Your speed-up might not save the energy you think.
Teams treat speed as a stand-in for energy, but the two do not move together. On llama.cpp, GreenCode research cut GPU energy by 4.7% while speed rose only 3%: energy and time told different stories.
GreenCode researchers found that speed gains and energy savings are not the same thing, so they measure both.
GreenCode helps tackle this by running a load test built from your system's own endpoints. It records energy as well as speed, and ranks the hotspots by the energy they use. More on this.
Your speed-up might not save the energy you think. Myth or fact: faster code always uses less energy? Myth. Energy is power multiplied by time. Finish sooner but draw more power, and you can still use more. More threads, higher clock speeds and busier memory all draw more power. GreenCode researchers tuned a widely used AI program. Its graphics card used nearly five percent less energy. But it ran only three percent faster. Speed and energy told different stories. Here is the fix you can use today. Measure energy directly, not only how long it takes. Compare before and after, on the same machine, under the same load. Report both numbers. A change can be faster and hungrier, or slower and leaner. Measure it with tools like GreenCode's green-languages, the Constitution profiler, or Linux perf. GreenCode helps tackle this by running a load test built from your own system. It records energy as well as speed, and ranks hotspots by energy. That was GreenCode Tech Tip number twelve. Follow for the next tip, and find more on greencode.ai/tech-tips.
Share this
