Skip to content

build/wasm: retain final Go and C++ DWARF after optimization - #2662

Merged
cpunion merged 3 commits into
xgo-dev:mainfrom
cpunion:codex/wasm-final-dwarf
Sep 27, 2026
Merged

build/wasm: retain final Go and C++ DWARF after optimization#2662
cpunion merged 3 commits into
xgo-dev:mainfrom
cpunion:codex/wasm-final-dwarf

Conversation

@cpunion

@cpunion cpunion commented Sep 23, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator

Optimized W32 debug builds ran wasm-opt -O2 before Asyncify without -g. The program still ran, but final DWARF marked Go and C++ functions as dead and source lookup failed. Preserve debug information through that pre-Asyncify optimization when the build retains DWARF.

Add a Go-to-C++ fixture that checks the final module after all Binaryen rewrites: llvm-dwarfdump --verify, Go/C++ DIEs and variables, source-line lookup with llvm-addr2line, and Node/Wasmtime execution at O0 and O2. The pinned llgo-v132.3 release includes the Memory64 DWARF fix from xgo-dev/binaryen#4, so CI now runs J32, J64, and W32 instead of gating J64.

This PR is stacked on #2652, which pins llgo-v132.3 for Emscripten and LLGo's standalone post-link. The diff against main includes that dependency.

Validation: all six J32/J64/W32 × O0/O2 cases passed locally using the published macOS ARM64 llgo-v132.3 archive and Emscripten 6.0.8. The full dev/test_wasm_single_worker.sh test-command suite, focused internal/build post-link tests, actionlint, and git diff --check passed.

fennoai[bot]
fennoai Bot reviewed Sep 23, 2026
View reviewed changes

@fennoai fennoai Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Review: Final WebAssembly DWARF verification + LLGo-patched Binaryen

Overall this is a clean, well-tested change. The Go post-link code (resolveWasmOpt, wasmOptIdentity, and the new debug threading into wasmPreAsyncifyArgs) is single-purpose and covered by both success and failure-path tests. The CI action validates the checksum format on both bash and PowerShell paths, asserts the extracted binaries are executable, and downloads over HTTPS. Fixtures and the new dev/test_wasm_debug_info.py integration test line up with the asserted source lines and DWARF names.

Two things worth a look inline, plus one architectural note below. Nothing blocking.

Supply-chain note (non-blocking): The Binaryen download source moves from WebAssembly/binaryen to the xgo-dev/binaryen fork (llgo-v132.2). The pinned SHA-256s guarantee byte-integrity of the fork's published archives, but not that they were built from audited upstream source, so CI's build-tool trust now depends on the xgo-dev org's release process. Consider documenting the fork's build provenance (ideally reproducible builds or SLSA/gh attestation) and restricting who can publish releases there.

Comment thread .github/actions/setup-binaryen/checksums.txt Outdated
Comment thread dev/test_wasm_debug_info.py Outdated
@codecov

