build: native target-cpu by default, debug 0 everywhere

.cargo/config-v3.toml

@@ -0,0 +1,6 @@
+# Portable distribution baseline: x86-64-v3 (AVX2), every x86_64 since
+# Haswell. REQUIRED for anything shipped from this tree, now that the default
+# `.cargo/config.toml` is `native`. Mirrors ndarray/.cargo/config-v3.toml.
+#   cargo build --config .cargo/config-v3.toml ...
+[target.'cfg(target_arch = "x86_64")']
+rustflags = ["-Ctarget-cpu=x86-64-v3"]

.cargo/config.toml

@@ -16,19 +16,52 @@
 # scalar, because the polyfill's scalar arm is emulating eight lanes through
 # arrays. A build with no target-cpu is a regression, not a no-op.
 #
-# Default here is v3 (AVX2) for the same reason ndarray chose it: portable
-# across all x86_64 since Haswell, and what CI runs (`.github/workflows/*.yml`
-# set `-C target-cpu=x86-64-v3` explicitly). Anything below v3 SIGILLs on the
-# `simd_avx2` intrinsics.
+# ⊘ SUPERSEDED 2026-10-12: the default here WAS v3 (AVX2), "for the same
+# reason ndarray chose it". ndarray itself flipped its default to `native` on
+# 2026-09-16 (see its `.cargo/config.toml`): a default that names a tier the
+# host is not makes every faster-tier measurement need an incantation, and a
+# forgotten one does not fail, it grades the wrong tier. Measured here the
+# same way: a cost probe under this file reported AVX2 timings on an AVX-512
+# host with nothing in the output saying so.
 #
-# AVX-512 (v4):  cargo build --config .cargo/config-avx512.toml ...
-#                or CARGO_BUILD_RUSTFLAGS='-Ctarget-cpu=x86-64-v4'
-# Build machine: cargo build --config .cargo/config-native.toml ...
+# Default is now `native`: rustc resolves the host CPUID, so the polyfill
+# compiles to the arm this machine can run and can never SIGILL here. A
+# measurement is labelled by the arm the program reports
+# (`cfg!(target_feature = "avx512f")`), never by "it was the default".
+#
+# A default build is therefore tuned to the BUILD HOST and is NOT portable.
+# Anything shipped pins its tier explicitly:
+#
+#   portable baseline (AVX2):  cargo build --config .cargo/config-v3.toml ...
+#   AVX-512 (v4):              cargo build --config .cargo/config-avx512.toml ...
+#
+# CI is unaffected: every workflow sets `RUSTFLAGS` with an explicit
+# `-C target-cpu=x86-64-v3`, and a RUSTFLAGS env REPLACES every cargo-config
+# rustflags entry. Everything else that ships pins x86_64 to v3 without
+# touching other architectures: the Dockerfiles that copy this tree replace
+# this file with config-v3.toml (x86_64-only), the thinking-engine images set
+# CARGO_TARGET_X86_64_UNKNOWN_LINUX_GNU_RUSTFLAGS, and `make publish` in
+# python/ sets RUSTFLAGS only on an x86_64 host. Note that the per-target
+# variable is JOINED with this file's entry, with `native` last (measured), so
+# it cannot pin a build that sees this file. Anything below v3 still SIGILLs on the
+# `simd_avx2` intrinsics, so never build with no target-cpu at all.
 #
 # The two crypto cfgs ndarray's config also carries (`curve25519_dalek_backend`,
 # `poly1305_force_soft`) are deliberately NOT repeated: neither crate is in
 # this workspace's Cargo.lock (verified 2026-09-02, 0 hits), so there is no
 # AVX2 crypto surface here to compile out. Re-check if `encryption` or
 # `ed25519-dalek` ever enters the tree.
 [target.'cfg(target_arch = "x86_64")']
-rustflags = ["-Ctarget-cpu=x86-64-v3"]
+rustflags = ["-Ctarget-cpu=native"]
+
+# Debug info off for every dev/test build under this tree, including the
+# crates that are their own workspace (19 of them: they never see the root
+# Cargo.toml's `[profile.dev] debug = 0`). Debug info is what fills the disk,
+# not the code. A config-file profile overrides a manifest profile; an
+# explicit CARGO_PROFILE_* env var still overrides this. Release profiles are
+# left to each crate.
+[profile.dev]
+debug = 0
+
+[profile.test]
+debug = 0

