tomoaki/hakmem
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
3daf75e57f61f90f7c3fdfbe17de54fc48d8901e
hakmem/core/hakmem_build_flags.h
Moe Charm (CI) e0aa51dba1 Phase 4-Step3: Add Front Config Box (+2.7-4.9% dead code elimination) 
Implement compile-time configuration system for dead code elimination in Tiny
allocation hot paths. The Config Box provides dual-mode configuration:
- Normal mode: Runtime ENV checks (backward compatible, flexible)
- PGO mode: Compile-time constants (dead code elimination, performance)

PERFORMANCE:
- Baseline (runtime config): 50.32 M ops/s (avg of 5 runs)
- Config Box (PGO mode): 52.77 M ops/s (avg of 5 runs)
- Improvement: +2.45 M ops/s (+4.87% with outlier, +2.72% without)
- Target: +5-8% (partially achieved)

IMPLEMENTATION:

1. core/box/tiny_front_config_box.h (NEW):
   - Defines TINY_FRONT_*_ENABLED macros for all config checks
   - PGO mode (#if HAKMEM_TINY_FRONT_PGO): Macros expand to constants (0/1)
   - Normal mode (#else): Macros expand to function calls
   - Functions remain in their original locations (no code duplication)

2. core/hakmem_build_flags.h:
   - Added HAKMEM_TINY_FRONT_PGO build flag (default: 0, off)
   - Documentation: Usage with make EXTRA_CFLAGS="-DHAKMEM_TINY_FRONT_PGO=1"

3. core/box/hak_wrappers.inc.h:
   - Replaced front_gate_unified_enabled() with TINY_FRONT_UNIFIED_GATE_ENABLED
   - 2 call sites updated (malloc and free fast paths)
   - Added config box include

EXPECTED DEAD CODE ELIMINATION (PGO mode):
  if (TINY_FRONT_UNIFIED_GATE_ENABLED) { ... }
  → if (1) { ... }  // Constant, always true
  → Compiler optimizes away the branch, keeps body

SCOPE:
  Currently only front_gate_unified_enabled() is replaced (2 call sites).
  To achieve full +5-8% target, expand to other config checks:
  - ultra_slim_mode_enabled()
  - tiny_heap_v2_enabled()
  - sfc_cascade_enabled()
  - tiny_fastcache_enabled()
  - tiny_metrics_enabled()
  - tiny_diag_enabled()

BUILD USAGE:
  Normal mode (runtime config, default):
    make bench_random_mixed_hakmem

  PGO mode (compile-time config, dead code elimination):
    make EXTRA_CFLAGS="-DHAKMEM_TINY_FRONT_PGO=1" bench_random_mixed_hakmem

BOX PATTERN COMPLIANCE:
✅ Single Responsibility: Configuration management ONLY
✅ Clear Contract: Dual-mode (PGO = constants, Normal = runtime)
✅ Observable: Config report function (debug builds)
✅ Safe: Backward compatible (default is normal mode)
✅ Testable: Easy A/B comparison (PGO vs normal builds)

WHY +2.7-4.9% (below +5-8% target)?
- Limited scope: Only 2 call sites for 1 config function replaced
- Lazy init overhead: front_gate_unified_enabled() cached after first call
- Need to expand to more config checks for full benefit

NEXT STEPS:
- Expand config macro usage to other functions (optional)
- OR proceed with PGO re-enablement (Final polish)