codecov Bot commented Sep 23, 2026 •
edited
Loading

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

@github-actions

github-actions Bot commented Sep 24, 2026 •
edited
Loading

Copy link
Copy Markdown

LLGo baseline benchmarks

1ef7918258a7 | workflow run | long-term charts

Program measurements

Platform Workload File size vs base Text size vs base Build vs base Run vs base
Linux cprintf 7160 B 0 B / +0.0% 387 B 0 B / +0.0% 492.926 ms -1.16 ms / -0.2% (better) 1.245 ms -62.77 us / -4.8% (better)
Linux cprintf-lto 6912 B 0 B / +0.0% 368 B 0 B / +0.0% 497.995 ms -744.4 us / -0.1% (better) 1.266 ms +22.6 us / +1.8% (worse)
Linux fmtprintf 1637960 B 0 B / +0.0% 484730 B 0 B / +0.0% 4.081 s -29.92 ms / -0.7% (better) 3.119 ms -37.42 us / -1.2% (better)
Linux fmtprintf-lto 1478416 B 0 B / +0.0% 422446 B 0 B / +0.0% 11.475 s -187.1 ms / -1.6% (better) 2.916 ms -157 us / -5.1% (better)
Linux println 61344 B 0 B / +0.0% 14171 B 0 B / +0.0% 512.947 ms -5.567 ms / -1.1% (better) 1.717 ms +55.2 us / +3.3% (worse)
Linux println-lto 54152 B 0 B / +0.0% 12073 B 0 B / +0.0% 708.130 ms -29.47 ms / -4.0% (better) 1.773 ms +218.5 us / +14.1% (worse)
macOS cprintf 68064 B 0 B / +0.0% 4429 B 0 B / +0.0% 747.372 ms +151.8 ms / +25.5% (worse) 3.745 ms +726 us / +24.0% (worse)
macOS cprintf-lto 84288 B 0 B / +0.0% 193 B 0 B / +0.0% 637.831 ms +5.773 ms / +0.9% (worse) 3.185 ms +144.7 us / +4.8% (worse)
macOS fmtprintf 1483856 B 0 B / +0.0% 860140 B 0 B / +0.0% 3.357 s -143.7 ms / -4.1% (better) 4.794 ms -1.46 ms / -23.3% (better)
macOS fmtprintf-lto 1175808 B 0 B / +0.0% 832752 B 0 B / +0.0% 8.253 s +65.4 ms / +0.8% (worse) 4.880 ms -552.7 us / -10.2% (better)
macOS println 114512 B 0 B / +0.0% 34330 B 0 B / +0.0% 602.830 ms -52.57 ms / -8.0% (better) 3.819 ms +142.9 us / +3.9% (worse)
macOS println-lto 118720 B 0 B / +0.0% 32308 B 0 B / +0.0% 748.854 ms -18.6 ms / -2.4% (better) 3.739 ms +188.3 us / +5.3% (worse)
Windows MinGW cprintf 19456 B 0 B / +0.0% 4550 B 0 B / +0.0% 1.095 s -103.3 ms / -8.6% (better) 3.436 ms -124.6 us / -3.5% (better)
Windows MinGW cprintf-lto 17920 B 0 B / +0.0% 4486 B 0 B / +0.0% 1.112 s -494.3 us / -0.04444% (better) 3.382 ms -57.2 us / -1.7% (better)
Windows MinGW fmtprintf 1910784 B 0 B / +0.0% 586502 B 0 B / +0.0% 3.983 s +24.06 ms / +0.6% (worse) 7.750 ms -299 us / -3.7% (better)
Windows MinGW fmtprintf-lto 1934848 B 0 B / +0.0% 535206 B 0 B / +0.0% 9.897 s -13.76 ms / -0.1% (better) 8.239 ms +56.3 us / +0.7% (worse)
Windows MinGW println 70656 B 0 B / +0.0% 23062 B 0 B / +0.0% 1.079 s -24.48 ms / -2.2% (better) 6.612 ms +246.7 us / +3.9% (worse)
Windows MinGW println-lto 65024 B 0 B / +0.0% 20358 B 0 B / +0.0% 1.301 s +3.909 ms / +0.3% (worse) 6.575 ms +88.9 us / +1.4% (worse)
Windows MinGW 386 cprintf 43008 B 0 B / +0.0% 5326 B 0 B / +0.0% 1.088 s -53.67 ms / -4.7% (better) 5.661 ms +248.3 us / +4.6% (worse)
Windows MinGW 386 cprintf-lto 20992 B 0 B / +0.0% 5094 B 0 B / +0.0% 1.115 s -71.39 ms / -6.0% (better) 5.125 ms -45.6 us / -0.9% (better)
Windows MinGW 386 fmtprintf 1876480 B 0 B / +0.0% 462574 B 0 B / +0.0% 4.111 s -133.6 ms / -3.1% (better) 10.024 ms -1.098 ms / -9.9% (better)
Windows MinGW 386 fmtprintf-lto 2152448 B 0 B / +0.0% 440194 B 0 B / +0.0% 9.505 s -187.4 ms / -1.9% (better) 10.106 ms -68.4 us / -0.7% (better)
Windows MinGW 386 println 91136 B 0 B / +0.0% 19814 B 0 B / +0.0% 1.110 s -61.11 ms / -5.2% (better) 8.409 ms -1.099 ms / -11.6% (better)
