InvokeAI/tests/nodes/test_invoker.py

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

135 lines
4.7 KiB
Python
Raw Normal View History

feat(nodes): add invocation cache The invocation cache provides simple node memoization functionality. Nodes that use the cache are memoized and not re-executed if their inputs haven't changed. Instead, the stored output is returned. ## Results This feature provides anywhere some significant to massive performance improvement. The improvement is most marked on large batches of generations where you only change a couple things (e.g. different seed or prompt for each iteration) and low-VRAM systems, where skipping an extraneous model load is a big deal. ## Overview A new `invocation_cache` service is added to handle the caching. There's not much to it. All nodes now inherit a boolean `use_cache` field from `BaseInvocation`. This is a node field and not a class attribute, because specific instances of nodes may want to opt in or out of caching. The recently-added `invoke_internal()` method on `BaseInvocation` is used as an entrypoint for the cache logic. To create a cache key, the invocation is first serialized using pydantic's provided `json()` method, skipping the unique `id` field. Then python's very fast builtin `hash()` is used to create an integer key. All implementations of `InvocationCacheBase` must provide a class method `create_key()` which accepts an invocation and outputs a string or integer key. ## In-Memory Implementation An in-memory implementation is provided. In this implementation, the node outputs are stored in memory as python classes. The in-memory cache does not persist application restarts. Max node cache size is added as `node_cache_size` under the `Generation` config category. It defaults to 512 - this number is up for discussion, but given that these are relatively lightweight pydantic models, I think it's safe to up this even higher. Note that the cache isn't storing the big stuff - tensors and images are store on disk, and outputs include only references to them. ## Node Definition The default for all nodes is to use the cache. The `@invocation` decorator now accepts an optional `use_cache: bool` argument to override the default of `True`. Non-deterministic nodes, however, should set this to `False`. Currently, all random-stuff nodes, including `dynamic_prompt`, are set to `False`. The field name `use_cache` is now effectively a reserved field name and possibly a breaking change if any community nodes use this as a field name. In hindsight, all our reserved field names should have been prefixed with underscores or something. ## One Gotcha Leaf nodes probably want to opt out of the cache, because if they are not cached, their outputs are not saved again. If you run the same graph multiple times, you only end up with a single image output, because the image storage side-effects are in the `invoke()` method, which is bypassed if we have a cache hit. ## Linear UI The linear graphs _almost_ just work, but due to the gotcha, we need to be careful about the final image-outputting node. To resolve this, a `SaveImageInvocation` node is added and used in the linear graphs. This node is similar to `ImagePrimitive`, except it saves a copy of its input image, and has `use_cache` set to `False` by default. This is now the leaf node in all linear graphs, and is the only node in those graphs with `use_cache == False` _and_ the only node with `is_intermedate == False`. ## Workflow Editor All nodes now have a footer with a new `Use Cache [ ]` checkbox. It defaults to the value set by the invocation in its python definition, but can be changed by the user. The workflow/node validation logic has been updated to migrate old workflows to use the new default values for `use_cache`. Users may still want to review the settings that have been chosen. In the event of catastrophic failure when running this migration, the default value of `True` is applied, as this is correct for most nodes. Users should consider saving their workflows after loading them in and having them updated. ## Future Enhancements - Callback A future enhancement would be to provide a callback to the `use_cache` flag that would be run as the node is executed to determine, based on its own internal state, if the cache should be used or not. This would be useful for `DynamicPromptInvocation`, where the deterministic behaviour is determined by the `combinatorial: bool` field. ## Future Enhancements - Persisted Cache Similar to how the latents storage is backed by disk, the invocation cache could be persisted to the database or disk. We'd need to be very careful about deserializing outputs, but it's perhaps worth exploring in the future.
