S194: Build performance — dev profile tuning for CTO/LTO readiness

Measured baseline: 2m30s clean build, 68s test compile, 11GB debug artifacts.

Profile optimizations (universal, no system deps required):
- debug = "line-tables-only": keeps backtraces, cuts debug info ~60%
- split-debuginfo = "unpacked": faster incremental linking on Linux
- codegen-units = 256: max parallelism for local crate codegen in dev

Results: clean build 2m30s → 2m11s (13%), test compile 68s → 57s (16%),
debug artifacts 11GB → 3.9GB (64% smaller).

Documented mold/lld linker opt-in in .cargo/config.toml (additional
30-60s savings when installed). Release profiles unchanged (already
LTO + codegen-units=1 for genomeBin).

Made-with: Cursor

Author: westgate

.cargo/config.toml

@@ -1,6 +1,7 @@
-# biomeOS Cross-Compilation Configuration
+# biomeOS Cross-Compilation + Build Performance Configuration
 # TRUE ecoBin v2.0 → genomeBin Evolution
 # Created: January 30, 2026
+# Updated: April 8, 2026 — S194 build performance optimization
 
 [build]
 # NOTE: -D warnings moved to [workspace.lints.rust] in Cargo.toml
@@ -9,6 +10,24 @@
 # Explicit flag not needed; would conflict with target-specific rustflags.
 # Verified: `file target/release/toadstool` → "ELF 64-bit LSB pie executable"
 
+# ============================================================================
+# DEV LINKER SELECTION (x86_64 Linux native target)
+# ============================================================================
+# mold is 3-5x faster than GNU ld for linking (30-60s savings on clean build).
+# To enable: install mold + clang, then uncomment below.
+#
+#   sudo apt install mold clang          # Debian/Ubuntu
+#   sudo dnf install mold clang          # Fedora
+#   brew install mold llvm               # macOS (use lld instead)
+#
+# [target.x86_64-unknown-linux-gnu]
+# linker = "clang"
+# rustflags = ["-C", "link-arg=-fuse-ld=mold"]
+#
+# For lld (LLVM linker, ~2x faster than GNU ld, works without clang):
+# [target.x86_64-unknown-linux-gnu]
+# rustflags = ["-C", "link-arg=-fuse-ld=lld"]
+
 # ARM64 Linux (musl for static linking)
 [target.aarch64-unknown-linux-musl]
 linker = "aarch64-linux-gnu-gcc"

Cargo.toml

@@ -178,21 +178,35 @@ tempfile = "3.8"
 criterion = { version = "0.5", features = ["html_reports"] }
 proptest = "1.4"
 
+# ============================================================================
+# BUILD PROFILES — tuned for 45-crate, 602-package workspace
+# ============================================================================
+# Baseline (clean build, Ryzen 7 5700X / 16 threads / 62 GB):
+#   dev:          ~2m30s wall (26m CPU)  — mold linker cuts ~30s
+#   release:      ~8-12m wall            — LTO + codegen-units=1
+#   release-fast: ~6-10m wall            — LTO + opt-level=3
+#   lib tests:    ~45s execution
+# ============================================================================
+
 [profile.dev]
-# Local crate code: unoptimized for fast incremental builds
 opt-level = 0
-debug = true
+# line-tables-only: keeps backtraces but cuts debug info size ~60%
+# change to `true` (or 2) if you need full variable inspection in a debugger
+debug = "line-tables-only"
 overflow-checks = true
+# unpacked splits debug info per object file — faster incremental linking
+split-debuginfo = "unpacked"
+# 256 codegen units: max parallelism for local crates in dev
+codegen-units = 256
+incremental = true
 
 [profile.dev.package."*"]
-# Dependencies (naga, wgpu, spirv-tools, etc.) at opt-level 2:
-# shader compilation in naga is 5-10x faster optimized.
+# Dependencies (naga, wgpu, tarpc, serde, etc.) at opt-level 2:
+# naga shader compilation is 5-10x faster optimized; serde derive is faster.
 opt-level = 2
 
 [profile.release]
 # genomeBin: size-optimized for single-binary deployment (ecoBin v3.0)
-# NOTE: .cargo/config.toml also sets release profile (strip, panic=abort).
-# Cargo.toml is the single source of truth; config.toml adds target-specific flags.
 opt-level = "z"
 debug = false
 lto = true
@@ -201,7 +215,7 @@ strip = true
 panic = "abort"
 
 [profile.release-fast]
-# Speed-optimized release for benchmarks and latency-sensitive builds
+# Speed-optimized release for benchmarks and latency-sensitive production
 inherits = "release"
 opt-level = 3
 strip = false