When we were reviewing the handful of newer AMD Chromebooks earlier this year, I noticed some strange benchmark scores and simply chalked them up to benchmarks being benchmarks. After all, if a device feels snappy and gets through my standard workload without breaking much of a sweat, I feel confident that performance isn’t really an issue. This scenario was absolutely the case with the recent review of the Lenovo ThinkPad C13 Yoga Chromebook for me. I generally spend my days plugged into a docking station and with power coursing through my Chromebook, so it never occured to me to consider speed differences when unplugged.
A few times during my review period, however, I remember remarking how snappy things felt and then being a little let down with the performance on benchmarks. There were a few times, however, that those same benchmark scores came back FAR higher and left me feeling like there was simply an issue with the AMD processors, Chrome OS and synthetic benchmarks. Since I never really felt the slowdown in my day-to-day, I didn’t see it as an issue. However, a week or so after my review period was finished, I was checking benchmarks again for some unrelated reason and I finally realized why I was experiencing such back-and-forth with speeds on my AMD-powered Chromebooks. And it all comes down to some strange throttling happening on these chips that is evident in multiple devices made by different manufacturers.
AMD throttling is seen in every benchmark
Once I realized the issue, I set out to test some things. The short version goes like this: plugged in, the Ryzen 5 3500C performs in a similar fashion to the 10th-gen Core i5 in most benchmarks. On battery, the story is drastically different, bringing scores down to something more like a Pentium Gold chip from last year. It’s a big dip in performance, and to be fair, I tested the same scenarios on Intel and MediaTek devices and there were no perceived drops in performance when those Chromebooks moved from plugged-in to battery-only. Just take a look at the scores I’m getting in 3 major benchmark tests:OCTANE:
- Plugged-in: 39,900
- On Battery: 23,700
- Plugged-in: 99.8
- On Battery: 63.4
- Plugged-in: 112.6
- On Battery: 72.742
Again, Intel and MediaTek devices don’t see any real changes in performance with the same testing setup. This is clearly an issue and, yes, I’ve submitted feedback. So far I’ve not seen any action on this or heard anything from Google about why this is an issue, but it is clearly a problem. I fully understand a bit of restraint when a Chromebook is on battery, but the performance drop-off we’re seeing here isn’t good. It makes these highly-capable processors feel like budget CPUs and it’s not a good look for the new kids on the block.
What can you do if you have one of these Chromebooks? Well, just know that plugging in will make performance far better and cross your fingers that Google is on the job. Just as an added bit of help, I’ve submitted feedback again as I’m writing this and I’d recommend you do the same if you are seeing the same slow downs we’re experiencing. Just hit SHIFT + ALT + I and fill out the short form to explain what you are seeing. This is not how Chrome OS is built to behave and we’ve never seen this on any processor prior, so I’m inclined to label this as a bug and the sooner Google is made aware, the sooner they’ll work on a fix. We’ll update as soon as we hear more about this issue.