🤖 Generated with Claude Code

Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-29 12:18:37 +09:00

183 lines
6.5 KiB
C
Raw Blame History

// hakmem_build_flags.h - Centralized compile-time feature switches
// Purpose: Define all build-time toggles in one place with safe defaults.
// Usage: Include from common public headers (e.g., hakmem.h / hakmem_tiny.h).
#ifndef HAKMEM_BUILD_FLAGS_H
#define HAKMEM_BUILD_FLAGS_H
// ------------------------------------------------------------
// Release/debug detection
// ------------------------------------------------------------
// HAKMEM_BUILD_RELEASE: 1 in release-like builds, 0 otherwise
#ifndef HAKMEM_BUILD_RELEASE
# if defined(NDEBUG)
# define HAKMEM_BUILD_RELEASE 1
# else
# define HAKMEM_BUILD_RELEASE 0
# endif
#endif
// ------------------------------------------------------------
// Instrumentation & counters (compile-time)
// ------------------------------------------------------------
// Enable lightweight path/debug counters (compiled out when 0)
#ifndef HAKMEM_DEBUG_COUNTERS
# define HAKMEM_DEBUG_COUNTERS 1
#endif
// Enable extended memory profiling (compiled out when 0)
#ifndef HAKMEM_DEBUG_MEMORY
# define HAKMEM_DEBUG_MEMORY 0
#endif
// Tiny refill optimization helpers (header-only)
#ifndef HAKMEM_TINY_REFILL_OPT
# define HAKMEM_TINY_REFILL_OPT 1
#endif
// Batch refill P0 (can be toggled for A/B)
#ifndef HAKMEM_TINY_P0_BATCH_REFILL
# define HAKMEM_TINY_P0_BATCH_REFILL 0
#endif
// Box refactor (Phase 6-1.7) — usually injected from build system
#ifndef HAKMEM_TINY_PHASE6_BOX_REFACTOR
# define HAKMEM_TINY_PHASE6_BOX_REFACTOR 1
#endif
// ------------------------------------------------------------
// Phase 7: Region-ID Direct Lookup (Header-based optimization)
// ------------------------------------------------------------
// Phase 7 Task 1: Header-based class_idx for O(1) free
// Default: OFF (enable after full validation in Task 5)
// Build: make HEADER_CLASSIDX=1 or make phase7
#ifndef HAKMEM_TINY_HEADER_CLASSIDX
# define HAKMEM_TINY_HEADER_CLASSIDX 1
#endif
// Phase 7 Task 2: Aggressive inline TLS cache access
// Default: OFF (enable after full validation in Task 5)
// Build: make AGGRESSIVE_INLINE=1 or make phase7
// Requires: HAKMEM_TINY_HEADER_CLASSIDX=1
#ifndef HAKMEM_TINY_AGGRESSIVE_INLINE
# define HAKMEM_TINY_AGGRESSIVE_INLINE 0
#endif
// Inline TLS SLL pop (experimental, A/B only)
// Default: OFF (HAKMEM_TINY_INLINE_SLL=0) to keep Box TLS-SLL API as the standard path.
// Enable explicitly via build flag: -DHAKMEM_TINY_INLINE_SLL=1 (bench/debug only).
#ifndef HAKMEM_TINY_INLINE_SLL
# define HAKMEM_TINY_INLINE_SLL 0
#endif
// Phase 7 Task 3: Pre-warm TLS cache at init
// Default: OFF (enable after implementation)
// Build: make PREWARM_TLS=1 or make phase7
#ifndef HAKMEM_TINY_PREWARM_TLS
# define HAKMEM_TINY_PREWARM_TLS 0
#endif
// Runtime verbosity (printf-heavy diagnostics). Keep OFF for benches.
#ifndef HAKMEM_DEBUG_VERBOSE
# define HAKMEM_DEBUG_VERBOSE 0
#endif
// Tiny/Mid safety checks on free path (mincore header validation).
// 0 = performance (boundary-only), 1 = strict (mincore for all)
#ifndef HAKMEM_TINY_SAFE_FREE
# define HAKMEM_TINY_SAFE_FREE 0
#endif
// Phase 10: Aggressive refill count defaults (tunable via env vars)
// Goal: Reduce backend transitions by refilling in larger batches
// HAKMEM_TINY_REFILL_COUNT: global default (default: 128)
