expand README.md #124
Merged
expand README.md #124
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
zero9178
reviewed
Aug 30, 2024
README.md
Outdated
Comment on lines
25
to
26
| The docker toolchain needs to be in sync with the snitch runtime, as the hardware gets continuously updated. | ||
| When updating the snitch runtime submodule, the docker image needs to be updated with the new commit hash to make sure the hardware and software are in sync. |
There was a problem hiding this comment.
This seems to be either a duplicate sentence
README.md
Outdated
| The docker toolchain needs to be in sync with the snitch runtime, as the hardware gets continuously updated. | ||
| When updating the snitch runtime submodule, the docker image needs to be updated with the new commit hash to make sure the hardware and software are in sync. | ||
|
|
||
| The easiest way to iterate on the docker image is to develop on it locally, and then pushing it to GitHub, which will detect that the toolchain directory is changed, and will release a new image automatically. |
There was a problem hiding this comment.
Suggested change
| The easiest way to iterate on the docker image is to develop on it locally, and then pushing it to GitHub, which will detect that the toolchain directory is changed, and will release a new image automatically. | |
| The easiest way to iterate on the docker image is to develop on it locally, and then push the changes to GitHub, which will detect that the toolchain directory is changed, and will release a new image automatically. |
README.md
Outdated
| ## Compiling Neural Networks | ||
|
|
||
| Any information you can find on how to use `iree-compile` is applicable, with the only major difference bing the `--iree-hal-target-backends=llvm-cpu`, which also supports our `quidditch` value for the flag, which leverages our flow. | ||
| The inputs to `iree-compile` are all `.mlir` files, so you need to somehow get the file, [here's](https://iree.dev/guides/deployment-configurations/cpu/#compile-and-run-a-program) the guide on how to get one. |
There was a problem hiding this comment.
Suggested change
| The inputs to `iree-compile` are all `.mlir` files, so you need to somehow get the file, [here's](https://iree.dev/guides/deployment-configurations/cpu/#compile-and-run-a-program) the guide on how to get one. | |
| The inputs to `iree-compile` are all `.mlir` files, so you need to somehow get the file, [here's](https://iree.dev/guides/ml-frameworks/) the guide on how to get one. |
is probably the better link
README.md
Outdated
|
|
||
| The PyTorch flow uses [iree-turbine](https://github.com/iree-org/iree-turbine), which is included in the virtualenv of the Docker container. | ||
| We use it to compile nsnet2 (`runtime/samples/nsnet2/NsNet2.py`). | ||
| The MLIR file extracted this way is that it will contain functions that are named the same as the Python model functions. |
There was a problem hiding this comment.
Suggested change
| The MLIR file extracted this way is that it will contain functions that are named the same as the Python model functions. | |
| The MLIR file extracted this way will contain functions that are named the same as the Python model functions. |
README.md
Outdated
|
|
||
| The `quiddtch_module` macro in the `CMakeLists.txt` turns the MLIR file into an executable. | ||
| `ASSERT_XDSL` is a flag that will error out if xDSL micro-kernel compilation fails. | ||
| This macro creates an executable called `mynewmodel` (removing the `.mlir` suffix). |
There was a problem hiding this comment.
Suggested change
| This macro creates an executable called `mynewmodel` (removing the `.mlir` suffix). | |
| This macro creates a library called `mynewmodel` (removing the `.mlir` suffix). |
README.md
Outdated
| 4. The cmake build system mirrors your source directory, so if you added `runtime/samples/mynewmodel/`, the build location will be `build/runtime/samples/mynewmodel`. | ||
| 5. Add your new executable to `tests/CMakeLists.txt` | ||
|
|
||
| > NB. you can override the compilation with xDSL by adding `LLVM` to the `uidditch_module` macro, which overrides the XDSL pipeline entirely and just computes everything with stock IREE `llvm-cpu` target. |
There was a problem hiding this comment.
Suggested change
| > NB. you can override the compilation with xDSL by adding `LLVM` to the `uidditch_module` macro, which overrides the XDSL pipeline entirely and just computes everything with stock IREE `llvm-cpu` target. | |
| > NB. you can override the compilation with xDSL by adding `LLVM` to the `quidditch_module` macro, which overrides the XDSL pipeline entirely and just computes everything with stock IREE `llvm-cpu` target. |
|
|
||
| ## Quidditch Compiler Structure | ||
|
|
||
| All source code is in `codegen/compiler/src/Quirritch`. |
There was a problem hiding this comment.
Suggested change
| All source code is in `codegen/compiler/src/Quirritch`. | |
| All source code is in `codegen/compiler/src/Quidditch`. |
README.md
Outdated
|
|
||
| `QuidditchTarget.cpp` | ||
|
|
||
| The QuidditchTargetBackend class sibclasses IREE::HAL::TargetBackend. |
There was a problem hiding this comment.
Suggested change
| The QuidditchTargetBackend class sibclasses IREE::HAL::TargetBackend. | |
| The QuidditchTargetBackend class subclasses IREE::HAL::TargetBackend. |
README.md
Outdated
|
|
||
| ### Dialect | ||
|
|
||
| There is currently only one dialect, `snitch`, which might be split up in the future. |
There was a problem hiding this comment.
Suggested change
| There is currently only one dialect, `snitch`, which might be split up in the future. | |
| There is currently only one dialect, `quidditch_snitch`, which might be split up in the future. |
zero9178
approved these changes
Sep 17, 2024
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.
These are the notes I took, as well as the existing README. Could you please double-check that the information is correct, and let me know if anything needs to changed?
CC @compor