⚠️ WARNING: POTENTIAL HARDWARE DAMAGE
This tool can cause system instability, crashes, data loss, and permanent hardware damage from overheating.
Requirements: Adequate compute power, proper cooling.
| Module | Target | Description |
|---|---|---|
Integer Arithmetic (3np1.asm/primes.asm) |
ALUs, Branch Prediction | Collatz conjecture & prime factorization |
AES Encryption/Decryption (aesENC.asm/aesDEC.asm) |
Crypto Accelerators | AES-256 XTS mode with key expansion |
AVX/FMA Floating-Point (avx.asm) |
Vector Units | 12-wave FPU torture with dependency chains |
Branch Prediction (branch.asm) |
Branch Predictors | Real-world patterns: gaming, database, compiler |
Cache test (cache.asm) |
L1/L2/L3 Memory Latency | Real-world operations |
CPU Rendering (render.asm) |
Mathematical Units | polynomial & trigonometric |
Disk I/O Stress (diskWrite.asm) |
Storage Subsystem | Multi-pattern 16GB write cycles |
Memory Flooding (flood.asm) |
DRAM & Cache Integrity | L1/L2/L3 cache hierarchy + rowhammer attacks |
LZMA Compression (lzma.module.cpp) |
CPU Compression | Multi-threaded LZMA level 9 + DEFLATE |
SHA-256 Hashing (sha256.asm) |
Crypto Extensions | SHA-NI instruction with multiple rounds |
System Monitor (systemMonitor.manage.cpp) |
Real-time Monitoring | ImGui interface with thermal detection |
avx- AVX/FMA vector unit destructionrender- Mathematical modemem- Memory controller + rowhammer attacksfull- All tests combined
branch- Branch prediction with realistic patterns3np1- Integer arithmetic with unpredictable workloadsprimes- Complex mathematical algorithmslzma- CPU compression workloads
aesenc/aesdec- AES encryption/decryptionsha- SHA-256 hashing performance
disk- Storage subsystem stress testing
| Version | Description |
|---|---|
SIFT (Full)  |
Complete system torture with real-time monitoring |
SIFTCLI  |
Command-line only (deprecated) |
SIFTLIB  |
Static library for integration |
All tests use standardized scoring: iterations/second
- Higher scores = better performance
- Consistent across all test modules
- Real-time monitoring with thermal throttling detection
| Guide | Description |
|---|---|
| 📖 USAGE.md | Complete user guide - how to run tests and interpret results |
| 🔧 BUILD.md | Build instructions for all platforms and configurations |
| ⚙️ PRESETS.md | Configuration file guide - create custom test presets |
| 🔬 TECHNICAL.md | Technical documentation - architecture and implementation details |
# 1. Build SIFT
git clone <repository>
cd SIFT
mkdir build && cd build
cmake ..
make -j$(nproc)
# 2. Run interactive mode
./sift
# 3. Or run a preset
./sift gaming.cfg📖 New to SIFT? Start with USAGE.md for complete instructions.
🔧 Build issues? Check BUILD.md for platform-specific help.
# Quick CPU torture
./sift
> avx
Iterations?: 1000
Lower bound?: 0.001
Upper bound?: 1000000
# Branch prediction testing
./sift
> branch
Iterations?: 5000
Pattern (1=Gaming, 2=Database, 3=Compiler, 4=Mixed): 4
# Nuclear option (all tests)
./sift
> full
Intensity (1 = default): 2Special thanks to the original project, ESST, which inspired the structure and intensity of this enhanced version. (I just make a fork because esst was not portable due to ROCm)
- Real-world correlation: Unlike synthetic benchmarks, SIFT tests predict actual application performance
- Thermal awareness: Automatically detects and reports CPU throttling
- Multi-threaded: Scales across all available CPU cores
- Assembly-optimized: Hand-tuned assembly for maximum hardware stress
This software comes with absolutely no warranty. Use at your own risk.
Monitor temperatures and ensure adequate cooling before running intensive tests.
The authors are not responsible for any hardware damage or system instability.
SIFT v0.9.0 - Don't sue me 🔥💀