Windows MinGW 386 println-lto 69120 B 0 B / +0.0% 17726 B 0 B / +0.0% 1.271 s -119 ms / -8.6% (better) 8.305 ms -382.7 us / -4.4% (better)
Windows MinGW ARM64 cprintf 18944 B 0 B / +0.0% 4408 B 0 B / +0.0% 1.173 s -27.23 ms / -2.3% (better) 5.384 ms -467 us / -8.0% (better)
Windows MinGW ARM64 cprintf-lto 17920 B 0 B / +0.0% 4340 B 0 B / +0.0% 1.237 s +27.53 ms / +2.3% (worse) 5.888 ms +510 us / +9.5% (worse)
Windows MinGW ARM64 fmtprintf 1798656 B 0 B / +0.0% 499196 B 0 B / +0.0% 3.902 s +61.66 ms / +1.6% (worse) 10.948 ms -42.2 us / -0.4% (better)
Windows MinGW ARM64 fmtprintf-lto 1857536 B 0 B / +0.0% 465268 B 0 B / +0.0% 9.187 s +515.7 ms / +5.9% (worse) 12.117 ms +1.127 ms / +10.3% (worse)
Windows MinGW ARM64 println 67072 B 0 B / +0.0% 21816 B 0 B / +0.0% 1.239 s -2.349 ms / -0.2% (better) 9.798 ms +134.6 us / +1.4% (worse)
Windows MinGW ARM64 println-lto 63488 B 0 B / +0.0% 19436 B 0 B / +0.0% 1.424 s +31.9 ms / +2.3% (worse) 9.548 ms -76.7 us / -0.8% (better)
Windows MSVC cprintf 120320 B 0 B / +0.0% 65782 B 0 B / +0.0% 917.808 ms -26.33 ms / -2.8% (better) 3.564 ms -886.4 us / -19.9% (better)
Windows MSVC cprintf-lto 119808 B 0 B / +0.0% 65718 B 0 B / +0.0% 1.073 s +120.3 ms / +12.6% (worse) 4.354 ms +627.3 us / +16.8% (worse)
Windows MSVC fmtprintf 1625088 B 0 B / +0.0% 682038 B 0 B / +0.0% 3.848 s +14.2 ms / +0.4% (worse) 10.297 ms +651.2 us / +6.8% (worse)
Windows MSVC fmtprintf-lto 1616384 B 0 B / +0.0% 635206 B 0 B / +0.0% 9.294 s +84.77 ms / +0.9% (worse) 10.606 ms +267.1 us / +2.6% (worse)
Windows MSVC println 192000 B 0 B / +0.0% 118486 B 0 B / +0.0% 956.273 ms +4.754 ms / +0.5% (worse) 8.098 ms -1.243 ms / -13.3% (better)
Windows MSVC println-lto 189952 B 0 B / +0.0% 116326 B 0 B / +0.0% 1.119 s -23.42 ms / -2.1% (better) 7.795 ms +50.4 us / +0.7% (worse)
Windows MSVC 386 cprintf 9728 B 0 B / +0.0% 3931 B 0 B / +0.0% 964.456 ms -131.7 ms / -12.0% (better) 7.251 ms +398.4 us / +5.8% (worse)
Windows MSVC 386 cprintf-lto 9216 B 0 B / +0.0% 3853 B 0 B / +0.0% 1.094 s +76.47 ms / +7.5% (worse) 6.625 ms -369.2 us / -5.3% (better)
Windows MSVC 386 fmtprintf 1190400 B 0 B / +0.0% 445957 B 0 B / +0.0% 4.108 s -16.92 ms / -0.4% (better) 13.145 ms -1.054 ms / -7.4% (better)
Windows MSVC 386 fmtprintf-lto 1224192 B 0 B / +0.0% 416649 B 0 B / +0.0% 9.289 s -96.61 ms / -1.0% (better) 13.855 ms +365.7 us / +2.7% (worse)
Windows MSVC 386 println 34304 B 0 B / +0.0% 18641 B 0 B / +0.0% 996.649 ms -20.49 ms / -2.0% (better) 11.527 ms +491.7 us / +4.5% (worse)
Windows MSVC 386 println-lto 32256 B 0 B / +0.0% 16839 B 0 B / +0.0% 1.174 s -26.87 ms / -2.2% (better) 11.440 ms -320.7 us / -2.7% (better)
Windows MSVC ARM64 cprintf 11776 B 0 B / +0.0% 4192 B 0 B / +0.0% 1.125 s +71.81 ms / +6.8% (worse) 7.524 ms +976.2 us / +14.9% (worse)
Windows MSVC ARM64 cprintf-lto 10752 B 0 B / +0.0% 4084 B 0 B / +0.0% 1.098 s +24.7 ms / +2.3% (worse) 7.094 ms +287.3 us / +4.2% (worse)
Windows MSVC ARM64 fmtprintf 1370624 B 0 B / +0.0% 499068 B 0 B / +0.0% 3.819 s +35.75 ms / +0.9% (worse) 14.578 ms +560.8 us / +4.0% (worse)
Windows MSVC ARM64 fmtprintf-lto 1387008 B 0 B / +0.0% 465900 B 0 B / +0.0% 8.567 s -168.6 ms / -1.9% (better) 13.730 ms -1.247 ms / -8.3% (better)
Windows MSVC ARM64 println 41984 B 0 B / +0.0% 21824 B 0 B / +0.0% 1.100 s +35.4 ms / +3.3% (worse) 12.666 ms +1.607 ms / +14.5% (worse)
