tests/internal_bench: Benchmarks for descriptor-related features. #16825
Conversation
This file contains hidden or 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
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #16825 +/- ##
=======================================
Coverage 98.38% 98.38%
=======================================
Files 171 171
Lines 22257 22257
=======================================
Hits 21898 21898
Misses 359 359 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
04d3702 to
2409a66
Compare
This was referenced Feb 26, 2025
|
It looks like the unix port / settrace_stackless test failure is spurious? It failed in |
833837a to
270b002
Compare
dpgeorge
reviewed
Jul 21, 2025
There was a problem hiding this comment.
Thanks for this, it's a nice addition, quite a comprehensive set of tests. Also great to be able to now run it on a remote target.
I noticed that tests like internal_bench/var-1-constant.py fail on remote targets because they take too long, about 1 minute. I'd be OK to just leave that for now, unless you want to try and fix it.
dpgeorge
reviewed
Jul 21, 2025
Yeah, I figured it might just be a timeout --- the root cause here is actually just that the test functions in the Fixing this probably means needing to rewrite them to generate the code under test with a string interpolation and then |
|
Ok actually I found a much dumber way of making the var tests run --- I can just have the runner string-replace Probably an even worse syntactic sin than trying to parse HTML with regex --- but if you think about it, you can just define a sub-language with a grammar that only accepts the specific benchmarks that are actually in the folder as valid, and at least over that sub-language the literal string replacement is a correct parse. |
AJMansfield
added a commit
to AJMansfield/micropython
that referenced
this pull request
Jul 21, 2025
This PR adds support for the `__set_name__` data model method specified by PEP487 - Simpler customisation of class creation. This includes support for methods that mutate the owner class, and avoids the naive modify-while-iterating hazard possible in a naive implementation like micropython#15503. Note that based on the benchmarks in micropython#16825, this is also as fast or faster than the naive implementation, thanks to clever data layout in setname_list_t, and the way this allows the capture step to run during an existing loop through the class dict. Other rejected approaches for dealing with the hazard include: - python/cpython#72983 During the implementation of this feature for MicroPython, it was discovered that some versions of CPython also have this naive hazard. CPython resolved this bug in BPO-28797 and now makes a complete flat copy of the class's dict to iterate. This design decision doesn't make much sense for a microcontroller though, even if it's perfectly reasonable in the desktop world where memcpy might actually be cheaper than a hard-to-branch-predict conditional; and it's also motivated in their case by error-tracing considerations. - micropython#16816 This is an equivalent implementation to CPython's approach that places this copy directly on the stack; however it is both slower and has larger code size than the approach taken here. - micropython#15503 The simplest implementation is to just not worry about it and let the user face the consequences if they mutate the owner class. That's not a very friendly behavior, though, and it's not actually much more performant than this implementation on either time or code size. - micropython#17693 Another alternative is to do the same as micropython#15503 but leverage MicroPython's existing `is_fixed` field in its dict type to convert attempted mutations of the owner dict into `AttributeError`s. This is safer than just leaving the open hazard, but there's still important use-cases for owner-mutating descriptors, and the performance ain is small enough that it isn't worth missing support for those cases. - combined micropython#17693 with this Another version of this feature used a new feature define, `MICROPY_PY_METACLASSES_LITE`, to control whether this algorithm or the naive version is used. This was rejected in favor of simplicity, based on the very limited performance margin the naive version has (which in some cases even goes _against_ it). Signed-off-by: Anson Mansfield <amansfield@mantaro.com>
AJMansfield
added a commit
to AJMansfield/micropython
that referenced
this pull request
Jul 21, 2025
This PR adds support for the `__set_name__` data model method specified by PEP487 - Simpler customisation of class creation. This includes support for methods that mutate the owner class, and avoids the naive modify-while-iterating hazard possible in a naive implementation like micropython#15503. Note that based on the benchmarks in micropython#16825, this is also as fast or faster than the naive implementation, thanks to clever data layout in `setname_list_t`, and the way this allows the capture step to run during an existing loop through the class dict. Other rejected approaches for dealing with the hazard include: - python/cpython#72983 During the implementation of this feature for MicroPython, it was discovered that some versions of CPython also have this naive hazard. CPython resolved this bug in BPO-28797 and now makes a complete flat copy of