2023-09-18 03:41:19 +00:00
import logging
2023-08-18 14:57:18 +00:00
import pytest
from invokeai.app.services.graph import Graph, GraphExecutionState, LibraryGraph
feat(nodes): add invocation cache The invocation cache provides simple node memoization functionality. Nodes that use the cache are memoized and not re-executed if their inputs haven't changed. Instead, the stored output is returned. ## Results This feature provides anywhere some significant to massive performance improvement. The improvement is most marked on large batches of generations where you only change a couple things (e.g. different seed or prompt for each iteration) and low-VRAM systems, where skipping an extraneous model load is a big deal. ## Overview A new `invocation_cache` service is added to handle the caching. There's not much to it. All nodes now inherit a boolean `use_cache` field from `BaseInvocation`. This is a node field and not a class attribute, because specific instances of nodes may want to opt in or out of caching. The recently-added `invoke_internal()` method on `BaseInvocation` is used as an entrypoint for the cache logic. To create a cache key, the invocation is first serialized using pydantic's provided `json()` method, skipping the unique `id` field. Then python's very fast builtin `hash()` is used to create an integer key. All implementations of `InvocationCacheBase` must provide a class method `create_key()` which accepts an invocation and outputs a string or integer key. ## In-Memory Implementation An in-memory implementation is provided. In this implementation, the node outputs are stored in memory as python classes. The in-memory cache does not persist application restarts. Max node cache size is added as `node_cache_size` under the `Generation` config category. It defaults to 512 - this number is up for discussion, but given that these are relatively lightweight pydantic models, I think it's safe to up this even higher. Note that the cache isn't storing the big stuff - tensors and images are store on disk, and outputs include only references to them. ## Node Definition The default for all nodes is to use the cache. The `@invocation` decorator now accepts an optional `use_cache: bool` argument to override the default of `True`. Non-deterministic nodes, however, should set this to `False`. Currently, all random-stuff nodes, including `dynamic_prompt`, are set to `False`. The field name `use_cache` is now effectively a reserved field name and possibly a breaking change if any community nodes use this as a field name. In hindsight, all our reserved field names should have been prefixed with underscores or something. ## One Gotcha Leaf nodes probably want to opt out of the cache, because if they are not cached, their outputs are not saved again. If you run the same graph multiple times, you only end up with a single image output, because the image storage side-effects are in the `invoke()` method, which is bypassed if we have a cache hit. ## Linear UI The linear graphs _almost_ just work, but due to the gotcha, we need to be careful about the final image-outputting node. To resolve this, a `SaveImageInvocation` node is added and used in the linear graphs. This node is similar to `ImagePrimitive`, except it saves a copy of its input image, and has `use_cache` set to `False` by default. This is now the leaf node in all linear graphs, and is the only node in those graphs with `use_cache == False` _and_ the only node with `is_intermedate == False`. ## Workflow Editor All nodes now have a footer with a new `Use Cache [ ]` checkbox. It defaults to the value set by the invocation in its python definition, but can be changed by the user. The workflow/node validation logic has been updated to migrate old workflows to use the new default values for `use_cache`. Users may still want to review the settings that have been chosen. In the event of catastrophic failure when running this migration, the default value of `True` is applied, as this is correct for most nodes. Users should consider saving their workflows after loading them in and having them updated. ## Future Enhancements - Callback A future enhancement would be to provide a callback to the `use_cache` flag that would be run as the node is executed to determine, based on its own internal state, if the cache should be used or not. This would be useful for `DynamicPromptInvocation`, where the deterministic behaviour is determined by the `combinatorial: bool` field. ## Future Enhancements - Persisted Cache Similar to how the latents storage is backed by disk, the invocation cache could be persisted to the database or disk. We'd need to be very careful about deserializing outputs, but it's perhaps worth exploring in the future.
2023-09-18 03:41:19 +00:00
from invokeai.app.services.invocation_cache.invocation_cache_memory import MemoryInvocationCache
2023-08-18 14:57:18 +00:00
from invokeai.app.services.invocation_queue import MemoryInvocationQueue
from invokeai.app.services.invocation_services import InvocationServices
from invokeai.app.services.invocation_stats import InvocationStatsService
from invokeai.app.services.invoker import Invoker
from invokeai.app.services.processor import DefaultInvocationProcessor
from invokeai.app.services.sqlite import SqliteItemStorage, sqlite_memory
2023-06-29 06:01:17 +00:00
from .test_nodes import (
ErrorInvocation,
PromptTestInvocation,
2023-08-18 14:57:18 +00:00
TestEventService,
TextToImageTestInvocation,
2023-06-29 06:01:17 +00:00
create_edge,
wait_until,
)
@pytest.fixture
def simple_graph():
g = Graph()
2023-06-29 06:01:17 +00:00
g.add_node(PromptTestInvocation(id="1", prompt="Banana sushi"))
g.add_node(TextToImageTestInvocation(id="2"))
g.add_edge(create_edge("1", "prompt", "2", "prompt"))
