Skip to content

Commit 6ce1e3e

Browse files
daanzudaanzu
committed
Implement CI caching of expensive native build
1 parent 734df98 commit 6ce1e3e

File tree

4 files changed

+247
-39
lines changed

4 files changed

+247
-39
lines changed

‎.github/workflows/build.yml‎

Lines changed: 203 additions & 25 deletions
Original file line numberDiff line numberDiff line change
@@ -15,6 +15,9 @@ jobs:
1515
build-linux:
1616
runs-on: ubuntu-latest
1717
if: true
18+
env:
19+
MKL_URL: ""
20+
# MKL_URL: "https://registrationcenter-download.intel.com/akdlm/irc_nas/tec/16917/l_mkl_2020.4.304.tgz"
1821
steps:
1922
- name: Checkout repository
2023
uses: actions/checkout@v5
@@ -26,29 +29,86 @@ jobs:
2629
export TAG=$(git tag --points-at HEAD)
2730
echo "TAG: $TAG"
2831
if [[ $TAG ]]; then
32+
echo "KALDI_BRANCH: kag-$TAG"
2933
echo "KALDI_BRANCH=kag-$TAG" >> $GITHUB_ENV
3034
echo "KALDI_BRANCH=kag-$TAG" >> $GITHUB_OUTPUT
3135
else
36+
echo "KALDI_BRANCH: ${GITHUB_REF/refs\/heads\//}"
3237
echo "KALDI_BRANCH=${GITHUB_REF/refs\/heads\//}" >> $GITHUB_ENV
3338
echo "KALDI_BRANCH=${GITHUB_REF/refs\/heads\//}" >> $GITHUB_OUTPUT
3439
fi
3540
41+
- name: Get Kaldi commit hash
42+
id: get-kaldi-commit
43+
run: |
44+
KALDI_COMMIT=$(git ls-remote https://github.com/daanzu/kaldi-fork-active-grammar.git $KALDI_BRANCH | cut -f1)
45+
echo "KALDI_COMMIT: $KALDI_COMMIT"
46+
echo "KALDI_COMMIT=$KALDI_COMMIT" >> $GITHUB_OUTPUT
47+
48+
- name: Restore cached native binaries
49+
id: cache-native-binaries-restore
50+
uses: actions/cache/restore@v4
51+
with:
52+
key: native-${{ runner.os }}-${{ steps.get-kaldi-commit.outputs.KALDI_COMMIT }}-${{ env.MKL_URL }}-v1
53+
path: |
54+
kaldi_active_grammar/exec/linux
55+
kaldi_active_grammar.libs
56+
3657
- name: Setup just
3758
run: |
3859
ls -al $HOME
3960
mkdir $HOME/bin
4061
curl --proto '=https' --tlsv1.2 -sSf https://just.systems/install.sh | bash -s -- --to $HOME/bin/
4162
echo "$HOME/bin" >> $GITHUB_PATH
4263
43-
- name: Build with dockcross
64+
- name: Modify native binaries restored from cache to work with auditwheel repair
65+
if: steps.cache-native-binaries-restore.outputs.cache-hit == 'true'
66+
run: |
67+
patchelf --force-rpath --set-rpath "$(PWD)/kaldi_active_grammar.libs" kaldi_active_grammar/exec/linux/libkaldi-dragonfly.so
68+
readelf -d kaldi_active_grammar/exec/linux/libkaldi-dragonfly.so | egrep 'NEEDED|RUNPATH|RPATH'
69+
LD_DEBUG=libs lddtree kaldi_active_grammar/exec/linux/libkaldi-dragonfly.so
70+
71+
- name: Build with dockcross (native binaries & python wheel)
4472
run: |
4573
echo "KALDI_BRANCH: $KALDI_BRANCH"
4674
echo "MKL_URL: $MKL_URL"
47-
# Example MKL_URL: https://registrationcenter-download.intel.com/akdlm/irc_nas/tec/16917/l_mkl_2020.4.304.tgz
48-
just build-dockcross $KALDI_BRANCH $MKL_URL
49-
# cp dist/* wheelhouse/
75+
just build-dockcross ${{ steps.cache-native-binaries-restore.outputs.cache-hit == 'true' && '--skip-native' || '' }} $KALDI_BRANCH $MKL_URL
5076
ls -al wheelhouse/
5177
78+
- name: Extract native binaries from wheel after auditwheel repair, to save to cache
79+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
80+
run: |
81+
# Assert there is only one wheel
82+
WHEEL_COUNT=$(ls wheelhouse/*.whl | wc -l)
83+
if [ "$WHEEL_COUNT" -ne 1 ]; then
84+
echo "Error: Expected exactly 1 wheel, found $WHEEL_COUNT"
85+
ls -al wheelhouse/
86+
exit 1
87+
fi
88+
WHEEL_FILE=$(ls wheelhouse/*.whl)
89+
echo "Extracting from wheel: $WHEEL_FILE"
90+
unzip -j -o $WHEEL_FILE 'kaldi_active_grammar/exec/linux/*' -d kaldi_active_grammar/exec/linux/
91+
unzip -o $WHEEL_FILE 'kaldi_active_grammar.libs/*'
92+
ls -al kaldi_active_grammar/exec/linux/
93+
readelf -d kaldi_active_grammar/exec/linux/libkaldi-dragonfly.so | egrep 'NEEDED|RUNPATH|RPATH'
94+
95+
- name: Save cached native binaries
96+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
97+
uses: actions/cache/save@v4
98+
with:
99+
key: ${{ steps.cache-native-binaries-restore.outputs.cache-primary-key }}
100+
path: |
101+
kaldi_active_grammar/exec/linux
102+
kaldi_active_grammar.libs
103+
104+
- name: Upload native binaries to artifacts
105+
uses: actions/upload-artifact@v4
106+
with:
107+
name: native-linux
108+
path: |
109+
kaldi_active_grammar/exec/linux
110+
kaldi_active_grammar.libs
111+
52112
- name: Upload Linux wheels
53113
uses: actions/upload-artifact@v4
54114
with:
@@ -74,33 +134,54 @@ jobs:
74134
path: main
75135