the class's dict to iterate. This design decision doesn't make much sense for a microcontroller though, even if it's perfectly reasonable in the desktop world where memcpy might actually be cheaper than a hard-to-branch-predict conditional; and it's also motivated in their case by error-tracing considerations. - micropython#16816 This is an equivalent implementation to CPython's approach that places this copy directly on the stack; however it is both slower and has larger code size than the approach taken here. - micropython#15503 The simplest implementation is to just not worry about it and let the user face the consequences if they mutate the owner class. That's not a very friendly behavior, though, and it's not actually much more performant than this implementation on either time or code size. - micropython#17693 Another alternative is to do the same as micropython#15503 but leverage MicroPython's existing `is_fixed` field in its dict type to convert attempted mutations of the owner dict into `AttributeError`s. This is safer than just leaving the open hazard, but there's still important use-cases for owner-mutating descriptors, and the performance ain is small enough that it isn't worth missing support for those cases. - combined micropython#17693 with this Another version of this feature used a new feature define, `MICROPY_PY_METACLASSES_LITE`, to control whether this algorithm or the naive version is used. This was rejected in favor of simplicity, based on the very limited performance margin the naive version has (which in some cases even goes _against_ it). Signed-off-by: Anson Mansfield <amansfield@mantaro.com>
AJMansfield
added a commit
to AJMansfield/micropython
that referenced
this pull request
Jul 23, 2025
This PR adds support for the `__set_name__` data model method specified by PEP487 - Simpler customisation of class creation. This includes support for methods that mutate the owner class, and avoids the naive modify-while-iterating hazard possible in a naive implementation like micropython#15503. Note that based on the benchmarks in micropython#16825, this is also as fast or faster than the naive implementation, thanks to clever data layout in `setname_list_t`, and the way this allows the capture step to run during an existing loop through the class dict. Other rejected approaches for dealing with the hazard include: - python/cpython#72983 During the implementation of this feature for MicroPython, it was discovered that some versions of CPython also have this naive hazard. CPython resolved this bug in BPO-28797 and now makes a complete flat copy of the class's dict to iterate. This design decision doesn't make much sense for a microcontroller though, even if it's perfectly reasonable in the desktop world where memcpy might actually be cheaper than a hard-to-branch-predict conditional; and it's also motivated in their case by error-tracing considerations. - micropython#16816 This is an equivalent implementation to CPython's approach that places this copy directly on the stack; however it is both slower and has larger code size than the approach taken here. - micropython#15503 The simplest implementation is to just not worry about it and let the user face the consequences if they mutate the owner class. That's not a very friendly behavior, though, and it's not actually much more performant than this implementation on either time or code size. - micropython#17693 Another alternative is to do the same as micropython#15503 but leverage MicroPython's existing `is_fixed` field in its dict type to convert attempted mutations of the owner dict into `AttributeError`s. This is safer than just leaving the open hazard, but there's still important use-cases for owner-mutating descriptors, and the performance ain is small enough that it isn't worth missing support for those cases. - combined micropython#17693 with this Another version of this feature used a new feature define, `MICROPY_PY_METACLASSES_LITE`, to control whether this algorithm or the naive version is used. This was rejected in favor of simplicity, based on the very limited performance margin the naive version has (which in some cases even goes _against_ it). Signed-off-by: Anson Mansfield <amansfield@mantaro.com>
AJMansfield
added a commit
to AJMansfield/micropython
that referenced
this pull request
Jul 28, 2025
This PR adds support for the `__set_name__` data model method specified by PEP487 - Simpler customisation of class creation. This includes support for methods that mutate the owner class, and avoids the naive modify-while-iterating hazard possible in a naive implementation like micropython#15503. Note that based on the benchmarks in micropython#16825, this is also as fast or faster than the naive implementation, thanks to clever data layout in `setname_list_t`, and the way this allows the capture step to run during an existing loop through the class dict. Other rejected approaches for dealing with the hazard include: - python/cpython#72983 During the implementation of this feature for MicroPython, it was discovered that some versions of CPython also have this naive hazard. CPython resolved this bug in BPO-28797 and now makes a complete flat copy of the class's dict to iterate. This design decision doesn't make much sense for a microcontroller though, even if it's perfectly reasonable