Windows MSVC ARM64 println-lto 40448 B 0 B / +0.0% 19556 B 0 B / +0.0% 1.266 s +63.46 ms / +5.3% (worse) 12.198 ms +834.4 us / +7.3% (worse)
Core language and compiler benchmarks
Platform Benchmark ns/op vs base
Linux BenchmarkLookupPCRandom 14.580 ns/op -0.36 ns/op / -2.4% (better)
Linux BenchmarkMergeCompilerFlags 215.600 ns/op +12.6 ns/op / +6.2% (worse)
Linux BenchmarkMergeLinkerFlags 135.800 ns/op +0.8 ns/op / +0.6% (worse)
Linux BenchmarkChannelBuffered 55.170 ns/op -0.24 ns/op / -0.4% (better)
Linux BenchmarkChannelHandoff 13774 ns/op +188 ns/op / +1.4% (worse)
Linux BenchmarkDefer 52.790 ns/op +1.86 ns/op / +3.7% (worse)
Linux BenchmarkDirectCall 1.596 ns/op +0.023 ns/op / +1.5% (worse)
Linux BenchmarkGlobalRead 1.559 ns/op -0.003 ns/op / -0.2% (better)
Linux BenchmarkGlobalWrite 7.801 ns/op -0.339 ns/op / -4.2% (better)
Linux BenchmarkGoroutine 31164 ns/op +9641 ns/op / +44.8% (worse)
Linux BenchmarkInterfaceCall 6.442 ns/op +0.007 ns/op / +0.1% (worse)
Linux BenchmarkRuntimeGetG 2.440 ns/op 0 ns/op / +0.0%
macOS BenchmarkLookupPCRandom 14.690 ns/op +0.63 ns/op / +4.5% (worse)
macOS BenchmarkMergeCompilerFlags 148.600 ns/op +10.8 ns/op / +7.8% (worse)
macOS BenchmarkMergeLinkerFlags 98.420 ns/op +24.67 ns/op / +33.5% (worse)
macOS BenchmarkChannelBuffered 30.770 ns/op -0.89 ns/op / -2.8% (better)
macOS BenchmarkChannelHandoff 10200 ns/op +3855 ns/op / +60.8% (worse)
macOS BenchmarkDefer 40.870 ns/op +0.65 ns/op / +1.6% (worse)
macOS BenchmarkDirectCall 1.196 ns/op -0.022 ns/op / -1.8% (better)
macOS BenchmarkGlobalRead 1.296 ns/op -0.109 ns/op / -7.8% (better)
macOS BenchmarkGlobalWrite 1.274 ns/op +0.009 ns/op / +0.7% (worse)
macOS BenchmarkGoroutine 53475 ns/op +15458 ns/op / +40.7% (worse)
macOS BenchmarkInterfaceCall 4.821 ns/op +0.048 ns/op / +1.0% (worse)
macOS BenchmarkRuntimeGetG 2.584 ns/op +0.001 ns/op / +0.03871% (worse)
Windows MinGW BenchmarkLookupPCRandom 12.400 ns/op +0.01 ns/op / +0.1% (worse)
Windows MinGW BenchmarkMergeCompilerFlags 527.700 ns/op -13.3 ns/op / -2.5% (better)
Windows MinGW BenchmarkMergeLinkerFlags 476.200 ns/op +1.2 ns/op / +0.3% (worse)
Windows MinGW BenchmarkChannelBuffered 30.650 ns/op -0.25 ns/op / -0.8% (better)
Windows MinGW BenchmarkChannelHandoff 1343 ns/op +28 ns/op / +2.1% (worse)
Windows MinGW BenchmarkDefer 52.740 ns/op +0.07 ns/op / +0.1% (worse)
Windows MinGW BenchmarkDirectCall 1.746 ns/op 0 ns/op / +0.0%
Windows MinGW BenchmarkGlobalRead 1.777 ns/op -0.154 ns/op / -8.0% (better)
Windows MinGW BenchmarkGlobalWrite 2.790 ns/op -0.001 ns/op / -0.03583% (better)
Windows MinGW BenchmarkGoroutine 73662 ns/op +852 ns/op / +1.2% (worse)
Windows MinGW BenchmarkInterfaceCall 8.749 ns/op +0.009 ns/op / +0.1% (worse)
Windows MinGW BenchmarkRuntimeGetG 1.811 ns/op +0.001 ns/op / +0.1% (worse)
Windows MinGW 386 BenchmarkLookupPCRandom 26.650 ns/op +0.02 ns/op / +0.1% (worse)
Windows MinGW 386 BenchmarkMergeCompilerFlags 763.500 ns/op -27.9 ns/op / -3.5% (better)
Windows MinGW 386 BenchmarkMergeLinkerFlags 719.800 ns/op +13.6 ns/op / +1.9% (worse)
Windows MinGW 386 BenchmarkChannelBuffered 38.840 ns/op -0.38 ns/op / -1.0% (better)
Windows MinGW 386 BenchmarkChannelHandoff 932.800 ns/op +22.5 ns/op / +2.5% (worse)
Windows MinGW 386 BenchmarkDefer 42.630 ns/op -1.41 ns/op / -3.2% (better)
Windows MinGW 386 BenchmarkDirectCall 1.551 ns/op +0.004 ns/op / +0.3% (worse)