76136
- name: Get KALDI_BRANCH (kag-$TAG tag if commit is tagged; current branch name if not)
77-
id: get_kaldi_branch
137+
id: get-kaldi-branch
138+
working-directory: main
78139
run: |
79140
# Fetch tags on the one fetched commit (shallow clone)
80141
git fetch --depth=1 origin "+refs/tags/*:refs/tags/*"
81142
export TAG=$(git tag --points-at HEAD)
143+
echo "TAG: $TAG"
82144
if [[ $TAG ]]; then
145+
echo "KALDI_BRANCH: kag-$TAG"
83146
echo "KALDI_BRANCH=kag-$TAG" >> $GITHUB_ENV
84147
echo "KALDI_BRANCH=kag-$TAG" >> $GITHUB_OUTPUT
85148
else
149+
echo "KALDI_BRANCH: ${GITHUB_REF/refs\/heads\//}"
86150
echo "KALDI_BRANCH=${GITHUB_REF/refs\/heads\//}" >> $GITHUB_ENV
87151
echo "KALDI_BRANCH=${GITHUB_REF/refs\/heads\//}" >> $GITHUB_OUTPUT
88152
fi
89153
154+
- name: Get Kaldi commit hash
155+
id: get-kaldi-commit
156+
run: |
157+
KALDI_COMMIT=$(git ls-remote https://github.com/daanzu/kaldi-fork-active-grammar.git $KALDI_BRANCH | cut -f1)
158+
echo "KALDI_COMMIT: $KALDI_COMMIT"
159+
echo "KALDI_COMMIT=$KALDI_COMMIT" >> $GITHUB_OUTPUT
160+
161+
- name: Restore cached native binaries
162+
id: cache-native-binaries-restore
163+
uses: actions/cache/restore@v4
164+
with:
165+
key: native-${{ runner.os }}-${{ steps.get-kaldi-commit.outputs.KALDI_COMMIT }}-${{ env.VS_VERSION }}-${{ env.PLATFORM_TOOLSET }}-${{ env.WINDOWS_TARGET_PLATFORM_VERSION }}-${{ env.MKL_VERSION }}-v1
166+
path: main/kaldi_active_grammar/exec/windows
167+
90168
- name: Checkout OpenFST repository
169+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
91170
uses: actions/checkout@v5
92171
with:
93172
repository: daanzu/openfst
94173
path: openfst
95174

96175
- name: Checkout Kaldi repository
176+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
97177
uses: actions/checkout@v5
98178
with:
99179
repository: daanzu/kaldi-fork-active-grammar
100180
path: kaldi
101-
ref: ${{ steps.get_kaldi_branch.outputs.KALDI_BRANCH }}
181+
ref: ${{ steps.get-kaldi-branch.outputs.KALDI_BRANCH }}
102182

103183
- name: Gather system information
184+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
104185
run: |
105186
echo $GITHUB_WORKSPACE
106187
df -h
@@ -118,6 +199,7 @@ jobs:
118199
vswhere -find 'VC\Redist\**\VC_redist.x64.exe'
119200
120201
- name: Setup Kaldi build configuration
202+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
121203
run: |
122204
cd kaldi/windows
123205
cp kaldiwin_mkl.props kaldiwin.props
@@ -136,28 +218,49 @@ jobs:
136218
perl get_version.pl
137219
138220
- name: Add msbuild to PATH
221+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
139222
uses: microsoft/setup-msbuild@v2
140223

141224
- name: Install MKL
225+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
142226
run: winget install --id=Intel.oneMKL -v "${MKL_VERSION}" -e --accept-package-agreements --accept-source-agreements --disable-interactivity
143227

144228
- name: Build OpenFST
229+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
145230
shell: cmd
146231
run: msbuild -t:Build -p:Configuration=Release -p:Platform=x64 -p:PlatformToolset=%PLATFORM_TOOLSET% -maxCpuCount -verbosity:minimal openfst/openfst.sln
147232

148233
- name: Build Kaldi
234+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
149235
shell: cmd
150236
run: msbuild -t:Build -p:Configuration=Release -p:Platform=x64 -p:PlatformToolset=%PLATFORM_TOOLSET% -p:WindowsTargetPlatformVersion=%WINDOWS_TARGET_PLATFORM_VERSION% -maxCpuCount -verbosity:minimal kaldi/kaldiwin_%VS_VERSION%_MKL/kaldiwin/kaldi-dragonfly/kaldi-dragonfly.vcxproj
151237

238+
- name: Copy native binaries
239+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
240+
run: |
241+
mkdir -p main/kaldi_active_grammar/exec/windows
242+
cp kaldi/kaldiwin_${VS_VERSION}_MKL/kaldiwin/kaldi-dragonfly/x64/Release/kaldi-dragonfly.dll main/kaldi_active_grammar/exec/windows/
243+
244+
- name: Save cached native binaries
245+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
246+
uses: actions/cache/save@v4
247+
with:
248+
key: ${{ steps.cache-native-binaries-restore.outputs.cache-primary-key }}
249+
path: main/kaldi_active_grammar/exec/windows
250+
251+
- name: Upload native binaries to artifacts
252+
uses: actions/upload-artifact@v4
253+
with:
254+
name: native-windows
255+
path: main/kaldi_active_grammar/exec/windows
256+
152257
- name: Build Python wheel
258+
working-directory: main
153259
run: |
154-
cd main
155260
python -m pip -V
156261
python -m pip install --upgrade setuptools wheel scikit-build cmake ninja
157262
# ls -alR ../
158-
mkdir -p kaldi_active_grammar/exec/windows
159-
cp ../kaldi/kaldiwin_${VS_VERSION}_MKL/kaldiwin/kaldi-dragonfly/x64/Release/kaldi-dragonfly.dll kaldi_active_grammar/exec/windows/
160-
env KALDIAG_SETUP_RAW=1 python setup.py bdist_wheel
263+
env KALDIAG_BUILD_SKIP_NATIVE=1 python setup.py bdist_wheel
161264
ls -al dist/
162265
163266
- name: Upload Windows wheels
@@ -188,27 +291,38 @@ jobs:
188291
uses: actions/checkout@v5
189292