return g
2023-06-29 06:01:17 +00:00
# This must be defined here to avoid issues with the dynamic creation of the union of all invocation types
# Defining it in a separate module will cause the union to be incomplete, and pydantic will not validate
# the test invocations.
@pytest.fixture
def mock_services() -> InvocationServices:
# NOTE: none of these are actually called by the test invocations
2023-08-02 22:31:10 +00:00
graph_execution_manager = SqliteItemStorage[GraphExecutionState](
filename=sqlite_memory, table_name="graph_executions"
)
2023-06-29 06:01:17 +00:00
return InvocationServices(
model_manager=None, # type: ignore
events=TestEventService(),
feat(nodes): add invocation cache The invocation cache provides simple node memoization functionality. Nodes that use the cache are memoized and not re-executed if their inputs haven't changed. Instead, the stored output is returned. ## Results This feature provides anywhere some significant to massive performance improvement. The improvement is most marked on large batches of generations where you only change a couple things (e.g. different seed or prompt for each iteration) and low-VRAM systems, where skipping an extraneous model load is a big deal. ## Overview A new `invocation_cache` service is added to handle the caching. There's not much to it. All nodes now inherit a boolean `use_cache` field from `BaseInvocation`. This is a node field and not a class attribute, because specific instances of nodes may want to opt in or out of caching. The recently-added `invoke_internal()` method on `BaseInvocation` is used as an entrypoint for the cache logic. To create a cache key, the invocation is first serialized using pydantic's provided `json()` method, skipping the unique `id` field. Then python's very fast builtin `hash()` is used to create an integer key. All implementations of `InvocationCacheBase` must provide a class method `create_key()` which accepts an invocation and outputs a string or integer key. ## In-Memory Implementation An in-memory implementation is provided. In this implementation, the node outputs are stored in memory as python classes. The in-memory cache does not persist application restarts. Max node cache size is added as `node_cache_size` under the `Generation` config category. It defaults to 512 - this number is up for discussion, but given that these are relatively lightweight pydantic models, I think it's safe to up this even higher. Note that the cache isn't storing the big stuff - tensors and images are store on disk, and outputs include only references to them. ## Node Definition The default for all nodes is to use the cache. The `@invocation` decorator now accepts an optional `use_cache: bool` argument to override the default of `True`. Non-deterministic nodes, however, should set this to `False`. Currently, all random-stuff nodes, including `dynamic_prompt`, are set to `False`. The field name `use_cache` is now effectively a reserved field name and possibly a breaking change if any community nodes use this as a field name. In hindsight, all our reserved field names should have been prefixed with underscores or something. ## One Gotcha Leaf nodes probably want to opt out of the cache, because if they are not cached, their outputs are not saved again. If you run the same graph multiple times, you only end up with a single image output, because the image storage side-effects are in the `invoke()` method, which is bypassed if we have a cache hit. ## Linear UI The linear graphs _almost_ just work, but due to the gotcha, we need to be careful about the final image-outputting node. To resolve this, a `SaveImageInvocation` node is added and used in the linear graphs. This node is similar to `ImagePrimitive`, except it saves a copy of its input image, and has `use_cache` set to `False` by default. This is now the leaf node in all linear graphs, and is the only node in those graphs with `use_cache == False` _and_ the only node with `is_intermedate == False`. ## Workflow Editor All nodes now have a footer with a new `Use Cache [ ]` checkbox. It defaults to the value set by the invocation in its python definition, but can be changed by the user. The workflow/node validation logic has been updated to migrate old workflows to use the new default values for `use_cache`. Users may still want to review the settings that have been chosen. In the event of catastrophic failure when running this migration, the default value of `True` is applied, as this is correct for most nodes. Users should consider saving their workflows after loading them in and having them updated. ## Future Enhancements - Callback A future enhancement would be to provide a callback to the `use_cache` flag that would be run as the node is executed to determine, based on its own internal state, if the cache should be used or not. This would be useful for `DynamicPromptInvocation`, where the deterministic behaviour is determined by the `combinatorial: bool` field. ## Future Enhancements - Persisted Cache Similar to how the latents storage is backed by disk, the invocation cache could be persisted to the database or disk. We'd need to be very careful about deserializing outputs, but it's perhaps worth exploring in the future.