Windows MinGW 386 BenchmarkGlobalRead 1.552 ns/op +0.002 ns/op / +0.1% (worse)
Windows MinGW 386 BenchmarkGlobalWrite 7.789 ns/op +0.016 ns/op / +0.2% (worse)
Windows MinGW 386 BenchmarkGoroutine 88355 ns/op +6047 ns/op / +7.3% (worse)
Windows MinGW 386 BenchmarkInterfaceCall 8.389 ns/op +0.018 ns/op / +0.2% (worse)
Windows MinGW 386 BenchmarkRuntimeGetG 2.170 ns/op +0.002 ns/op / +0.1% (worse)
Windows MinGW ARM64 BenchmarkLookupPCRandom 12.070 ns/op -0.01 ns/op / -0.1% (better)
Windows MinGW ARM64 BenchmarkMergeCompilerFlags 567.400 ns/op -1 ns/op / -0.2% (better)
Windows MinGW ARM64 BenchmarkMergeLinkerFlags 533 ns/op +14.8 ns/op / +2.9% (worse)
Windows MinGW ARM64 BenchmarkChannelBuffered 37.790 ns/op -0.05 ns/op / -0.1% (better)
Windows MinGW ARM64 BenchmarkChannelHandoff 1701 ns/op -52 ns/op / -3.0% (better)
Windows MinGW ARM64 BenchmarkDefer 53.270 ns/op +2.06 ns/op / +4.0% (worse)
Windows MinGW ARM64 BenchmarkDirectCall 0.589 ns/op -0.0006 ns/op / -0.1% (better)
Windows MinGW ARM64 BenchmarkGlobalRead 0.664 ns/op +0.0006 ns/op / +0.1% (worse)
Windows MinGW ARM64 BenchmarkGlobalWrite 0.663 ns/op -0.0009 ns/op / -0.1% (better)
Windows MinGW ARM64 BenchmarkGoroutine 52535 ns/op +570 ns/op / +1.1% (worse)
Windows MinGW ARM64 BenchmarkInterfaceCall 4.156 ns/op -0.003 ns/op / -0.1% (better)
Windows MinGW ARM64 BenchmarkRuntimeGetG 1.802 ns/op +0.01 ns/op / +0.6% (worse)
Windows MSVC BenchmarkLookupPCRandom 11.690 ns/op -0.07 ns/op / -0.6% (better)
Windows MSVC BenchmarkMergeCompilerFlags 573.400 ns/op +0.2 ns/op / +0.03489% (worse)
Windows MSVC BenchmarkMergeLinkerFlags 527.800 ns/op +20.3 ns/op / +4.0% (worse)
Windows MSVC BenchmarkChannelBuffered 44.630 ns/op +0.46 ns/op / +1.0% (worse)
Windows MSVC BenchmarkChannelHandoff 3225 ns/op +1463 ns/op / +83.0% (worse)
Windows MSVC BenchmarkDefer 67.600 ns/op -0.83 ns/op / -1.2% (better)
Windows MSVC BenchmarkDirectCall 1.213 ns/op +0.036 ns/op / +3.1% (worse)
Windows MSVC BenchmarkGlobalRead 1.066 ns/op +0.006 ns/op / +0.6% (worse)
Windows MSVC BenchmarkGlobalWrite 8.713 ns/op +0.071 ns/op / +0.8% (worse)
Windows MSVC BenchmarkGoroutine 66705 ns/op +203 ns/op / +0.3% (worse)
Windows MSVC BenchmarkInterfaceCall 6.558 ns/op -0.109 ns/op / -1.6% (better)
Windows MSVC BenchmarkRuntimeGetG 1.656 ns/op +0.102 ns/op / +6.6% (worse)
Windows MSVC 386 BenchmarkLookupPCRandom 72.250 ns/op +0.28 ns/op / +0.4% (worse)
Windows MSVC 386 BenchmarkMergeCompilerFlags 807.600 ns/op +17 ns/op / +2.2% (worse)
Windows MSVC 386 BenchmarkMergeLinkerFlags 759.100 ns/op +3.9 ns/op / +0.5% (worse)
Windows MSVC 386 BenchmarkChannelBuffered 55.900 ns/op +0.02 ns/op / +0.03579% (worse)
Windows MSVC 386 BenchmarkChannelHandoff 3838 ns/op +177 ns/op / +4.8% (worse)
Windows MSVC 386 BenchmarkDefer 52.460 ns/op -1.19 ns/op / -2.2% (better)
Windows MSVC 386 BenchmarkDirectCall 1.169 ns/op +0.084 ns/op / +7.7% (worse)
Windows MSVC 386 BenchmarkGlobalRead 1.383 ns/op +0.01 ns/op / +0.7% (worse)
Windows MSVC 386 BenchmarkGlobalWrite 17.750 ns/op -0.02 ns/op / -0.1% (better)
Windows MSVC 386 BenchmarkGoroutine 85200 ns/op +669 ns/op / +0.8% (worse)
Windows MSVC 386 BenchmarkInterfaceCall 5.866 ns/op +0.141 ns/op / +2.5% (worse)
Windows MSVC 386 BenchmarkRuntimeGetG 1.378 ns/op +0.001 ns/op / +0.1% (worse)
Windows MSVC ARM64 BenchmarkLookupPCRandom 12.080 ns/op -0.03 ns/op / -0.2% (better)
Windows MSVC ARM64 BenchmarkMergeCompilerFlags 571.500 ns/op +1 ns/op / +0.2% (worse)
Windows MSVC ARM64 BenchmarkMergeLinkerFlags 535.500 ns/op +2.4 ns/op / +0.5% (worse)
Windows MSVC ARM64 BenchmarkChannelBuffered 37.780 ns/op -1.83 ns/op / -4.6% (better)
Windows MSVC ARM64 BenchmarkChannelHandoff 2328 ns/op -80 ns/op / -3.3% (better)
Windows MSVC ARM64 BenchmarkDefer 61.660 ns/op -1.26 ns/op / -2.0% (better)
Windows MSVC ARM64 BenchmarkDirectCall 0.590 ns/op 0 ns/op / +0.0%
Windows MSVC ARM64 BenchmarkGlobalRead 0.664 ns/op +0.0002 ns/op / +0.03014% (worse)
Windows MSVC ARM64 BenchmarkGlobalWrite 3.753 ns/op -0.006 ns/op / -0.2% (better)
Windows MSVC ARM64 BenchmarkGoroutine 55285 ns/op +1030 ns/op / +1.9% (worse)
Windows MSVC ARM64 BenchmarkInterfaceCall 4.144 ns/op -0.001 ns/op / -0.02413% (better)
Windows MSVC ARM64 BenchmarkRuntimeGetG 2.243 ns/op 0 ns/op / +0.0%
Timer runtime benchmarks
Platform Operation and runtime ns/op vs base
Linux AfterFuncZeroDelivery/Go 898.500 ns/op -17.4 ns/op / -1.9% (better)
Linux AfterFuncZeroDelivery/LLGo 40611 ns/op -1601 ns/op / -3.8% (better)
Linux CreateStop/Go 296.200 ns/op -6 ns/op / -2.0% (better)
Linux CreateStop/LLGo 1453 ns/op -282 ns/op / -16.3% (better)
Linux RearmStopped/Go 114.700 ns/op -4 ns/op / -3.4% (better)
Linux RearmStopped/LLGo 1314 ns/op +66 ns/op / +5.3% (worse)
Linux ResetActive/Go 67.540 ns/op -2.08 ns/op / -3.0% (better)
Linux ResetActive/LLGo 783.900 ns/op +110.6 ns/op / +16.4% (worse)
Linux ResetHeap1024/Go 67.210 ns/op -0.15 ns/op / -0.2% (better)
Linux ResetHeap1024/LLGo 184.700 ns/op +1.7 ns/op / +0.9% (worse)
macOS AfterFuncZeroDelivery/Go 522.500 ns/op +18.9 ns/op / +3.8% (worse)
macOS AfterFuncZeroDelivery/LLGo 90654 ns/op +11432 ns/op / +14.4% (worse)
macOS CreateStop/Go 150.900 ns/op +0.3 ns/op / +0.2% (worse)
macOS CreateStop/LLGo 456.300 ns/op -16.8 ns/op / -3.6% (better)
macOS RearmStopped/Go 70.280 ns/op +2.61 ns/op / +3.9% (worse)
macOS RearmStopped/LLGo 511.700 ns/op +34.6 ns/op / +7.3% (worse)
macOS ResetActive/Go 49.990 ns/op -1.07 ns/op / -2.1% (better)
macOS ResetActive/LLGo 169.300 ns/op +2.9 ns/op / +1.7% (worse)
macOS ResetHeap1024/Go 48.910 ns/op +1.43 ns/op / +3.0% (worse)
macOS ResetHeap1024/LLGo 96.100 ns/op -6.2 ns/op / -6.1% (better)
Windows MinGW AfterFuncZeroDelivery/Go 474.100 ns/op -13.7 ns/op / -2.8% (better)
Windows MinGW AfterFuncZeroDelivery/LLGo 132302 ns/op -2539 ns/op / -1.9% (better)
Windows MinGW CreateStop/Go 115.200 ns/op -1.8 ns/op / -1.5% (better)
Windows MinGW CreateStop/LLGo 477.800 ns/op +30.5 ns/op / +6.8% (worse)
Windows MinGW RearmStopped/Go 31.500 ns/op -0.06 ns/op / -0.2% (better)
Windows MinGW RearmStopped/LLGo 283.500 ns/op -4.7 ns/op / -1.6% (better)
Windows MinGW ResetActive/Go 19.020 ns/op -0.04 ns/op / -0.2% (better)
Windows MinGW ResetActive/LLGo 165.500 ns/op -10.5 ns/op / -6.0% (better)
Windows MinGW ResetHeap1024/Go 19.110 ns/op -0.53 ns/op / -2.7% (better)
Windows MinGW ResetHeap1024/LLGo 139 ns/op +1.7 ns/op / +1.2% (worse)
Windows MinGW 386 AfterFuncZeroDelivery/Go 952.100 ns/op -12.3 ns/op / -1.3% (better)
Windows MinGW 386 AfterFuncZeroDelivery/LLGo 184941 ns/op +1775 ns/op / +1.0% (worse)
Windows MinGW 386 CreateStop/Go 196.900 ns/op +3.3 ns/op / +1.7% (worse)
Windows MinGW 386 CreateStop/LLGo 498 ns/op +1.9 ns/op / +0.4% (worse)
Windows MinGW 386 RearmStopped/Go 63.360 ns/op -0.74 ns/op / -1.2% (better)
Windows MinGW 386 RearmStopped/LLGo 353.900 ns/op +21.7 ns/op / +6.5% (worse)
