---
icon: material/racing-helmet
---

The 10x Engine [transforms](https://doc.log10x.com/run/transform/) log/trace events into well-defined, typed [TenXObjects](https://doc.log10x.com/api/js/#TenXObject) to efficiently report on, regulate and optimize in edge and cloud environments. 

### :material-racing-helmet: Optimization Model

The 10x Engine employs a **JIT optimization** model inspired by Chrome V8’s [_intelligent design_](https://v8.dev/docs/hidden-classes#why-have-hidden-classes%3F){target="\_blank"} principle. Rather than treating each log event or JavaScript object as a "bag of properties," both engines assumes only a finite number of object shapes will emerge and that these shapes will follow stereotypical usage patterns.

Because neither JavaScript objects nor log events are defined or compiled ahead of time, both V8 and the 10x Engine infer structure dynamically at runtime by "listening" for patterns, gradually recognizing shared object shapes. Once identified, V8/10x assign cached _hidden classes_ to TenXObjects to enable optimized, structured access.

To learn more see [**TenXObjects**](https://doc.log10x.com/run/transform).

### :material-cogs: AOT Compiler

To dynamically create class and metadata information at run time, the 10x [AOT Compile](https://doc.log10x.com/compile/) pipeline scans GitHub and container repos ahead-of-time to generate [symbol library](https://doc.log10x.com/run/symbol/) files, similar to package artifacts (e.g., npm, .jar, .NET assembly).

The 10x [run](https://doc.log10x.com/run/) pipeline utilizes symbol library files to assign a shared [TenXTemplate](https://doc.log10x.com/run/template/) (i.e., _hidden class_) to input log/trace events with the same [structure](https://doc.log10x.com/run/transform/structure), creating a cached optimized schema for each event type.

Operating on typed objects enables direct access to [symbol](https://doc.log10x.com/run/transform/structure/#symbols) and [variable](https://doc.log10x.com/run/transform/structure/#variables) values at runtime without repeatedly parsing JSON structures or evaluating complex, brittle regular expressions for each event.