2023-09-18 03:41:19 +00:00
logger=logging, # type: ignore
2023-06-29 06:01:17 +00:00
images=None, # type: ignore
latents=None, # type: ignore
boards=None, # type: ignore
board_images=None, # type: ignore
queue=MemoryInvocationQueue(),
graph_library=SqliteItemStorage[LibraryGraph](filename=sqlite_memory, table_name="graphs"),
2023-08-02 22:31:10 +00:00
graph_execution_manager=graph_execution_manager,
2023-06-29 06:01:17 +00:00
processor=DefaultInvocationProcessor(),
2023-08-02 22:31:10 +00:00
performance_statistics=InvocationStatsService(graph_execution_manager),
2023-06-29 06:01:17 +00:00
configuration=None, # type: ignore
feat(nodes): add invocation cache The invocation cache provides simple node memoization functionality. Nodes that use the cache are memoized and not re-executed if their inputs haven't changed. Instead, the stored output is returned. ## Results This feature provides anywhere some significant to massive performance improvement. The improvement is most marked on large batches of generations where you only change a couple things (e.g. different seed or prompt for each iteration) and low-VRAM systems, where skipping an extraneous model load is a big deal. ## Overview A new `invocation_cache` service is added to handle the caching. There's not much to it. All nodes now inherit a boolean `use_cache` field from `BaseInvocation`. This is a node field and not a class attribute, because specific instances of nodes may want to opt in or out of caching. The recently-added `invoke_internal()` method on `BaseInvocation` is used as an entrypoint for the cache logic. To create a cache key, the invocation is first serialized using pydantic's provided `json()` method, skipping the unique `id` field. Then python's very fast builtin `hash()` is used to create an integer key. All implementations of `InvocationCacheBase` must provide a class method `create_key()` which accepts an invocation and outputs a string or integer key. ## In-Memory Implementation An in-memory implementation is provided. In this implementation, the node outputs are stored in memory as python classes. The in-memory cache does not persist application restarts. Max node cache size is added as `node_cache_size` under the `Generation` config category. It defaults to 512 - this number is up for discussion, but given that these are relatively lightweight pydantic models, I think it's safe to up this even higher. Note that the cache isn't storing the big stuff - tensors and images are store on disk, and outputs include only references to them. ## Node Definition The default for all nodes is to use the cache. The `@invocation` decorator now accepts an optional `use_cache: bool` argument to override the default of `True`. Non-deterministic nodes, however, should set this to `False`. Currently, all random-stuff nodes, including `dynamic_prompt`, are set to `False`. The field name `use_cache` is now effectively a reserved field name and possibly a breaking change if any community nodes use this as a field name. In hindsight, all our reserved field names should have been prefixed with underscores or something. ## One Gotcha Leaf nodes probably want to opt out of the cache, because if they are not cached, their outputs are not saved again. If you run the same graph multiple times, you only end up with a single image output, because the image storage side-effects are in the `invoke()` method, which is bypassed if we have a cache hit. ## Linear UI The linear graphs _almost_ just work, but due to the gotcha, we need to be careful about the final image-outputting node. To resolve this, a `SaveImageInvocation` node is added and used in the linear graphs. This node is similar to `ImagePrimitive`, except it saves a copy of its input image, and has `use_cache` set to `False` by default. This is now the leaf node in all linear graphs, and is the only node in those graphs with `use_cache == False` _and_ the only node with `is_intermedate == False`. ## Workflow Editor All nodes now have a footer with a new `Use Cache [ ]` checkbox. It defaults to the value set by the invocation in its python definition, but can be changed by the user. The workflow/node validation logic has been updated to migrate old workflows to use the new default values for `use_cache`. Users may still want to review the settings that have been chosen. In the event of catastrophic failure when running this migration, the default value of `True` is applied, as this is correct for most nodes. Users should consider saving their workflows after loading them in and having them updated. ## Future Enhancements - Callback A future enhancement would be to provide a callback to the `use_cache` flag that would be run as the node is executed to determine, based on its own internal state, if the cache should be used or not. This would be useful for `DynamicPromptInvocation`, where the deterministic behaviour is determined by the `combinatorial: bool` field. ## Future Enhancements - Persisted Cache Similar to how the latents storage is backed by disk, the invocation cache could be persisted to the database or disk. We'd need to be very careful about deserializing outputs, but it's perhaps worth exploring in the future.