Windows MinGW 386 ResetActive/Go 39.100 ns/op +0.01 ns/op / +0.02558% (worse)
Windows MinGW 386 ResetActive/LLGo 1000 ns/op -14 ns/op / -1.4% (better)
Windows MinGW 386 ResetHeap1024/Go 39.410 ns/op -0.13 ns/op / -0.3% (better)
Windows MinGW 386 ResetHeap1024/LLGo 186.600 ns/op +1 ns/op / +0.5% (worse)
Windows MinGW ARM64 AfterFuncZeroDelivery/Go 661.100 ns/op -7.6 ns/op / -1.1% (better)
Windows MinGW ARM64 AfterFuncZeroDelivery/LLGo 136359 ns/op +7755 ns/op / +6.0% (worse)
Windows MinGW ARM64 CreateStop/Go 195.500 ns/op -3.2 ns/op / -1.6% (better)
Windows MinGW ARM64 CreateStop/LLGo 371.800 ns/op +10.6 ns/op / +2.9% (worse)
Windows MinGW ARM64 RearmStopped/Go 70.520 ns/op -0.08 ns/op / -0.1% (better)
Windows MinGW ARM64 RearmStopped/LLGo 251.300 ns/op +4.3 ns/op / +1.7% (worse)
Windows MinGW ARM64 ResetActive/Go 31.020 ns/op +0.17 ns/op / +0.6% (worse)
Windows MinGW ARM64 ResetActive/LLGo 125.800 ns/op -0.6 ns/op / -0.5% (better)
Windows MinGW ARM64 ResetHeap1024/Go 31.090 ns/op +0.02 ns/op / +0.1% (worse)
Windows MinGW ARM64 ResetHeap1024/LLGo 127.200 ns/op +1.2 ns/op / +1.0% (worse)
Windows MSVC AfterFuncZeroDelivery/Go 710.200 ns/op +0.8 ns/op / +0.1% (worse)
Windows MSVC AfterFuncZeroDelivery/LLGo 141499 ns/op +3537 ns/op / +2.6% (worse)
Windows MSVC CreateStop/Go 190.600 ns/op -1.6 ns/op / -0.8% (better)
Windows MSVC CreateStop/LLGo 909.700 ns/op +159.1 ns/op / +21.2% (worse)
Windows MSVC RearmStopped/Go 71.500 ns/op +0.06 ns/op / +0.1% (worse)
Windows MSVC RearmStopped/LLGo 368.600 ns/op -28.2 ns/op / -7.1% (better)
Windows MSVC ResetActive/Go 31.790 ns/op +0.05 ns/op / +0.2% (worse)
Windows MSVC ResetActive/LLGo 208.500 ns/op +9.4 ns/op / +4.7% (worse)
Windows MSVC ResetHeap1024/Go 32.080 ns/op +0.02 ns/op / +0.1% (worse)
Windows MSVC ResetHeap1024/LLGo 135.500 ns/op +3.5 ns/op / +2.7% (worse)
Windows MSVC 386 AfterFuncZeroDelivery/Go 1052 ns/op +8 ns/op / +0.8% (worse)
Windows MSVC 386 AfterFuncZeroDelivery/LLGo 175867 ns/op -1344 ns/op / -0.8% (better)
Windows MSVC 386 CreateStop/Go 266.700 ns/op +0.7 ns/op / +0.3% (worse)
Windows MSVC 386 CreateStop/LLGo 776.800 ns/op +42.3 ns/op / +5.8% (worse)
Windows MSVC 386 RearmStopped/Go 96.520 ns/op +0.09 ns/op / +0.1% (worse)
Windows MSVC 386 RearmStopped/LLGo 427.900 ns/op +1.8 ns/op / +0.4% (worse)
Windows MSVC 386 ResetActive/Go 46.580 ns/op 0 ns/op / +0.0%
Windows MSVC 386 ResetActive/LLGo 314.900 ns/op -5.8 ns/op / -1.8% (better)
Windows MSVC 386 ResetHeap1024/Go 47 ns/op +0.06 ns/op / +0.1% (worse)
Windows MSVC 386 ResetHeap1024/LLGo 171.900 ns/op -1 ns/op / -0.6% (better)
Windows MSVC ARM64 AfterFuncZeroDelivery/Go 662.500 ns/op -0.5 ns/op / -0.1% (better)
Windows MSVC ARM64 AfterFuncZeroDelivery/LLGo 122460 ns/op -13921 ns/op / -10.2% (better)
Windows MSVC ARM64 CreateStop/Go 201.700 ns/op +3.4 ns/op / +1.7% (worse)
Windows MSVC ARM64 CreateStop/LLGo 385.400 ns/op -1.4 ns/op / -0.4% (better)
Windows MSVC ARM64 RearmStopped/Go 70.560 ns/op +0.01 ns/op / +0.01417% (worse)
Windows MSVC ARM64 RearmStopped/LLGo 269.200 ns/op -0.8 ns/op / -0.3% (better)
Windows MSVC ARM64 ResetActive/Go 30.980 ns/op -0.07 ns/op / -0.2% (better)
Windows MSVC ARM64 ResetActive/LLGo 134.700 ns/op +1.5 ns/op / +1.1% (worse)
Windows MSVC ARM64 ResetHeap1024/Go 31.110 ns/op -0.01 ns/op / -0.03213% (better)
Windows MSVC ARM64 ResetHeap1024/LLGo 137.300 ns/op +0.7 ns/op / +0.5% (worse)