.github/workflows/build.yml

@@ -16,7 +16,7 @@ concurrency:
 
 env:
   CARGO_TERM_COLOR: always
-  RUSTFLAGS: "-C debuginfo=1 -C target-cpu=x86-64-v3"
+  RUSTFLAGS: "-C debuginfo=0 -C target-cpu=x86-64-v3"
   RUST_BACKTRACE: "1"
   CARGO_INCREMENTAL: "0"
 
@@ -25,7 +25,7 @@ jobs:
     runs-on: ubuntu-24.04
     timeout-minutes: 30
     env:
-      # Override the workflow-level debuginfo=1 for this job (parity with the
+      # Pin debuginfo=0 here too (the workflow level is also 0 since 2026-10-12) for this job (parity with the
       # `test` + `test-with-coverage` jobs in rust-test.yml, TD-CI-TEST-JOB-DEBUGINFO0
       # / TD-CI-COVERAGE-MOLD-1). This job links the SAME full lance+datafusion
       # test-binary set at the SAME disk/RSS cliff and was the one job still

.github/workflows/rust-publish.yml

@@ -20,7 +20,7 @@ on:
 
 env:
   CARGO_TERM_COLOR: always
-  RUSTFLAGS: "-C debuginfo=1 -C target-cpu=x86-64-v3"
+  RUSTFLAGS: "-C debuginfo=0 -C target-cpu=x86-64-v3"
   RUST_BACKTRACE: "1"
   CARGO_INCREMENTAL: "0"
   CARGO_BUILD_JOBS: "1"

.github/workflows/rust-test.yml

@@ -16,7 +16,7 @@ concurrency:
 
 env:
   CARGO_TERM_COLOR: always
-  RUSTFLAGS: "-C debuginfo=1 -C target-cpu=x86-64-v3"
+  RUSTFLAGS: "-C debuginfo=0 -C target-cpu=x86-64-v3"
   RUST_BACKTRACE: "1"
   CARGO_INCREMENTAL: "0"
 
@@ -30,7 +30,7 @@ jobs:
     runs-on: ubuntu-24.04
     timeout-minutes: 30
     env:
-      # Override the workflow-level debuginfo=1 for this job too (parity with
+      # Pin debuginfo=0 here too (the workflow level is also 0 since 2026-10-12) for this job too (parity with
       # test-with-coverage, TD-CI-COVERAGE-MOLD-1). The `test` job links the
       # full lance+datafusion integration-test set at the SAME disk/RSS cliff
       # the coverage job hit — and #507 (+4055 lines across causal-edge +
@@ -506,7 +506,7 @@ jobs:
     runs-on: ubuntu-24.04
     timeout-minutes: 30
     env:
-      # Override the workflow-level debuginfo=1 for this job only. llvm-cov
+      # Pin debuginfo=0 here too (the workflow level is also 0 since 2026-10-12) for this job only. llvm-cov
       # coverage lives in __llvm_covmap/__llvm_prf_* ELF sections, NOT in
       # DWARF (verified locally: 600/600 contract tests pass under
       # `-C instrument-coverage -C debuginfo=0`, covmap sections present,

.github/workflows/style.yml

@@ -16,7 +16,7 @@ concurrency:
 
 env:
   CARGO_TERM_COLOR: always
-  RUSTFLAGS: "-C debuginfo=1 -C target-cpu=x86-64-v3"
+  RUSTFLAGS: "-C debuginfo=0 -C target-cpu=x86-64-v3"
 
 # Least-privilege: these jobs only read the repo (checkout + build + lint).
 permissions:

Dockerfile.hhtld

@@ -48,6 +48,12 @@ RUN if [ ! -d /build/ndarray ]; then \
     fi
 
 WORKDIR /build/lance-graph
+# Portable baseline. The repo's .cargo/config.toml defaults to `native`
+# (tuned to the build host); an image must run on any x86_64 since Haswell.
+# The copied tree's config is replaced by config-v3.toml, which only applies
+# to x86_64 targets, so an arm64 build keeps its own aarch64 baseline (a
+# global RUSTFLAGS would hand the aarch64 compiler an x86 CPU name).
+RUN cp .cargo/config-v3.toml .cargo/config.toml
 
 # Build the decode example (the inference binary)
 RUN cargo build --release --example tts_17b_hhtld_decode \

crates/cognitive-stack/Dockerfile

@@ -35,6 +35,12 @@ WORKDIR /build
 # The lance-graph repo (this build context). cognitive-stack path-deps the
 # in-repo crates (lance-graph, the four template crates, lance-graph-ogar).
 COPY . /build/lance-graph
+# Portable baseline. The repo's .cargo/config.toml defaults to `native`
+# (tuned to the build host); an image must run on any x86_64 since Haswell.
+# The copied tree's config is replaced by config-v3.toml, which only applies
+# to x86_64 targets, so an arm64 build keeps its own aarch64 baseline (a
+# global RUSTFLAGS would hand the aarch64 compiler an x86 CPU name).
+RUN cp /build/lance-graph/.cargo/config-v3.toml /build/lance-graph/.cargo/config.toml
 
 # lance-graph path-deps the ndarray fork as a SIBLING (`../../../ndarray`).
 # Clone it next to the repo so that path resolves.

crates/symbiont/Dockerfile

@@ -35,6 +35,12 @@ WORKDIR /build
 # The lance-graph repo (this build context). symbiont path-deps lance-graph +
 # perturbation-sim from inside it.
 COPY . /build/lance-graph
+# Portable baseline. The repo's .cargo/config.toml defaults to `native`
+# (tuned to the build host); an image must run on any x86_64 since Haswell.
+# The copied tree's config is replaced by config-v3.toml, which only applies
+# to x86_64 targets, so an arm64 build keeps its own aarch64 baseline (a
+# global RUSTFLAGS would hand the aarch64 compiler an x86 CPU name).
+RUN cp /build/lance-graph/.cargo/config-v3.toml /build/lance-graph/.cargo/config.toml
 
 # lance-graph's own Cargo.toml path-deps the ndarray fork as a SIBLING
 # (`../../../ndarray`). Clone it next to the repo so that path resolves.

crates/thinking-engine/Dockerfile

@@ -1,4 +1,8 @@
 FROM rust:1.98.1-bookworm AS builder
+# Portable baseline for x86_64 only: this image does not copy .cargo, so the
+# per-target variable is the only flag source, and it does not apply when
+# building for another architecture.
+ENV CARGO_TARGET_X86_64_UNKNOWN_LINUX_GNU_RUSTFLAGS="-C target-cpu=x86-64-v3"
 WORKDIR /app
 
 # Dependencies first (cached layer)

crates/thinking-engine/Dockerfile.railway

@@ -7,6 +7,10 @@
 # The codebooks run standalone for reasoning at 5,676 queries/sec
 
 FROM rust:1.98.1-bookworm AS builder
+# Portable baseline for x86_64 only: this image does not copy .cargo, so the
+# per-target variable is the only flag source, and it does not apply when
+# building for another architecture.
+ENV CARGO_TARGET_X86_64_UNKNOWN_LINUX_GNU_RUSTFLAGS="-C target-cpu=x86-64-v3"
 WORKDIR /app
 
 # Copy workspace

python/Makefile

@@ -24,5 +24,11 @@ clean:
 	rm -rf .venv
 	rm -rf ../target
 
+# The repo's .cargo/config.toml defaults to `native`, which would tune the
+# wheel to the build host; a published x86_64 wheel must run on any x86_64
+# since Haswell. RUSTFLAGS replaces the config's flags; it is set only on an
+# x86_64 host so an arm64 build keeps its own baseline.
+PORTABLE_RUSTFLAGS := $(if $(filter x86_64,$(shell uname -m)),RUSTFLAGS="-C target-cpu=x86-64-v3",)
+
 publish:
-	maturin build --release --out dist --sdist
+	$(PORTABLE_RUSTFLAGS) maturin build --release --out dist --sdist