in the desktop world where memcpy might actually be cheaper than a hard-to-branch-predict conditional; and it's also motivated in their case by error-tracing considerations. - micropython#16816 This is an equivalent implementation to CPython's approach that places this copy directly on the stack; however it is both slower and has larger code size than the approach taken here. - micropython#15503 The simplest implementation is to just not worry about it and let the user face the consequences if they mutate the owner class. That's not a very friendly behavior, though, and it's not actually much more performant than this implementation on either time or code size. - micropython#17693 Another alternative is to do the same as micropython#15503 but leverage MicroPython's existing `is_fixed` field in its dict type to convert attempted mutations of the owner dict into `AttributeError`s. This is safer than just leaving the open hazard, but there's still important use-cases for owner-mutating descriptors, and the performance ain is small enough that it isn't worth missing support for those cases. - combined micropython#17693 with this Another version of this feature used a new feature define, `MICROPY_PY_METACLASSES_LITE`, to control whether this algorithm or the naive version is used. This was rejected in favor of simplicity, based on the very limited performance margin the naive version has (which in some cases even goes _against_ it). Signed-off-by: Anson Mansfield <amansfield@mantaro.com>
dpgeorge
pushed a commit
to AJMansfield/micropython
that referenced
this pull request
Jul 28, 2025
This commit adds support for the `__set_name__` data model method specified by PEP487 - Simpler customisation of class creation. This includes support for methods that mutate the owner class, and avoids the naive modify-while-iterating hazard possible in a naive implementation like micropython#15503. Note that based on the benchmarks in micropython#16825, this is also as fast or faster than the naive implementation, thanks to clever data layout in `setname_list_t`, and the way this allows the capture step to run during an existing loop through the class dict. Other rejected approaches for dealing with the hazard include: - python/cpython#72983 During the implementation of this feature for MicroPython, it was discovered that some versions of CPython also have this naive hazard. CPython resolved this bug in BPO-28797 and now makes a complete flat copy of the class's dict to iterate. This design decision doesn't make much sense for a microcontroller though, even if it's perfectly reasonable in the desktop world where memcpy might actually be cheaper than a hard-to-branch-predict conditional; and it's also motivated in their case by error-tracing considerations. - micropython#16816 This is an equivalent implementation to CPython's approach that places this copy directly on the stack; however it is both slower and has larger code size than the approach taken here. - micropython#15503 The simplest implementation is to just not worry about it and let the user face the consequences if they mutate the owner class. That's not a very friendly behavior, though, and it's not actually much more performant than this implementation on either time or code size. - micropython#17693 Another alternative is to do the same as micropython#15503 but leverage MicroPython's existing `is_fixed` field in its dict type to convert attempted mutations of the owner dict into `AttributeError`s. This is safer than just leaving the open hazard, but there's still important use-cases for owner-mutating descriptors, and the performance gain is small enough that it isn't worth missing support for those cases. - combined micropython#17693 with this Another version of this feature used a new feature define, `MICROPY_PY_METACLASSES_LITE`, to control whether this algorithm or the naive version is used. This was rejected in favor of simplicity, based on the very limited performance margin the naive version has (which in some cases even goes _against_ it). Signed-off-by: Anson Mansfield <amansfield@mantaro.com>
Signed-off-by: Anson Mansfield <amansfield@mantaro.com>
Signed-off-by: Anson Mansfield <amansfield@mantaro.com>
dpgeorge
approved these changes
Jul 29, 2025
Signed-off-by: Anson Mansfield <amansfield@mantaro.com>
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.
Summary
This includes new
internal_benchbenchmarks for how quickly micropython can process class definitions, created in order to gauge the performance penalty for implementing__set_name__in the different ways proposed in #15503, #16806, and #16816.It also includes an update to
run-internalbench.pythat adds in the same ability to run the benchmarks on remote micropython instances (e.g. actual microcontroller hardware) that the mainrun-tests.pyhas.Testing
I've tested these benchmarks using the unix port inside WSL on my development machine, with the following results:
Trade-offs and Alternatives
Due to the overall slow speed of class creation in general, the
class_createtests reduce the number of iterations from 20,000,000 to just 500,000; but this should still be wide enough to give the results adequate statistical significance.For microcontroller remotes it uses 200,000 as the base default (which in the class_create case gets divided down to 5,000 runs).