Yes, generally cheaper CPUs tend to be more power-efficient.
Yes, generally cheaper CPUs tend to be more power-efficient.
The question explores how performance and power consumption relate to processor speed and graphics limits. It considers whether a less powerful CPU is more energy-efficient compared to a more powerful one, and if restricting game frame rates affects GPU power use. The discussion highlights the trade-offs between speed, efficiency, and energy usage in modern hardware.
A CPU with fewer cores tends to consume less power when idle. At moderate workloads, usage remains stable, while under heavy load, the lower-end model may draw more current than a higher-end one delivering the same performance (efficiency drops at peak). It’s unlikely to be useful to forecast these situations in real life, as results depend on the specific program and its core utilization.
Above is a chart I made previously by testing and 8066k and 3600 at default and setting power limits. I should have titled it better, but the vertical axis isn't performance, but relative performance/watt. Note the 3600 at stock runs at 88W power limit, NOT the 65W TDP it is rated at. For both CPUs, there is a clear gain in efficiency (work done per power) as you set lower power limits. My mobo runs the 8086k at unlimited power limit by default. This is NOT considered to be an overclock by Intel, but would be by AMD. Tests run at the time were Cinebench R15 and Prime95 with a small-ish FFT. Generally speaking, it takes more proportionately more power to increase performance a given amount. Not actual numbers, but for example, relative to some operating condition, you might need 120% more power to reach 110% performance. There's two parts to this. If you take efficiency as low power, and don't care about performance, maybe. If you take efficiency as performance/watt, then it depends on the software and how well it can make use of more threads. Assuming we have software that scales well with threads, like Cinebench, then if you adjust different core count CPUs to give the same multi-thread score, the higher core count models will probably have an advantage. In other words, 4 cores running at 4 GHz is less power efficient than 8 similar cores running at 2 GHz. There are some other factors to consider, like if the rest of system power is significant compared to CPU alone. So optimising for CPU only may make system power efficiency worse. This is a common error to make. Similar applies as for CPU. The 2080 has more "cores" and they can do the same work at lower clock than the 2060. There are more complications here, in that VRAM might also start to play a bigger factor in power consumption. Original thread: