 2.5%]	6
Regressions ❌ (secondary)	1.1%	[0.4%, 2.6%]	8
Improvements ✅ (primary)	-0.7%	[-1.0%, -0.4%]	5
Improvements ✅ (secondary)	-1.2%	[-2.7%, -0.4%]	6
All ❌✅ (primary)	0.4%	[-1.0%, 2.5%]	11

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 489.577s -> 489.898s (0.07%)
Artifact size: 390.24 MiB -> 390.29 MiB (0.01%)