AMD Versus Intel, Watts The Difference In Mobile Power?
![AMD Versus Intel, Watts The Difference In Mobile Power?](https://pcper.com/wp-content/uploads/2020/11/6427570.jpg)
Zero To 51W In No Time!
It seems you have an interesting choice when picking a laptop powered by either a Ryzen 4000 series or by a Tiger Lake processor. Some research from Intel has demonstrated that the two architectures behave quite differently for the first ten seconds or so of a heavy load. Their take on it is that AMD is throttling performance for the initial processing, but as it turns out Intel is also playing with the CPU power load at the start of a task. This means you can use this information to decide which you would prefer when picking up a new laptop.
At a press briefing, Intel successfully demonstrated that when unplugged a Zen 2 laptop delays ramping power and voltage up to their maximum states for up to 11 seconds after a workload begins. This results in a ~30% drop in performance for short tasks on these 15W cTDP laptops, just like the slides demonstrate. This is only true for shorter benchmarks, as after that delay the AMD processor finally does get full power and can run at it’s rated frequencies. It can even outperform an Intel powered laptop on a much larger workload which can make use of all the cores.
Intel went the other way, starting big and then backing down to normal. Ars Technica measured the laptop running at the full PL2 power limit of 51W, with temperatures hitting 98°C during the first ~10 seconds of a Cinebench R23 run. After that initial burst the chip settled down to a more reasonable 34W and the temperature dropped to a much more comfortable range. This initial burst of power helps Intel on short workloads but not so much on longer ones.
Armed with this information you can now ponder what your usage would benefit more from. Do you want short sprints of exceptional performance, with the drawback of increased fan noise while it is bursting with power and a hit to your battery life? Would you benefit more from long running multithreaded performance, with much longer equivalent battery life and lower temps to keep your fan speed under control?
If you heavily lean towards one or the other of these scenarios then it will behoove you to pick the architecture which matches your style. We will be keeping an eye out for changes as well, as it is possible that one or both companies may respond to these findings and change how their products perform.
In an embargoed presentation Friday morning, Intel Chief Performance Strategist Ryan Shrout (who dat?) walked a group of tech journalists through a presentation aimed at taking AMD's Zen 2 (Ryzen 4000 series) laptop CPUs down a peg.
More Tech News From Around The Web
- The GIMP turns 25 and promises to carry on being the FOSS not-Photoshop @ The Register
- Bet You Didn’t Know Arduinos Are Psychic @ Hackaday
- Intel Shuts Down Its AR/VR Volumetric Capture Stage @ Slashdot
- Windows Insiders give thanks for latest release… Hang on. Where’s the 64-bit Intel on Arm stuffing @ The Register
- ‘Smart’ Doorbells For Sale On Amazon, eBay Came Stocked With Security Vulnerabilities @ Slashdot
- How to turn your Xbox Series X/S into an emulation powerhouse @ Ars Technica
- QNAP QGD-1602P 2-Bay NAS and POE Switch @ TechPowerUp
Intel conveniently left out that this is configured by the OEM and is not a trait inherent to the architecture. Oops.
Have you read the article on Extremetech about this? https://www.extremetech.com/extreme/317657-intel-is-spreading-fud-about-supposedly-huge-ryzen-4000-performance-drops-on-battery
Good stuff by Ryan as usual, but I’m not sure they covered themselves with glory.
When I’m on battery I don’t expect full performance and from independent testing I’ve not seen anyone complaining about slow response on AMD systems.
The AMD CPU governor makes more sense, the ramping fan noise in TL would be quite annoying.
Even more desirable is let the client choose the CPU governor, even if hidden in advanced settings somewhere. In rooted Android phones you can do just that.