Media and Entertainment, Gaming
Activision Accelerates Game Development With NVIDIA vGPU
Objective
Activision, the global leader behind the Call of Duty franchise, streamlines its game development pipeline by building a revolutionary global testing and deployment platform with NVIDIA Virtual GPU (vGPU) technology to accelerate deployment cycles, enhance stability, and optimize testing across every stage of the game’s lifecycle.
Customer
Activision
Use Case
Data Center / Cloud
Products
NVIDIA vGPU
NVIDIA RTX GPUs
NVIDIA RTX Virtual Workstation
Key Business Impact
- Achieved an 82% reduction in data center footprint and a 72% decrease in power consumption.
- Faster validation before code commits and broader test coverage enabled over 250,000 tasks daily, reducing continuous integration delays and boosting team velocity across 3,000+ developers.
- Reduced infrastructure fragmentation, streamlined global management, and scaled testing across 500+ servers with less downtime and complexity.
A New Era for Game Development
In the high-stakes world of multi-platform video game development, graphics fidelity and technical performance are just as important as compelling gameplay and story. To address the increasing demand for testing, Activision developed Compass, an automated system designed to evaluate and validate game code, game maps, release builds, and branch maintenance. This system operates on Activision’s continuous integration and continuous development (CI/CD) infrastructure.
Initially designed for single-studio development, Compass quickly faced three key challenges as Call of Duty expanded into a multi-studio operation with complex environments:
- Stability: Ensuring high code quality and promptly identifying regressions in a multi-studio codebase is essential. Regression bugs from tool changes, asset pipeline updates, or graphics code modifications must be quickly detected and resolved to prevent them from impacting QA or reaching players.
- Scalability: Supporting thousands of concurrent tests for diverse game modes, platforms, and massive environments like Warzone. The infrastructure needed to accommodate 3,000 to 5,000 compute nodes running continuously, without bottlenecking during peak loads.
- Flexibility: Rapidly reconfiguring environments for varied hardware targets and development needs, ranging from lightweight GPU tasks and heavier GPU-intensive workloads such as offline map compiles, demanding fine-grained resource allocation.
Before implementing NVIDIA Virtual GPU (vGPU), Activision’s CI infrastructure relied on workstation-class systems and small GPU server clusters with a maximum of four GPUs per system spread across sites. These setups worked well individually, but without centralized control, resource sharing was limited, and testing efficiency suffered. Compute resources sat idle in one region while queues built up in another. Developers faced delays, coordination overhead grew, and scaling the system only added complexity. Activision needed a unified infrastructure with dense virtualization, flexible GPU allocation, and centralized management to evolve Compass into a global CI/CD backbone.
“The biggest trend we see is scale… We went from 6v6 multiplayer maps to 4x4 kilometer environments with 150–200 players. That’s an enormous change in scale, and an enormous change in testing.”
Michael Vance,
SVP at Activision
Transforming Activision’s Robust, Scalable CI/CD Solution
Activision revolutionized its CI/CD infrastructure with NVIDIA vGPU technology and NVIDIA GPUs, enabling a robust, enterprise-grade pipeline. This architectural overhaul turned Compass into a powerhouse, handling more than 250,000 tasks daily for 3,000 developers. Key features of the new solution include:
- High-Density Virtualization: 4–24 virtual machines (VMs) per server with NVIDIA RTX™ Virtual Workstation profiles and a minimum of 8 GB VRAM per vGPU. This allowed Activision to map compute profiles more efficiently to specific task demands, eliminating resource waste and enabling parallel execution of thousands of jobs.
- Centralized Management: Streamlined global operations and resource allocation. A fleet of NVIDIA vGPU-powered virtual workstations provided consistent performance and monitoring across on-prem and hybrid cloud environments.
- Flexible Resource Partitioning: Optimized for diverse testing needs, from lightweight tasks to heavy offline computations. Workloads that previously failed or stalled due to 24 GB memory ceilings could now be processed at scale with higher reliability and less manual triage.
- Enterprise-Class Reliability: Relied on enterprise support to maintain business continuity and avoid critical development and engineering downtime.
This purpose-built infrastructure makes short work of complex tasks, including automated multiplayer validation, visual regression checks through screenshot comparisons, and performance testing across a full range of game modes, maps, and hardware targets. Operating 3,000–5,000 compute nodes across 500 hosts in on-premises and cloud infrastructure, with numerous smaller installations worldwide, Compass delivers the next generation of performance.
Measurable Gains: Activision’s Leap in Productivity and Performance
The rollout of the NVIDIA vGPU-powered infrastructure had an immediate and significant impact on development for Activision:
- Data Center Consolidation: Reduced from 100 individual 1U servers to just six 3U units, slashing rack space requirements by 82%. This consolidation was enabled by high-density vGPU support across fewer hosts, reducing physical infrastructure while increasing capacity.
- Energy and Cooling Savings: Power usage dropped by 72%, and thermal load by 73%, significantly lowering operational costs. These gains reflect both the improved performance-per-watt of NVIDIA GPUs and reduced chassis count in data center environments.
- Enhanced Productivity: Higher vGPU density enabled more concurrent test runners, expanding test coverage and accelerating feedback. Teams were able to run more parallel jobs, improving throughput for CI tasks like Radiant validation, runtime checks, and automated preflight testing.
- Improved Code Quality: Broader test coverage led to earlier issue detection, reducing bugs reaching QA and retail, and elevating the player experience. Faster detection of rendering, asset, and performance regressions allowed engineering teams to resolve issues before they impacted release builds.
By embedding a smart debugging system, scalable test infrastructure, and real-time performance analysis deep into the development pipeline, Compass accelerated by NVIDIA vGPU didn’t just help meet expectations—it helped redefine them.
“With the changes we’ve made around the GPU project, we’ve drastically reduced the amount of physical space our GPU chassis took—from around 100 racks down to about six, through the density improvements with the NVIDIA GPU solution. We’ve also seen enormous improvement in power draw efficiency and cooling in our data centers, which directly translates into savings and reduced maintenance overhead.”
Michael Vance
SVP at Activision
Redefining Game Development Pipelines at Enterprise Scale
Activision’s journey demonstrates how NVIDIA vGPU technology, along with NVIDIA GPUs, including NVIDIA RTX A5000, RTX A6000, L40, L40S, and L4, can transform distributed and siloed infrastructure into a unified, efficient, and scalable enterprise environment. Organizations seeking similar results can leverage NVIDIA’s solutions to modernize their infrastructure and drive innovation at scale. With plans to adopt RDMA-based compute-memory fabric and NVIDIA® ConnectX® adapters, Activision is continuing to push the boundaries of CI/CD infrastructure performance by reducing latency and improving throughput for large map builds and asset compiles. Close collaboration with NVIDIA ensures Activision can tune its fleet and adopt new platforms as developer needs evolve.
Learn more about NVIDIA vGPU solutions.