Cross-Platform Performance Telemetry Overlay for Windows, Linux, and macOS.
TelemetryX is a high-performance, near-zero overhead performance telemetry overlay designed for enthusiasts, gamers, and developers. It unifies hardware metrics from NVIDIA, AMD, Intel, and Apple Silicon into a single, canonical performance model.
- Near-Zero Overhead: Built with C17 and lock-free SPSC ring buffers for minimal CPU/GPU impact.
- Vulkan-First Architecture: Low-latency rendering with fallbacks for Metal and OpenGL.
- Unified Telemetry Schema: A single model for over 80+ hardware and software metrics.
- ML-Assisted Normalization: Uses lightweight on-device embeddings to map disparate vendor counters to a canonical schema.
- Advanced Bottleneck Attribution: Heuristic analysis for GPU, CPU, VRAM, and I/O stalls.
- Platform Agnostic: First-class support for Windows (WMI/PDH), Linux (sysfs/proc), and macOS (Mach/Metal).
Comprehensive documentation of the project structure and subsystems:
- Architecture Overview: docs/architecture.md
- Telemetry Engine: docs/telemetry-engine.md
- ML Normalization: docs/normalization-pipeline.md
- Graphics Abstraction Layer: docs/overlay-renderer.md
- Injection Strategy: docs/injection-strategy.md
- Roadmap: docs/roadmap.md
TelemetryX is a closed-source project. This repository serves as the public documentation hub, community issue tracker, and release portal. The proprietary implementation is maintained in a private core repository.
- Bug Reports: Please use the Bug Report Template.
- Feature Requests: Open an issue using the Feature Request Template.
- Performance Issues: Submit a Performance Report.
TelemetryX is strictly offline-by-design. It contains no networking code and never "phones home." All telemetry data is processed and stored locally on your device.
See Privacy and Data Handling for more details.
This documentation and the TelemetryX public-facing project are licensed under the MIT License. See LICENSE for details.