Skip to content

Add a cell-cache executor for batched generation #22234

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
kiymetakdemir wants to merge 3 commits into
base: add-runner
Choose a base branch
from
Draft

Add a cell-cache executor for batched generation #22234

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
3e31cf1
Add a cell-cache executor for batched generation
kiymetakdemir Aug 27, 2026
ef6400a
Address review: build the cache from the program, add a start hook
kiymetakdemir Aug 28, 2026
cb15008
Address review: build the cache from the program, read the backend fr…
kiymetakdemir Aug 29, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -997,7 +997,9 @@ if(EXECUTORCH_BUILD_EXTENSION_LLM)
add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/extension/llm/cache)
list(APPEND _executorch_extensions extension_llm_cache)
add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/extension/llm/batching)
list(APPEND _executorch_extensions extension_llm_batching)
list(APPEND _executorch_extensions extension_llm_batching
extension_llm_batching_cell
)
endif()

if(EXECUTORCH_BUILD_EXTENSION_RUNNER_UTIL)
Expand Down
20 changes: 19 additions & 1 deletion extension/llm/batching/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,10 @@
# scheduler and the executor seam are header-only and free of ExecuTorch runtime
# types; the runner owns a thread, so this is a static library rather than an
# INTERFACE target.
#
# extension_llm_batching_cell is a separate target because it implements that
# seam against a program and a KV cache, and so carries the runtime types the
# seam itself is kept clear of.

if(NOT EXECUTORCH_ROOT)
set(EXECUTORCH_ROOT ${CMAKE_CURRENT_SOURCE_DIR}/../../..)
Expand All @@ -26,8 +30,22 @@ target_compile_options(extension_llm_batching PUBLIC ${_common_compile_options})
find_package(Threads REQUIRED)
target_link_libraries(extension_llm_batching PUBLIC Threads::Threads)

add_library(extension_llm_batching_cell cell_executor.cpp)
target_link_libraries(
extension_llm_batching_cell
PUBLIC extension_llm_batching extension_llm_cache extension_module
extension_tensor
PRIVATE extension_llm_sampler
)
target_include_directories(
extension_llm_batching_cell PUBLIC ${_common_include_directories}
)
target_compile_options(
extension_llm_batching_cell PUBLIC ${_common_compile_options}
)

install(
TARGETS extension_llm_batching
TARGETS extension_llm_batching extension_llm_batching_cell
EXPORT ExecuTorchTargets
DESTINATION ${CMAKE_INSTALL_LIBDIR}
INCLUDES
Expand Down
Loading
Loading