Enable SIMD optimizations by default with auto-detection #982
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This commit enables SIMD optimizations automatically based on CPU capabilities, providing significant performance improvements for JSON string parsing without requiring manual configuration via --with-sse42 flag. Key changes: 1. Simplified extconf.rb for auto-detection: - Automatically tries -msse4.2, falls back to -msse2 - No user configuration needed - works out of the box - Removed unnecessary platform-specific logic 2. Enhanced simd.h with unified architecture detection: - Defines HAVE_SIMD_SSE4_2, HAVE_SIMD_SSE2, HAVE_SIMD_NEON - Provides SIMD_TYPE macro for debugging - Uses compiler defines for cleaner conditional compilation - Priority: SSE4.2 > NEON > SSE2 > scalar 3. Added SSE2 fallback implementation: - Uses SSE2 instructions available on all x86_64 CPUs - Provides SIMD benefits even on older processors - Uses bit manipulation for efficient character matching 4. Updated parse.c to use new SIMD architecture: - scan_string_SSE42() for SSE4.2 capable CPUs - scan_string_SSE2() for older x86_64 CPUs - Automatic selection at initialization Performance: - Equivalent performance to baseline with --with-sse42 - All tests pass (445 runs, 986 assertions, 0 failures) - SIMD now enabled by default without any flags 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <[email protected]>
This commit improves SIMD performance by processing 64 bytes per iteration with prefetching and branch hints for better CPU utilization. Optimizations: 1. Process 64 bytes (4x16-byte chunks) per iteration instead of 16 2. Prefetch next cache line with __builtin_prefetch() 3. Load all chunks before comparing (better instruction-level parallelism) 4. Add __builtin_expect() branch hints (matches are unlikely in long strings) 5. Applied to both SSE4.2 and SSE2 implementations Performance improvements (50K iterations): - Strings with escape sequences: 8.3% faster (0.166s -> 0.152s) - Long strings (~2KB): 3.8% faster (0.145s -> 0.140s) - Short strings: 0.8% faster (1.945s -> 1.929s) All tests pass: 445 runs, 986 assertions, 0 failures 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <[email protected]>
Use only compiler-provided __SSE4_2__ define for SIMD detection. The old OJ_USE_SSE4_2 macro is no longer needed since we rely on compiler flags (-msse4.2) which automatically define __SSE4_2__. This simplifies the code and removes legacy configuration.
ohler55
approved these changes
Nov 24, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
This PR enables SIMD (Single Instruction, Multiple Data) optimizations automatically based on CPU capabilities, providing performance improvements for JSON string parsing without requiring manual configuration.
Previously, users had to pass
--with-sse42during gem installation to enable SIMD. Now it's enabled by default and automatically detects the best instruction set for the CPU.Performance Improvements
Benchmarked on:
Results (50,000 iterations):
Key Win: Best improvements on strings with escape sequences (most common real-world scenario).
Changes
1. Simplified extconf.rb (4 lines)
Before: Required
gem install oj -- --with-sse42After: Just
gem install oj- SIMD enabled automatically ✨2. Enhanced simd.h
HAVE_SIMD_SSE4_2,HAVE_SIMD_SSE2,HAVE_SIMD_NEON#ifdefbased conditional compilation3. Optimized SIMD String Scanner (parse.c)
SSE4.2 implementation (modern x86_64):
__builtin_prefetch()__builtin_expect()SSE2 fallback (older x86_64):
Testing
✅ All tests pass: 445 runs, 986 assertions, 0 failures, 0 errors
✅ Clean builds verified
✅ Proper baseline comparisons done
Breaking Changes
None. This is a pure improvement that maintains full backward compatibility.
Benefits
Development Process
This PR was developed with Claude Code AI, which assisted with:
Related Issues
Addresses user requests for automatic SIMD enablement and improved default performance.
🤖 Built with Claude Code
Co-Authored-By: Claude [email protected]