Introduce benchmarking API #2437
Merged
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
bjorn3
reviewed
Nov 20, 2020
|
I created a draft based on @fitzgen's comment re: a Wasmtime benchmark dylib. I don't really expect this to be the final product but I wanted to make sure that I could use the API we discussed over in sightglass (and what better way to know than to actually build it?). I am not quite sure about:
|
abrown
force-pushed
the
bench-api
branch
2 times, most recently
from
5b1bdbb to
0567e6e
Compare
fitzgen
reviewed
Nov 23, 2020
Comment on lines
+101
to
+107
| /// Opaque pointer type for hiding the engine state details. | ||
| #[repr(C)] | ||
| pub struct OpaqueEngineState { | ||
| _private: [u8; 0], | ||
| } |
There was a problem hiding this comment.
I don't think we need to actually have a new type here. We should be able to pass *mut EngineState out and expect that callers won't know how to access its fields or anything.
| struct EngineState<'a> { | ||
| bytes: &'a [u8], | ||
| engine: Engine, | ||
| store: Store, |
There was a problem hiding this comment.
Worth noting in a comment that if we ever want to measure more than one instantiation per process, we will probably want to create a new store for each instantiation.
Hi @abrown, to be clear, so this is adding a crate that when built will build a shared library that includes everything for embedding enough of the wasmtime runtime into whatever (sightglass for example) for instantiating and running a wasm file? Doesn't the c-api already expose api's to load and instantiate and execute modules? If so, what are the notable differences with what's here? Also the bench_start and bench_end .. if someone has a C program or rust files and want to add these markers around a specific function and recompile to wasm .. how is this shared library that is produced here used in that process? |
Yes.
It does, but this is quite a bit simpler. I did create bindings to use the Wasm C API from Rust but this was quite a bit more work than this approach (a lot of unsafe things are yet to be figured out there). Either way is fine with me.
This shared library is decoupled from that completely. That way benchmark artifacts can be created apart from how they are run. There is an example of how this works here |
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
abrown
added a commit
to abrown/sightglass
that referenced
this pull request
Dec 10, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
The new crate introduced here, `wasmtime-bench-api`, creates a shared library, e.g. `wasmtime_bench_api.so`, for executing Wasm benchmarks using Wasmtime. It allows us to measure several phases separately by exposing `engine_compile_module`, `engine_instantiate_module`, and `engine_execute_module`, which pass around an opaque pointer to the internally initialized state. This state is initialized and freed by `engine_create` and `engine_free`, respectively. The API also introduces a way of passing in functions to satisfy the `"bench" "start"` and `"bench" "end"` symbols that we expect Wasm benchmarks to import. The API is exposed in a C-compatible way so that we can dynamically load it (carefully) in our benchmark runner.
abrown
added a commit
to bytecodealliance/sightglass
that referenced
this pull request
Dec 11, 2020
This change brings in work done in a separate repository to create a new benchmark runner for sightglass. The original code is retained with a `legacy` moniker where appropriate; the new runner uses `next`. The `README.md` now contains a description of the different components present in the repository and links to documentation or implementation for each. In writing this, it clarified for me that a reorganization (in a future PR) is needed. The new runner employs a different paradigm than the old one: - benchmarks are Wasm files, not shared libraries, built using Dockerfiles and committed in-tree (accompanied by their WAT pair) under `benchmarks-next`; they import `bench.start` and `bench.end` functions which are placed around the section of code to measure - engines (i.e. Wasm VMs) are shared libraries, like the one introduced in bytecodealliance/wasmtime#2437, allowing the runner to control the compilation, instantiation, and execution of the Wasm benchmark - the building and running of the benchmarks is controlled with a CLI tool, see `cargo +nightly run -- --help` and `docs/next.md` for more information
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.
The new crate introduced here,
wasmtime-bench-api, creates a shared library, e.g.wasmtime_bench_api.so, for executing Wasm benchmarks using Wasmtime. It allows us to measure several phases separately by exposingengine_compile_module,engine_instantiate_module, andengine_execute_module, which pass around an opaque pointer to the internally initialized state. This state is initialized and freed byengine_createandengine_free, respectively. The API also introduces a way of passing in functions to satisfy the"bench" "start"and"bench" "end"symbols that we expect Wasm benchmarks to import. The API is exposed in a C-compatible way so that we can dynamically load it (carefully) in our benchmark runner.