Optimized execution across CPUs, GPUs, and multiple instruction sets.
Backend-specific optimization with runtime compilation.
Extend UHCR with custom kernels, compiler passes, and runtime plugins.
Fast search, responsive layouts, and offline-ready docs.
UHCR compiles a custom intermediate representation (IR) into native machine code while automatically selecting the optimal backend for your hardware.
JIT Compilation Trace Python functions and compile them to native instructions
Hardware Detection Automatic CPUID, GPU discovery, and cache-aware execution
Backend Flexibility CUDA, AVX2, AVX512, and generic CPU support
Install UHCR:
pip install uhcr
Create a compiled function:
import uhcr
@uhcr.jit(eager=True)
def compute(a, b):
return (a + b) * 2
result = compute(10, 11)
print(result) # 42
Get up and running with installation and first examples.
Learn how to trace and compile Python functions with UHCR.
Browse every module, class, and supported backend API.
Explore UHCR runtime, compiler, and storage design.
Automatic backend selection and IR optimizations tuned for real workloads.
Custom plugin support allows new backends and compiler passes.
Responsive guides, search, and offline-ready content.
Detects CPU, GPU, and platform details to choose optimal execution paths.
Browse comprehensive guides, reference material, and hardware documentation that help build fast, reliable applications.