Compared with a34388d0a9ab measured in the same runner job.

@github-actions

github-actions Bot commented Sep 24, 2026 •
edited
Loading

Copy link
Copy Markdown

LLGo WebAssembly build benchmarks

afa8a6c5ea8b | workflow run | long-term charts

WebAssembly output sizes
Example, profile and compiler Wasm module vs base Generated JS glue vs base
cprintf/j32-emscripten/LLGo 135610 B 0 B / +0.0% 70901 B 0 B / +0.0%
cprintf/j32-goos-js/LLGo 134108 B 0 B / +0.0% 69235 B 0 B / +0.0%
cprintf/j64-emscripten-memory64/LLGo 124046 B 0 B / +0.0% 74134 B 0 B / +0.0%
cprintf/w32-goos-wasip1/LLGo 138444 B 0 B / +0.0% 0 B 0 B / 0.0%
cprintf/w32-wasi/LLGo 138170 B 0 B / +0.0% 0 B 0 B / 0.0%
fmtprintf/j32-emscripten/LLGo 3087288 B 0 B / +0.0% 115325 B 0 B / +0.0%
fmtprintf/j32-goos-js/Go 2526852 B 0 B / +0.0% 0 B 0 B / 0.0%
fmtprintf/j32-goos-js/LLGo 3082171 B 0 B / +0.0% 98999 B 0 B / +0.0%
fmtprintf/j64-emscripten-memory64/LLGo 2826659 B 0 B / +0.0% 122264 B 0 B / +0.0%
fmtprintf/w32-goos-wasip1/Go 2500019 B 0 B / +0.0% 0 B 0 B / 0.0%
fmtprintf/w32-goos-wasip1/LLGo 2827999 B 0 B / +0.0% 0 B 0 B / 0.0%
fmtprintf/w32-wasi/LLGo 2694467 B 0 B / +0.0% 0 B 0 B / 0.0%
j32-emscripten/LLGo 134897 B 0 B / +0.0% 70901 B 0 B / +0.0%
j32-goos-js/Go 1895533 B 0 B / +0.0% 0 B 0 B / 0.0%
j32-goos-js/LLGo 133563 B 0 B / +0.0% 69235 B 0 B / +0.0%
j64-emscripten-memory64/LLGo 123371 B 0 B / +0.0% 74134 B 0 B / +0.0%
reflectcall/j32-emscripten/LLGo 1479250 B 0 B / +0.0% 89361 B 0 B / +0.0%
reflectcall/j32-goos-js/Go 2191221 B 0 B / +0.0% 0 B 0 B / 0.0%
reflectcall/j32-goos-js/LLGo 1482639 B 0 B / +0.0% 87695 B 0 B / +0.0%
reflectcall/j64-emscripten-memory64/LLGo 1366522 B 0 B / +0.0% 94654 B 0 B / +0.0%
reflectcall/w32-goos-wasip1/Go 2205707 B 0 B / +0.0% 0 B 0 B / 0.0%
reflectcall/w32-goos-wasip1/LLGo 1533364 B 0 B / +0.0% 0 B 0 B / 0.0%
reflectcall/w32-wasi/LLGo 1458829 B 0 B / +0.0% 0 B 0 B / 0.0%
w32-goos-wasip1/Go 1909947 B 0 B / +0.0% 0 B 0 B / 0.0%
w32-goos-wasip1/LLGo 137725 B 0 B / +0.0% 0 B 0 B / 0.0%
w32-wasi/LLGo 137518 B 0 B / +0.0% 0 B 0 B / 0.0%
LLGo WebAssembly build measurements
Example and profile Build vs base
j32-emscripten 5.687 s -118.7 ms / -2.0% (better)
j32-goos-js 5.874 s -305 ms / -4.9% (better)
j64-emscripten-memory64 4.757 s -394.5 ms / -7.7% (better)
reflectcall/w32-wasi 26.476 s -369.3 ms / -1.4% (better)
w32-goos-wasip1 4.706 s -468.3 ms / -9.0% (better)
w32-wasi 4.637 s +134.4 ms / +3.0% (worse)

Compared with be496b2cc5c3 measured in the same runner job.

This was referenced Sep 24, 2026
Merged
Open
@cpunion
cpunion force-pushed the codex/wasm-final-dwarf branch from 7f23f53 to 1ef7918 Compare September 26, 2026 14:39
@cpunion
cpunion merged commit 5c80bcf into xgo-dev:main Sep 27, 2026
88 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@fennoai fennoai[bot] fennoai[bot] left review comments
@visualfc visualfc visualfc approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@cpunion @visualfc