Software engineers at AMD are about to embark on a major change in their development environment. They spend just 40 percent of their work hours on developing and testing new code. On top of that, they commit another 10 percent to learning and implementing new technologies. AMD has one goal: make life easier for the software developers. They’re using new generative artificial intelligence (AI) tools that make them more productive and efficient.
AI’s integration arrives as AMD faces a crucial imperative to expand its long-standing Adrenalin graphics-software package. Each component in the suite undergoes a strict and iterative development cycle. That’s typically six months to the first release, followed by a further three to six months spent entirely in the support phase. Software-centric engineering Nearly half of AMD’s engineers are dedicated solely to software development. The company culture strongly emphasizes making the most productive use of their development resources.
Current Workflow and Time Allocation
In the day-to-day of a software engineer at AMD, the workday is spent in the majority on these four fundamental pillars. Most importantly, they set aside 20 percent of their time to triaging and debugging issues—that’s the magic pixie dust in keeping the software quality high. On top of that, engineers spend an entire fifth of their hours reworking and cleaning up the code they’ve already produced.
There’s that pesky documentation, which takes up just over 10 percent of their time. This helps make sure that developers who might come after you are able to quickly read and build upon the code you’ve written. I know that you all have spent a lot of time doing these things. Unquestionably, any progress made here would be a huge benefit.
Engineers are involved in different phases of software development at AMD. At peak development and test, there are 60 engineers on the project and this drops to 20 engineers during the support phase of the project. The design phase takes in 10 engineers’ worth of input, and the definition phase has 5 engineers working on it. This IPC-focused, organized mindset further emphasizes the collaborative approach to advanced software development on display at AMD.
Impact of AI on Functional Testing
Delivering AI tools in the functional testing stage represents a significant leap for AMD. These tools are able to reduce the time spent on manual testing by 15 percent. This automation allows engineers to focus on more value-added, innovative work that requires human expertise and creativity. AI increases the scenarios you can run by 20 times. This improvement greatly increases the general robustness of the software.
This increase in testing capacity speeds up the development cycle, producing higher quality software. With less time spent on manual tests, engineers can apply their skills to more productive tasks. This change makes for a far more productive and streamlined process.
Once we use these AI tools in every step of the SDLC, only then will we see an inevitable increase in team-wide productivity. In reality, this increase may be even larger than 25 percent! This ambitious goal is a testament to AMD’s desire to use technology as a tool for ongoing, systemic evolution.
Future Prospects for Software Engineering at AMD
With AMD’s increasing focus on AI driven solutions, these software engineering teams are in a great position for the future. The anticipated productivity boost will not only optimize current processes but allow engineers to spend more time on innovative projects and learning new technologies.
AI tools are rapidly becoming table stakes for engineers’ workflows. This fuels their ambition to take on the most intricate of challenges and continuously expand the leading edge of what is possible in software development. The effect is much more than increasing productivity. It represents a cultural change at AMD, one that is open to adopt and leverage bleeding-edge technologies to fuel innovation.
