Add a new tutorial for compile time caching

[oulgen]

Previous tutorial was not fleshed out and primary talked about configurations.

cc @williamwen42 @msaroufim @anijain2305

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/3277

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 New Failure

As of commit 8f5fd59 with merge base a543d05 ():

NEW FAILURE - The following job has failed:

• Check spelling / pyspelling (gh)
  Process completed with exit code 1.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[svekars]

Please put all the names of methods/APIs in double backticks:

Suggested change

Most of these caches are in-memory, only used within the same process, and are transparent to the user. An exception is caches that store compiled FX graphs (FXGraphCache, AOTAutogradCache). These caches allow Inductor to avoid recompilation across process boundaries when it encounters the same graph with the same Tensor input shapes (and the same configuration). The default implementation stores compiled artifacts in the system temp directory. An optional feature also supports sharing those artifacts within a cluster by storing them in a Redis database.

Most of these caches are in-memory, only used within the same process, and are transparent to the user. An exception is caches that store compiled FX graphs (``FXGraphCache``, ``AOTAutogradCache``). These caches allow Inductor to avoid recompilation across process boundaries when it encounters the same graph with the same Tensor input shapes (and the same configuration). The default implementation stores compiled artifacts in the system temp directory. An optional feature also supports sharing those artifacts within a cluster by storing them in a Redis database.

[svekars]

Suggested change

	The intented use case is after compiling and executing a model, the user calls ``torch.compiler.save_cache_artifacts()`` which will return the compiler artifacts in a portable form. Later, potentially on a different machine, the user may call ``torch.compiler.load_cache_artifacts()`` with these artifacts to prepopulate the ``torch.compile`` caches in order to jump-start their cache.
	The intended use case is after compiling and executing a model, the user calls ``torch.compiler.save_cache_artifacts()`` which will return the compiler artifacts in a portable form. Later, potentially on a different machine, the user may call ``torch.compiler.load_cache_artifacts()`` with these artifacts to prepopulate the ``torch.compile`` caches in order to jump-start their cache.

[svekars]

Suggested change

	``torch.compile`` end-to-end caching (a.k.a. ``Mega-Cache``)
	``torch.compile`` end-to-end caching

[svekars]

Suggested change

	``Mega-Cache`` provides two compiler APIs
	``Mega-Cache`` provides two compiler APIs:

[svekars]

Suggested change

	An example to this is as follows. First, compile and save the cache artifacts.
	Consider the following example. First, compile and save the cache artifacts:

[svekars]

Suggested change

	Later, the user can jump-start their cache by the following.
	Later, you can jump-start the cache by the following:

[svekars]

Suggested change

	The above described MegaCache is also compromised of individual components that can be used without any user intervention. By default, PyTorch Compiler comes with local on-disk caches for ``TorchDynamo``, ``TorchInductor``, and ``Triton``. These caches are as following.
	The aforementioned ``MegaCache``is composed of individual components that can be used without any user intervention. By default, PyTorch Compiler comes with local on-disk caches for ``TorchDynamo``, ``TorchInductor``, and ``Triton``. These caches include:

[svekars]

Suggested change

	Like ``TORCHINDUCTOR_FX_GRAPH_REMOTE_CACHE``, this setting enables the remote ``AOTAutogradCache`` feature. The current implementation uses Redis. ``1`` enables caching, and any other value disables it. The following environment variables configure the host and port of the ``Redis`` server:
	Similar to ``TORCHINDUCTOR_FX_GRAPH_REMOTE_CACHE``, this setting enables the remote ``AOTAutogradCache`` feature. The current implementation uses Redis. Setting it to ``1`` enables caching, while any other value disables it. The following environment variables are used to configure the host and port of the Redis server:

[svekars]

Suggested change

	``TORCHINDUCTOR_REDIS_HOST`` (defaults to ``localhost``)
	* ``TORCHINDUCTOR_REDIS_HOST`` (defaults to ``localhost``)

[svekars]

Suggested change

	``TORCHINDUCTOR_REDIS_PORT`` (defaults to ``6379``)
	* ``TORCHINDUCTOR_REDIS_PORT`` (defaults to ``6379``)

[svekars]

Suggested change

	`TORCHINDUCTOR_AUTOGRAD_REMOTE_CACHE`` depends on ``TORCHINDUCTOR_FX_GRAPH_REMOTE_CACHE`` to be enabled to work. The same Redis server can store both AOTAutograd and FXGraph cache results.
	``TORCHINDUCTOR_AUTOGRAD_REMOTE_CACHE`` requires ``TORCHINDUCTOR_FX_GRAPH_REMOTE_CACHE`` to be enabled in order to function. The same Redis server can be used to store both AOTAutograd and FXGraph cache results.

[svekars]

Suggested change

	This setting enables a remote cache for ``TorchInductor``’s autotuner. As with the remote FX graph cache, the current implementation uses Redis. ``1`` enables caching, and any other value disables it. The same host / port environment variables listed above apply to this cache.
	This setting enables a remote cache for ``TorchInductor``’s autotuner. Similar to the remote FX graph cache, the current implementation uses Redis. Setting it to ``1`` enables caching, while any other value disables it. The same host and port environment variables mentioned above apply to this cache.