190293
- name: Get KALDI_BRANCH (kag-$TAG tag if commit is tagged; current branch name if not)
294+
id: get-kaldi-branch
191295
run: |
192296
# Fetch tags on the one fetched commit (shallow clone)
193297
git fetch --depth=1 origin "+refs/tags/*:refs/tags/*"
194298
export TAG=$(git tag --points-at HEAD)
299+
echo "TAG: $TAG"
195300
if [[ $TAG ]]; then
301+
echo "KALDI_BRANCH: kag-$TAG"
196302
echo "KALDI_BRANCH=kag-$TAG" >> $GITHUB_ENV
197303
echo "KALDI_BRANCH=kag-$TAG" >> $GITHUB_OUTPUT
198304
else
305+
echo "KALDI_BRANCH: ${GITHUB_REF/refs\/heads\//}"
199306
echo "KALDI_BRANCH=${GITHUB_REF/refs\/heads\//}" >> $GITHUB_ENV
200307
echo "KALDI_BRANCH=${GITHUB_REF/refs\/heads\//}" >> $GITHUB_OUTPUT
201308
fi
202309
203-
- name: Install dependencies
310+
- name: Get Kaldi commit hash
311+
id: get-kaldi-commit
204312
run: |
205-
python3 -m pip install --break-system-packages --user --upgrade setuptools wheel scikit-build cmake ninja
206-
brew install automake sox libtool
207-
brew reinstall gfortran # For openblas
208-
# brew install autoconf
313+
KALDI_COMMIT=$(git ls-remote https://github.com/daanzu/kaldi-fork-active-grammar.git $KALDI_BRANCH | cut -f1)
314+
echo "KALDI_COMMIT: $KALDI_COMMIT"
315+
echo "KALDI_COMMIT=$KALDI_COMMIT" >> $GITHUB_OUTPUT
316+
317+
- name: Restore cached native binaries
318+
id: cache-native-binaries-restore
319+
uses: actions/cache/restore@v4
320+
with:
321+
key: native-${{ runner.os }}-arm-${{ steps.get-kaldi-commit.outputs.KALDI_COMMIT }}-${{ env.MACOSX_DEPLOYMENT_TARGET }}-${{ env.MKL_URL }}-v1
322+
path: kaldi_active_grammar/exec/macos
209323

210324
- name: Install MKL (if enabled)
211-
if: ${{ env.MKL_URL != '' }}
325+
if: ${{ env.MKL_URL != '' && steps.cache-native-binaries-restore.outputs.cache-hit != 'true' }}
212326
run: |
213327
echo "Installing MKL from: $MKL_URL"
214328
export MKL_FILE=${MKL_URL##*/}
@@ -219,12 +333,38 @@ jobs:
219333
sed -i.bak -e 's/decline/accept/g' silent.cfg
220334
sudo /Volumes/${MKL_FILE}/${MKL_FILE}.app/Contents/MacOS/install.sh --silent silent.cfg
221335
336+
- name: Install dependencies for native build
337+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
338+
run: |
339+
python3 -m pip install --break-system-packages --user --upgrade scikit-build cmake ninja
340+
brew install automake sox libtool
341+
brew reinstall gfortran # For openblas
342+
# brew install autoconf
343+
344+
- name: Install dependencies for python build
345+
run: |
346+
python3 -m pip install --break-system-packages --user --upgrade setuptools wheel
347+
222348
- name: Build Python wheel
223349
run: |
224350
echo "KALDI_BRANCH: $KALDI_BRANCH"
225-
python3 setup.py bdist_wheel
351+
echo "MKL_URL: $MKL_URL"
352+
${{ steps.cache-native-binaries-restore.outputs.cache-hit == 'true' && 'KALDIAG_BUILD_SKIP_NATIVE=1' || '' }} python3 setup.py bdist_wheel
226353
ls -al dist/
227354
355+
- name: Save cached native binaries
356+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
357+
uses: actions/cache/save@v4
358+
with:
359+
key: ${{ steps.cache-native-binaries-restore.outputs.cache-primary-key }}
360+
path: kaldi_active_grammar/exec/macos
361+
362+
- name: Upload native binaries to artifacts
363+
uses: actions/upload-artifact@v4
364+
with:
365+
name: native-macos-arm
366+
path: kaldi_active_grammar/exec/macos
367+
228368
- name: Upload MacOS ARM wheels
229369
uses: actions/upload-artifact@v4
230370
with:
@@ -243,27 +383,38 @@ jobs:
243383
uses: actions/checkout@v5
244384

245385
- name: Get KALDI_BRANCH (kag-$TAG tag if commit is tagged; current branch name if not)
386+
id: get-kaldi-branch
246387
run: |
247388
# Fetch tags on the one fetched commit (shallow clone)
248389
git fetch --depth=1 origin "+refs/tags/*:refs/tags/*"
249390
export TAG=$(git tag --points-at HEAD)
391+
echo "TAG: $TAG"
250392
if [[ $TAG ]]; then
393+
echo "KALDI_BRANCH: kag-$TAG"
251394
echo "KALDI_BRANCH=kag-$TAG" >> $GITHUB_ENV
252395
echo "KALDI_BRANCH=kag-$TAG" >> $GITHUB_OUTPUT
253396
else
397+
echo "KALDI_BRANCH: ${GITHUB_REF/refs\/heads\//}"
254398
echo "KALDI_BRANCH=${GITHUB_REF/refs\/heads\//}" >> $GITHUB_ENV
255399
echo "KALDI_BRANCH=${GITHUB_REF/refs\/heads\//}" >> $GITHUB_OUTPUT
256400
fi
257401
258-
- name: Install dependencies
402+
- name: Get Kaldi commit hash
403+
id: get-kaldi-commit
259404
run: |
260-
python3 -m pip install --break-system-packages --user --upgrade setuptools wheel scikit-build cmake ninja
261-
brew install automake sox
262-
brew reinstall gfortran # For openblas
263-
# brew install autoconf libtool
405+
KALDI_COMMIT=$(git ls-remote https://github.com/daanzu/kaldi-fork-active-grammar.git $KALDI_BRANCH | cut -f1)
406+
echo "KALDI_COMMIT: $KALDI_COMMIT"
407+
echo "KALDI_COMMIT=$KALDI_COMMIT" >> $GITHUB_OUTPUT
408+
409+
- name: Restore cached native binaries
410+
id: cache-native-binaries-restore
411+
uses: actions/cache/restore@v4
412+
with:
413+
key: native-${{ runner.os }}-intel-${{ steps.get-kaldi-commit.outputs.KALDI_COMMIT }}-${{ env.MACOSX_DEPLOYMENT_TARGET }}-${{ env.MKL_URL }}-v1
414+
path: kaldi_active_grammar/exec/macos
264415

265416
- name: Install MKL (if enabled)
266-
if: ${{ env.MKL_URL != '' }}
417+
if: ${{ env.MKL_URL != '' && steps.cache-native-binaries-restore.outputs.cache-hit != 'true' }}
267418
run: |
268419
echo "Installing MKL from: $MKL_URL"
269420
export MKL_FILE=${MKL_URL##*/}
@@ -274,16 +425,43 @@ jobs:
274425
sed -i.bak -e 's/decline/accept/g' silent.cfg
275426
sudo /Volumes/${MKL_FILE}/${MKL_FILE}.app/Contents/MacOS/install.sh --silent silent.cfg
276427
428+
- name: Install dependencies for native build
429+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
430+
run: |
431+
python3 -m pip install --break-system-packages --user --upgrade scikit-build cmake ninja
432+
brew install automake sox
433+
brew reinstall gfortran # For openblas
434+
# brew install autoconf libtool
435+
436+
- name: Install dependencies for python build
437+
run: |
438+
python3 -m pip install --break-system-packages --user --upgrade setuptools wheel
439+
277440
- name: Build Python wheel
278441
run: |
279-
python3 setup.py bdist_wheel
442+
echo "KALDI_BRANCH: $KALDI_BRANCH"
443+
echo "MKL_URL: $MKL_URL"
444+
${{ steps.cache-native-binaries-restore.outputs.cache-hit == 'true' && 'KALDIAG_BUILD_SKIP_NATIVE=1' || '' }} python3 setup.py bdist_wheel
280445
ls -al dist/
281446
447+
- name: Save cached native binaries
448+
if: steps.cache-native-binaries-restore.outputs.cache-hit != 'true'
449+
uses: actions/cache/save@v4
450+
with:
451+
key: ${{ steps.cache-native-binaries-restore.outputs.cache-primary-key }}
452+
path: kaldi_active_grammar/exec/macos
453+
454+
- name: Upload native binaries to artifacts
455+
uses: actions/upload-artifact@v4
456+
with:
457+
name: native-macos-intel
458+
path: kaldi_active_grammar/exec/macos
459+
282460
- name: Upload MacOS Intel wheels
283461
uses: actions/upload-artifact@v4
284462
with:
285463
name: wheels-macos-intel
286-
path: dist/*
464+
path: dist/*.whl
287465

288466
merge-wheels:
289467
runs-on: ubuntu-latest

0 commit comments

Comments
 (0)