2023-09-18 03:41:19 +00:00
invocation_cache=MemoryInvocationCache(),
2023-06-29 06:01:17 +00:00
)
@pytest.fixture()
def mock_invoker(mock_services: InvocationServices) -> Invoker:
2023-06-29 06:01:17 +00:00
return Invoker(services=mock_services)
def test_can_create_graph_state(mock_invoker: Invoker):
g = mock_invoker.create_execution_state()
mock_invoker.stop()
assert g is not None
assert isinstance(g, GraphExecutionState)
2023-06-29 06:01:17 +00:00
def test_can_create_graph_state_from_graph(mock_invoker: Invoker, simple_graph):
2023-06-29 06:01:17 +00:00
g = mock_invoker.create_execution_state(graph=simple_graph)
mock_invoker.stop()
assert g is not None
assert isinstance(g, GraphExecutionState)
assert g.graph == simple_graph
2023-06-29 06:01:17 +00:00
# @pytest.mark.xfail(reason = "Requires fixing following the model manager refactor")
def test_can_invoke(mock_invoker: Invoker, simple_graph):
2023-06-29 06:01:17 +00:00
g = mock_invoker.create_execution_state(graph=simple_graph)
invocation_id = mock_invoker.invoke(g)
assert invocation_id is not None
def has_executed_any(g: GraphExecutionState):
g = mock_invoker.services.graph_execution_manager.get(g.id)
return len(g.executed) > 0
2023-06-29 06:01:17 +00:00
wait_until(lambda: has_executed_any(g), timeout=5, interval=1)
mock_invoker.stop()
g = mock_invoker.services.graph_execution_manager.get(g.id)
assert len(g.executed) > 0
2023-06-29 06:01:17 +00:00
# @pytest.mark.xfail(reason = "Requires fixing following the model manager refactor")
def test_can_invoke_all(mock_invoker: Invoker, simple_graph):
2023-06-29 06:01:17 +00:00
g = mock_invoker.create_execution_state(graph=simple_graph)
invocation_id = mock_invoker.invoke(g, invoke_all=True)
assert invocation_id is not None
def has_executed_all(g: GraphExecutionState):
g = mock_invoker.services.graph_execution_manager.get(g.id)
return g.is_complete()
2023-06-29 06:01:17 +00:00
wait_until(lambda: has_executed_all(g), timeout=5, interval=1)
mock_invoker.stop()
g = mock_invoker.services.graph_execution_manager.get(g.id)
assert g.is_complete()
2023-06-29 06:01:17 +00:00
# @pytest.mark.xfail(reason = "Requires fixing following the model manager refactor")
def test_handles_errors(mock_invoker: Invoker):
g = mock_invoker.create_execution_state()
2023-06-29 06:01:17 +00:00
g.graph.add_node(ErrorInvocation(id="1"))
mock_invoker.invoke(g, invoke_all=True)
def has_executed_all(g: GraphExecutionState):
g = mock_invoker.services.graph_execution_manager.get(g.id)
return g.is_complete()
2023-06-29 06:01:17 +00:00
wait_until(lambda: has_executed_all(g), timeout=5, interval=1)
mock_invoker.stop()
g = mock_invoker.services.graph_execution_manager.get(g.id)
assert g.has_error()
assert g.is_complete()
2023-06-29 06:01:17 +00:00
assert all((i in g.errors for i in g.source_prepared_mapping["1"]))