Speedometer 2.1 benchmark competition

dhh posted a really interesting Speedometer 2.1 benchmark result on a Framework Laptop 13 with 7040 Series. Neither he nor we have been able to replicate this again though. So, we thought it would be fun to see if anyone in the Community is able to meet or beat this score on any configuration/OS/browser on Framework Laptop 13. The prize for the winning score is… the pride of knowing you have the fastest Framework Laptop 13.

You can access Speedometer 2.1 here: Speedometer 2.1

Please share an image of your results along with what your system configuration, OS (and kernel version if on Linux), browser version, and any other notable information is.

1 Like

Love this idea! It’s become a bit of a sport to get a great Speedometer score out of the PCs I have. This is the older Speedometer 2.1 test, but the competition is ~520 from an Intel 14900K and, I believe, 550 from an Apple M3.

But since that score was recorded, I’ve only been able to hit 300-350. Even on Linux, which is running about 20% faster than Windows normally.

Was wondering whether somehow that Linux run was happening with the zenbleed or other security precautions turned off?

That is crazy - my 1135G7 turns about 275 w/ Chrome 123 or 250 w/ Firefox 124, on Fedora 39 KDE (kernel, 64GB RAM, 1 TB USB storage. No real difference between X11 or Wayland.

Same machine w/ Windows 11 (from SK Hynix 512GB NVMe internal) returns about 283 w/ Chrome 123 and 240 w/ Firefox 124.

My 5800X desktop gets about 365 w/ Chrome 123 and about 285 w/ Firefox 124 on Fedora 39 KDE (

My Steam Deck (512GB oem NVMe) gets about 140 w/ Firefox 124 flatpak in Desktop mode no matter whether it’s locked to 60 fps, TDP limited to 15W or not.

This is on Framework 13 7840U, Arch Linux, running Chrome 123 from AUR (so, not Flatpak / AppImage / Snap).

Kernel 6.8.1-arch1-1, Plasma 6 Wayland, connected to power, with power-profiles-daemon set to “Performance.”

This comes close to my Apple M1 Pro, which hit 480. Pretty impressive! Speedometer 3 was 22.6, compared to 27.1 on the M1 Pro.


Oh, I’m remembering that there is a bug that can cause the power profile to get invisibly stuck after a suspend. So any benchmarking should probably be done right after a reboot, without suspending in between.

Intel Core i7-1260P
16 GB of RAM (2 X 8 GB)
Linux Mint 21.3
Kernel 6.5.0
Chrome 123.0.6312.58
16% CPU usage maximum while running this test

My 5900X desktop with 32 GB of RAM gets 331 +/- 25 with a dozen tabs open and other programs running. Tops out at just over 10% CPU.

That’s an incredible score!! What on earth accounts for such an extreme discrepancy to Ubuntu? Maybe the power profile is somehow broken on Ubuntu? I’m running 22.04, and I still can’t get above 320. You’re posting a score that’s 45% higher. That’s crazy.

Does Arch build Chrome with some special compile flags? It doesn’t just use the Google provided distribution?

I’m willing to run Arch if I can get that kind of performance.

Oh. I just saw the Speedometer 3 score. That’s no faster than what I can get under Ubuntu. Very strange!

I have no idea! This version of Chrome is just pulled out of the official .deb file, so I don’t think there would be a difference from just installing the deb on Ubuntu.

There have been some AMD-specific improvements to power-profiles-daemon and the kernel in just the last few months, so having newer versions of those might help.

It must be kernel optimizations. I just ran SP2.1 on Ubuntu 23.10 and I got 410! Required Performance mode AND being plugged in. But that seems to be the difference. Maybe that original screenshot I got was from 23.10? I’ve been running 22.04, and there I can’t get above 320-350.

Strangely, like you, I don’t see a big jump in SP3.0. That’s very similar between 22.04 and 23.10. Anyway. This settles it. I’m installing 23.10 :grin:

Yep. That did the trick. Upgraded to 23.10, and now I too am seeing 450 on SP2.1. Just awesome. Still fascinating why it barely/didn’t move the needle on SP3.0. Thanks a bunch!