GLMark2

GLMark2
Name	GLMark2
Title	GLMark2
Latest release version	2.0.0
Programming language	C++
Operating system	Linux, FreeBSD, Android
Genre	Benchmarking software
License	MIT

GLMark2 GLMark2 is an open-source 3D graphics benchmarking tool developed to evaluate OpenGL and OpenGL ES performance on a range of hardware and software stacks. It produces repeatable workload-driven scores that help compare graphics drivers, GPU architectures, and compositor effects across devices and distributions. The project is frequently used alongside other profiling and testing tools in performance labs and continuous integration environments.

Overview

GLMark2 runs a series of scene-based tests exercising shaders, texturing, fragment processing, and buffer management to produce quantitative metrics. It is intended for use by system integrators, distribution maintainers, and hardware vendors to compare results across platforms such as those targeted by Android (operating system), Ubuntu (distribution), and FreeBSD. The benchmark yields numerical scores used by developers at organizations like Intel, NVIDIA, and ARM Ltd. to assess driver changes, and is often cited in reports from projects such as Mesa (computer graphics), Wayland (display server protocol), and X.Org Server.

History and Development

GLMark2 originated as a successor to earlier micro-benchmarks aligned with the needs of open-source graphics stacks. Its development intersected with contributions from communities around Mesa (computer graphics), maintainers of the X.Org Server, and engineers at hardware vendors including Intel Corporation and NVIDIA Corporation. The project saw iterations that paralleled major shifts in graphics APIs and compositors, including the rise of OpenGL ES, the adoption of Wayland (display server protocol), and performance tuning efforts related to releases of Linux kernel graphics subsystems and driver stacks. Contributors have included individuals affiliated with distributions like Debian and Fedora Project and organizations such as Collabora.

Features and Benchmarks

GLMark2 provides a suite of scenarii that exercise pipeline stages: shader compilation, fragment lighting, texture filtering, buffer swaps, and framebuffer operations. Scenes are designed to reveal differences among implementations such as those in Mesa (computer graphics), proprietary drivers from NVIDIA Corporation and AMD, and mobile GPU stacks from ARM Holdings partners. Tests include heavy use of GLSL shaders relevant to shader compilers in projects like LLVM and drivers influenced by work in DRM (Direct Rendering Manager). The benchmark supports customizable scene presets and repeatable run modes useful to teams at Canonical (company) and performance groups at Google for regression detection.

Usage and Command-Line Options

GLMark2 is invoked from a terminal and accepts options to select the rendering backend, windowing system, and list of scenes to run. Typical deployments in test farms integrate GLMark2 with continuous integration tools used by Travis CI historically and current systems like Jenkins (software). Command-line options allow specifying backend interfaces compatible with EGL, selecting windowing targets such as X.Org Server or Wayland (display server protocol), and toggling features for headless rendering useful in environments run by companies like Red Hat and SUSE. Test automation often combines GLMark2 invocations with profiling tools from Perf (Linux) and tracing utilities linked to SystemTap.

Implementation and Architecture

The codebase of GLMark2 is primarily written in C++ and interfaces with platform-specific loader libraries for OpenGL and OpenGL ES via EGL, GLX, or platform wrappers. Its architecture separates scene description from rendering execution, enabling scene reuse and extension by contributors from projects like Khronos Group and vendors including Broadcom. The framework interacts with GPU kernel interfaces provided by the Linux kernel's DRM subsystem when running on platforms using drivers maintained in the Mesa (computer graphics) project. Rendering paths are modular so tests can run under compositor environments such as Mutter (software) and Weston (compositor).

Results Interpretation and Scoring

GLMark2 reports per-scene frame rates and an aggregated score reflecting overall throughput; these numbers are used to compare relative performance across hardware generations like those from Intel Corporation, AMD, and ARM Holdings licensees. Interpreting results requires understanding driver optimizations in stacks like Mesa (computer graphics), the effect of compositor latency in Wayland (display server protocol) sessions, and power management behaviors influenced by ACPI. Analysts often correlate GLMark2 output with low-level counters exposed by vendor tools such as Intel Graphics Performance Analyzers or proprietary profilers from NVIDIA Corporation and AMD to attribute bottlenecks.

Platform Support and Compatibility

GLMark2 runs on a variety of Unix-like systems and embedded platforms, with builds available in package repositories of distributions including Debian, Ubuntu (distribution), and Fedora Project. Mobile and single-board computer deployments commonly use GLMark2 on devices running firmware stacks from vendors such as Broadcom, Qualcomm, and NVIDIA Corporation's Tegra line. Compatibility depends on support for EGL/GLX and shader language versions implemented by drivers, which are maintained by projects like Mesa (computer graphics) and vendors such as ARM Ltd. and Imagination Technologies.

Category:Benchmarking software