How much faster is FastMM4-AVX compared to the original FastMM4?

In multi-threaded benchmarks, FastMM4-AVX can be over twice as fast, with tests showing up to 69.75% speed improvement in NexusDB scenarios. However, single-threaded performance may have minimal gains or even slowdowns due to AVX overhead.

Does FastMM4-AVX work with FreePascal on Linux?

Yes, it supports FreePascal/Lazarus on Linux 64-bit, with version history noting fixes for compilation and AVX-512 support. It has been tested with Lazarus 1.6.4 and later versions for cross-platform compatibility.

How to disable AVX in FastMM4-AVX to avoid slowdowns?

Define DisableAVX, DisableAVX1, DisableAVX2, or DisableAVX512 in your project to turn off AVX instructions. This is recommended if your CPU has transition penalties or reduced turbo frequency, as per the README's advice.

What's the difference between FastMM4-AVX and FastMM5?

FastMM4-AVX focuses on AVX optimizations and efficient synchronization for multi-threaded gains, while FastMM5 may perform better in some single-threaded or high-contention scenarios. Benchmarks in the README show mixed results, with FastMM5 sometimes faster in multi-threaded tests.

Is FastMM4-AVX safe for production use in critical applications?

It has performance benefits but security gaps: no safe-unlinking and double-free detection only in debug mode. For high-security needs, enable FullDebugMode during testing and consider additional safeguards like CheckHeapForCorruption in production.

Can I use FastMM4-AVX in a DLL without issues?

Yes, but avoid defining UseReleaseStack, as it causes deadlocks during DLL unload on modern Windows. The README explicitly warns against this, recommending alternative synchronization methods or feature disablement.

What alignment should I use for best performance with AVX?

With AVX enabled, blocks are aligned by 32 bytes by default, but this increases memory overhead. If your CPU supports FSRM, you can disable AVX and use 8-byte alignment for better efficiency, as suggested in the README.

Open-Awesome

FastMM4-AVX

GPL-3.0Pascalv1.0.12

A high-performance fork of FastMM4 with AVX/AVX2/AVX512 support, efficient synchronization, and FreePascal compatibility.

GitHub

154 stars24 forks0 contributors

What is FastMM4-AVX?

FastMM4-AVX is an optimized memory manager for Delphi and FreePascal applications, designed to replace the default memory allocator with faster, more scalable performance. It enhances the original FastMM4 with support for AVX instructions and improved synchronization techniques, making it particularly effective in multi-threaded environments.

Target Audience

Delphi and FreePascal developers building high-performance, multi-threaded applications where memory allocation speed and low contention are critical.

Value Proposition

Developers choose FastMM4-AVX for its proven performance gains in multi-threaded scenarios, modern CPU feature utilization, and compatibility with both Delphi and FreePascal compilers without requiring external DLLs.

Overview

Fork of FastMM4 by Pierre le Riche with AVX/AVX2/AVX512 support, efficient synchronization, and FreePascal (FPC/Lazarus) compatibility

Use Cases

Best For

Multi-threaded Delphi applications requiring low-latency memory allocation
FreePascal/Lazarus projects needing a high-performance memory manager
Applications running on AVX/AVX2/AVX512-capable CPUs
Scenarios with high memory allocation contention across many threads
Replacing the default memory manager in Embarcadero Delphi projects
Performance-critical applications where memory copy speed is a bottleneck

Not Ideal For

Single-threaded applications where the advanced synchronization overhead provides no benefit over the original FastMM4
Legacy systems with CPUs lacking AVX/ERMS/FSRM support, as performance gains heavily depend on these modern features
DLL projects that require the UseReleaseStack feature, due to documented deadlock risks during unload on Windows Server 2016+
Projects prioritizing minimal memory overhead, as 32-byte alignment increases padding waste compared to 8-byte alignment

Pros & Cons

Pros

Multi-Threaded Performance

Benchmarks show over 50% speed improvements in high-contention scenarios, such as the NexusDB test with 64 threads running twice as fast, due to efficient pause-based spin-wait loops and optimized synchronization.

Modern CPU Optimization

Leverages AVX1/AVX2/AVX512, ERMS, and FSRM for faster memory copies, with secure register clearing via vxorps/vpxor to prevent data exposure in XMM/YMM/ZMM registers.

Cross-Compiler Compatibility

Supports Embarcadero Delphi and FreePascal/Lazarus on Win32, Win64, and Linux 64-bit, with code safety enhancements like stricter type adherence and support for compiler overflow checking.

Configurable Synchronization

Offers flexible locking via critical sections, WaitPKG instructions, or spin-waits, with compile-time options to tune for specific contention levels and avoid unnecessary bus locking.

Cons

AVX Performance Trade-Offs

AVX instructions can cause slowdowns from AVX-SSE transition penalties and reduced CPU turbo frequency, with the README admitting that benefits are negligible or negative in some cases, advising to disable AVX.

Incomplete Security Features

Safe-unlinking protection is not implemented (as noted in remaining issues), and double-free detection is only available in FullDebugMode, leaving production builds vulnerable to memory corruption attacks.

Complex Configuration

Requires managing numerous conditional defines (e.g., DisableAVX, EnableWaitPKG) and alignment settings, which can be daunting for integration and error-prone without deep understanding.

DLL Deadlock Risks

The UseReleaseStack feature must be avoided in DLLs due to deadlocks during unload, as documented in known limitations, forcing developers to compromise on performance or risk system hangs.

Frequently Asked Questions

Related Projects

FastMM

A memory manager for Delphi and C++ Builder with powerful debugging facilities

Stars493

Forks171

Last commit3 months ago

FastMM5

FastMM is a fast replacement memory manager for Embarcadero Delphi applications that scales well across multiple threads and CPU cores, is not prone to memory fragmentation, and supports shared memory without the use of external .DLL files.

Fast scaling memory manager for Delphi

Stars105

Forks25

Last commit4 years ago

BrainMM

Extremely fast memory manager for Delphi

Stars99

Forks17

Last commit6 years ago

Community-curated · Updated weekly · 100% open source

Found a gem we're missing?

Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.

Submit a project Star on GitHub