Skip to content

GAP9: -O3 hot kernels + hoist L2->L1 tile-control tables to L2 #199

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

Open
runwangdl wants to merge 2 commits into
base: devel
Choose a base branch
from
Open

GAP9: -O3 hot kernels + hoist L2->L1 tile-control tables to L2 #199

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
57e7b82
perf(GAP9): -O3 on hot forward kernels (Conv / DWConv / Gemm)
runwangdl Jul 2, 2026
1e91ff7
fix(GAP9): hoist L2->L1 tile-control tables to L2
runwangdl Jul 2, 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
13 changes: 12 additions & 1 deletion Deeploy/TilingExtension/CodeTransformationPasses/TilingHoistingMixIn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,7 +60,18 @@ def _hoistValues(self,
else:
cb._type = PointerClass(BasicDataTypes.minimalIntegerType(values))
cb._instance = cb._type(cb.name, ctxt)
cb._memoryLevel = self.memory
# These are constant tile *control* tables (numTiles / DMA cmd / size /
# dims / offsets) read by the (cluster) controller to drive the tiling
# loop and program DMAs -- not bulk tile data. Putting them in the
# innermost tile memory (L1/TCDM) wastes scarce L1 and, on GAP9, places
# them in the contended L1 region next to the cluster master stack: a
# deep stack write can clobber a single table entry, turning a DMA `cmd`
# into a garbage code pointer so mchan_transfer_wait() hangs forever
# (observed on MobileNetV1 training). Keep them in the controller-
# addressable outer memory (L2) instead. Only redirect the L2->L1 pass;
# the L3->L2 pass keeps its tables in L2 (== self.memory), never L3.
# Platforms that don't tile into a level named "L1" are unaffected.
cb._memoryLevel = "L2" if self.memory == "L1" else self.memory
return cb

def _hoistReference(self,
Expand Down
10 changes: 10 additions & 0 deletions TargetLibraries/GAP9/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,16 @@ target_compile_options(deeploygap9 PRIVATE

target_link_libraries(deeploygap9 PUBLIC pmsis)

# Compile the hot forward kernels at -O3 (set last so it wins over the SDK's
# default -Os). Conv / depthwise-conv / Gemm dominate GAP9 inference cycles;
# -O3 turns on the RISC-V (XpulpV2) hardware loops on their tight inner loops.
set(_KERNEL_O3_FILES
${CMAKE_CURRENT_LIST_DIR}/../PULPOpen/src/Convolution_fp32.c
${CMAKE_CURRENT_LIST_DIR}/../PULPOpen/src/DWConvolution_fp32.c
${CMAKE_CURRENT_LIST_DIR}/../PULPOpen/src/Gemm.c
)
set_source_files_properties(${_KERNEL_O3_FILES} PROPERTIES COMPILE_OPTIONS "-O3")

#RW: Link PULP-NN
#RW: Set PULP-NN version and bitwidth for pulp-nn-mixed
set(PULPNNVERSION XPULPV2)
Expand Down
Loading