intel/opencl-intercept-layer

Intercept Layer for Debugging and Analyzing OpenCL Applications

What it solves

This tool provides a way to debug and analyze the performance of OpenCL applications without requiring any changes to the application's source code or the underlying drivers.

How it works

The Intercept Layer acts as a middleman between the OpenCL application and the real OpenCL ICD loader. It masquerades as the loader (or occasionally as an implementation) to intercept OpenCL calls. Once intercepted, the tool can modify these calls or pass them through to the actual OpenCL implementation as is.

Who it’s for

Developers and performance engineers who need to inspect, modify, or profile OpenCL-based software.

Highlights

  • Zero Modification: No need to rewrite application code or update drivers.
  • Kerenel ISA Inspection: Ability to dump and disassemble kernel ISA.
  • Injection Capabilities: Support for injecting modified programs, buffers, and images.
  • Performance Analysis: Integration with VTune, Chrome tracing, and MDAPI performance metrics.
  • Capture and Replay: Ability to capture and capture and replay single kernels.

Related

  • Project
    KhronosGroup/OpenCL-ICD-Loader

    The official Khronos OpenCL ICD Loader allows applications to interface with multiple OpenCL hardware implementations without linking to a specific vendor driver.

  • Project
    Mouseww/anything-analyzer

    Anything Analyzer is an Electron‑based desktop app that captures HTTP/HTTPS traffic from browsers, desktop apps, CLI tools, scripts, mobile devices and IoT via an embedded Chromium CDP engine and a built‑in MITM proxy. Captured sessions are fed to a configurable LLM (OpenAI, Anthropic, etc.) which automatically produces reverse‑engineering, security‑audit, performance or encryption‑analysis reports. The tool includes JS hook injection, MCP integration for AI‑agent workflows, SQLite session storage, cross‑platform installers and an MIT‑style UI.

  • Project
    uccl-project/uccl

    An efficient, portable GPU communication library that provides high-performance collectives, P2P transfers, and expert-parallel communication across heterogeneous GPU and NIC hardware.

  • Project
    mlc-ai/web-llm

    A high-performance in-browser LLM inference engine that uses WebGPU to run language models locally without a server.

  • Project
    starbaser/ccproxy

    A transparent network interceptor for LLM tooling that allows for the inspection, transformation, and cross-provider routing of AI API traffic.