Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
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
20 changes: 20 additions & 0 deletions .github/workflows/ci.yml
Original file line number Diff line number Diff line change
Expand Up @@ -107,6 +107,15 @@ jobs:
run: |
python3.12 -m pytest tests/test_pr37_regression.py -v --tb=short

- name: Run topic29 SIMD-vectorize regressions
run: |
python3.12 -m pytest \
tests/test_vectorize.py \
tests/test_vector_lowering.py \
tests/test_vector_encoder.py \
tests/test_topic29_vectorize_case_report.py \
-v --tb=short

- name: Generate test visualization page
if: github.ref == 'refs/heads/main'
run: |
Expand Down Expand Up @@ -218,6 +227,14 @@ jobs:
--json benchmark_reports/const_merge_report.json \
--markdown benchmark_reports/const_merge_report.md

# ── 3.1.2 课题29:SIMD 向量化 case 报告(A/B + RV32 执行等价) ─────
- name: Topic 29 SIMD vectorize case report
run: |
mkdir -p benchmark_reports
python3.12 benchmarks/run_topic29_vectorize_case.py \
--json benchmark_reports/vectorize_report.json \
--markdown benchmark_reports/vectorize_report.md

# ── 3.2 DSL 用例编译 + 模拟基准 ────────────────────────────────────
- name: DSL case compilation benchmarks
run: |
Expand Down Expand Up @@ -363,6 +380,9 @@ jobs:
if [ -f benchmark_reports/const_merge_report.md ]; then
cat benchmark_reports/const_merge_report.md >> $GITHUB_STEP_SUMMARY
fi
if [ -f benchmark_reports/vectorize_report.md ]; then
cat benchmark_reports/vectorize_report.md >> $GITHUB_STEP_SUMMARY
fi
echo "" >> $GITHUB_STEP_SUMMARY
if [ -f benchmark_reports/github_summary.md ]; then
cat benchmark_reports/github_summary.md >> $GITHUB_STEP_SUMMARY
Expand Down
13 changes: 13 additions & 0 deletions benchmarks/cases/topic29_vectorize_feature.dsl
Original file line number Diff line number Diff line change
@@ -0,0 +1,13 @@
# SIMD vectorize feature case (Topic 29, phase 1).
#
# Scalar shape mirrored by the report's vectorizable IR case:
# out[i] = relu(a[i] + a[i]) for i in [0, 16)
# The phase-1 DSL grammar has no array load/store syntax, so the report
# builds the canonical element-addressing IR (design doc appendix 5.1)
# and compiles it through CompilerDriver; this file is still read and
# validated by every driver call and compiled off/on in the wiring check.
for i = 0, 16
t = add(x, x)
y = relu(t)
endfor
return y
Loading
Loading