[RFC007] Extract runtime contract equality out of typecheck #2130
+528
−8
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This continues the preparative work to migrate the typechecker to the new AST. One current blocker is that the `typecheck::eq` is used both by the typechecker and at runtime for contract deduplication; however we will get rid of it eventually, while still needing to perform contract deduplication (until the landing of the bytecode virtual machine). This commit extract the code from `typecheck::eq` for contract deduplication in a new module, so that the runtime isn't dependent on `typecheck` anymore. The code is specialized (as it doesn't need to handle static contract equality), and in order to eliminate the dependence on `UnifType`, we drop the ability to compare polymorphic type. The latter is the only reason `typecheck::eq` operated on `UnifType` instead of `Type`, but the motivation is not very convincing: how many times do we encounter a type annotation, itself inside a contract annotation, such that annotation is a polymorphic type AND it happens to be a candidate for deduplication? Our intuition would be: zero. This hypothesis will be validated on our working benchmarks (instrumentation counts each contract deduplication)
|
Bencher Report| Branch | rfc007/extract-contract-dedup |
| Testbed | ubuntu-latest |
Click to view all benchmark results
| Benchmark | Latency | nanoseconds (ns) |
|---|---|---|
| fibonacci 10 | 📈 view plot 🚷 view threshold | 493.22 |
| foldl arrays 50 | 📈 view plot 🚷 view threshold | 1,767.10 |
| foldl arrays 500 | 📈 view plot 🚷 view threshold | 6,883.70 |
| foldr strings 50 | 📈 view plot 🚷 view threshold | 7,156.70 |
| foldr strings 500 | 📈 view plot 🚷 view threshold | 62,739.00 |
| generate normal 250 | 📈 view plot 🚷 view threshold | 45,808.00 |
| generate normal 50 | 📈 view plot 🚷 view threshold | 2,172.60 |
| generate normal unchecked 1000 | 📈 view plot 🚷 view threshold | 3,438.70 |
| generate normal unchecked 200 | 📈 view plot 🚷 view threshold | 784.22 |
| pidigits 100 | 📈 view plot 🚷 view threshold | 3,189.00 |
| pipe normal 20 | 📈 view plot 🚷 view threshold | 1,516.70 |
| pipe normal 200 | 📈 view plot 🚷 view threshold | 10,362.00 |
| product 30 | 📈 view plot 🚷 view threshold | 847.27 |
| scalar 10 | 📈 view plot 🚷 view threshold | 1,541.30 |
| sum 30 | 📈 view plot 🚷 view threshold | 846.48 |
jneem
approved these changes
Dec 24, 2024
Co-authored-by: jneem <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR continues the preparation work to migrate the typechecker to the new AST. One current blocker is that
typecheck::eqis used both by the typechecker and at runtime for contract deduplication. We need to get rid of this module when switching the typechecker implementation, while still needing to perform contract deduplication at runtime (until the landing of the bytecode virtual machine).This PR extracts the code from
typecheck::eqfor contract deduplication in a new module, so that the runtime isn't dependent ontypecheckanymore. The code is specialized (as it doesn't need to handle static contract equality), and in order to eliminate the dependence onUnifType, we drop the ability to compare polymorphic types. The latter is the only reasontypecheck::eqoperated onUnifTypeinstead ofType(which makes it dependent on the old typechecker), but the motivation is not very convincing: how many times do we encounter a type annotation, itself inside a contract annotation, such that annotation is a polymorphic type AND it happens to be a candidate for deduplication? Our intuition would be: zero.This hypothesis has been validated on our working benchmarks, as
--metricsshow detailed data on contract deduplication - including the total count of deduplicated contracts - which doesn't change between this PR and master.