// HAKMEM_TINY_REFILL_COUNT_HOT: class 0-3 (default: 128)
// HAKMEM_TINY_REFILL_COUNT_MID: class 4-7 (default: 96)
// Larson Fix (Priority 1): Increased from 64 to 128 to reduce lock contention
// Expected impact: Lock frequency reduction 19K → ~1.6K locks/sec (12x)
// NOTE: Multi-threaded Larson has pre-existing crash bug (not caused by this change)
#ifndef HAKMEM_TINY_REFILL_DEFAULT
# define HAKMEM_TINY_REFILL_DEFAULT 128
#endif
// ------------------------------------------------------------
// Tiny front architecture toggles (compile-time defaults)
// ------------------------------------------------------------
// New 3-layer Tiny front (A/B via build flag)
#ifndef HAKMEM_TINY_USE_NEW_3LAYER
# define HAKMEM_TINY_USE_NEW_3LAYER 0
#endif
// Minimal/strict front variants (bench/debug only)
#ifndef HAKMEM_TINY_MINIMAL_FRONT
# define HAKMEM_TINY_MINIMAL_FRONT 1
#endif
#ifndef HAKMEM_TINY_STRICT_FRONT
# define HAKMEM_TINY_STRICT_FRONT 0
#endif
// ------------------------------------------------------------
// Phase 4-Step3: Tiny Front PGO Config Box
// ------------------------------------------------------------
// HAKMEM_TINY_FRONT_PGO:
// 0 = Normal build with runtime configuration (default, backward compatible)
// Configuration checked via ENV variables at runtime (flexible)
// 1 = PGO-optimized build with compile-time configuration (performance)
// Configuration fixed at compile time (dead code elimination)
// Eliminates runtime branches for maximum performance.
// Use with: make CFLAGS="-DHAKMEM_TINY_FRONT_PGO=1" bench_random_mixed_hakmem
// Expected benefit: +5-8% improvement via dead code elimination (57.2 → 60-62 M ops/s)
#ifndef HAKMEM_TINY_FRONT_PGO
# define HAKMEM_TINY_FRONT_PGO 0
#endif
// Route fingerprint (compile-time gate; runtime ENV still required)
#ifndef HAKMEM_ROUTE
# define HAKMEM_ROUTE 0
#endif
// Bench-only knobs (default values; can be overridden via build flags)
#ifndef HAKMEM_TINY_BENCH_REFILL
# define HAKMEM_TINY_BENCH_REFILL 8
#endif
#ifndef HAKMEM_TINY_BENCH_REFILL8
# define HAKMEM_TINY_BENCH_REFILL8 HAKMEM_TINY_BENCH_REFILL
#endif
#ifndef HAKMEM_TINY_BENCH_REFILL16
# define HAKMEM_TINY_BENCH_REFILL16 HAKMEM_TINY_BENCH_REFILL
#endif
#ifndef HAKMEM_TINY_BENCH_REFILL32
# define HAKMEM_TINY_BENCH_REFILL32 HAKMEM_TINY_BENCH_REFILL
#endif
#ifndef HAKMEM_TINY_BENCH_REFILL64
# define HAKMEM_TINY_BENCH_REFILL64 HAKMEM_TINY_BENCH_REFILL
#endif
#ifndef HAKMEM_TINY_BENCH_WARMUP8
# define HAKMEM_TINY_BENCH_WARMUP8 64
#endif
#ifndef HAKMEM_TINY_BENCH_WARMUP16
# define HAKMEM_TINY_BENCH_WARMUP16 96
#endif
#ifndef HAKMEM_TINY_BENCH_WARMUP32
# define HAKMEM_TINY_BENCH_WARMUP32 160
#endif
#ifndef HAKMEM_TINY_BENCH_WARMUP64
# define HAKMEM_TINY_BENCH_WARMUP64 192
#endif
// ------------------------------------------------------------
// Helper enum (for documentation / logging)
// ------------------------------------------------------------
typedef enum {
HAK_FLAG_BUILD_RELEASE = HAKMEM_BUILD_RELEASE,
HAK_FLAG_DEBUG_COUNTERS = HAKMEM_DEBUG_COUNTERS,
HAK_FLAG_DEBUG_MEMORY = HAKMEM_DEBUG_MEMORY,
HAK_FLAG_REFILL_OPT = HAKMEM_TINY_REFILL_OPT,
HAK_FLAG_P0_BATCH = HAKMEM_TINY_P0_BATCH_REFILL,
HAK_FLAG_BOX_REFACTOR = HAKMEM_TINY_PHASE6_BOX_REFACTOR,
HAK_FLAG_NEW_3LAYER = HAKMEM_TINY_USE_NEW_3LAYER,
} hak_build_flags_t;
#endif // HAKMEM_BUILD_FLAGS_H
Reference in New Issue View Git Blame Copy Permalink