12th gen Framework Benchmarking (Please submit your scores)

Given the extreme price difference between the three 12th gen CPU choices, it seems necessary to make a clean comparison.

If you own a 12th gen Framework Laptop, please help out by submitting your Geekbench scores:

How to Benchmark
  1. Download and install Geekbench (don’t use the browser version): Geekbench 5 - Cross-Platform Benchmark
  2. Check Task Manager to make sure your system is at idle (no updates running, etc.)
    Close all significant background apps. You should be somewhere under 3% cpu usage.
  3. Set Power Mode to “Best Performance” (Click your battery icon)
  4. Launch Geekbench, Click Later and “Run CPU Benchmark”
  5. Post the resulting URL here. All necessary information is included by Geekbench. No private information is shared.

Select average of scores from multiple devices (outliers are ignored)
The scores in brackets are the individually submitted results

  • i5-1240p (base price):
    Single-Core: 1560 (1560)
    Multi-Core: 7663 (7663)

  • i7-1260p (+300$):
    Single-Core: 1854 (1759, 1921, 1883)
    Multi-Core: 9685 (9687, 9658, 9711)

  • i7-1280p (+710$):
    Single-Core: 1755 (1755)
    Multi-Core: 10075 (10075)

Its hard to give a verdict without more scores.

1 Like

Quick note! Don’t forget i7 models have a 20% stronger iGPU!!!
The 1280P vs 1260P score seems very sus, I suppose we’ll have to wait for people with actual Frameworks to benchmark those 2!

The 1185G7 in my unit just edged out slightly ahead (very barely) in the single core scores. But I’m guessing the 12th gen procs are burning throw less energy to achieve that.

On Linux, my 1260p is over 9% faster on single-core, multicore about the same as the other test you dug out. Framework Laptop (12th Gen Intel Core) - Geekbench Browser

  • 1921 single-core
  • 9658 multi-core

This was run on a 5.18 a while back. I just did a new run on 5.19.7 and it basically matches Framework Laptop (12th Gen Intel Core) vs Framework Laptop (12th Gen Intel Core) - Geekbench Browser

  • 1883 single-core (98% of prior run)
  • 9711 multi-core (100.5% of prior run)
1 Like

The scores are from actual framework laptops in high performance mode, but benchmark scores vary a lot. Without a few scores from different devices its not possible to give a clear verdict.

I’d avoid comparing 11th and 12th gen Geekbench because its different architectures. Synthetic benchmarks almost always score in favor of a particular architecture.

Notice that my post wasn’t a score submission…I’m merely saying / hinting that those scores in your post look low for 12th gen. i.e. You would expect higher scores. The single core scores from @lhl look inline with 12th gen from the score browser.

Why is my score so low compared to others?..

Set your wifi to “metered connection” to prevent Windows Update from downloading anything. Make sure Windows Update isn’t running. Plug in USB-PD. Start your laptop, start Geekbench (the app, not the benchmarking yet). Make sure it’s in “Best Performance” power mode. Let it idle for 15 minutes. Then run the benchmark.

The scores bumped just a little bit. My cinebench scores are also much lower than the ones on the official blog. I am contacting support to help troubleshoot.

There is always a fair amount of variation between CPUs and it could just be bad luck. Looking at i7-1260p Geekbench scores your score does seem about right.

My scores are even lower than the official 1240P scores. Bad luck for spending $300 more for nothing. Hmmm…

Reach out to support, they may / may not provide thermal paste and / or heatsink replacement.

Performance variation in processors of the same SKU are typically very minimal (the tolerances are very tight on Intel side). The performance difference you typically see is due to external causes…usually TIM or heatsink (much loosely controlled, relatively speaking). These are the two areas where assembly variations are introduced. Specifically, heatpipe assembly quality could induce +/- 10-15% cooling capacity differences.

(The joint on between the copper contact plate and and heat pipe on one end, and between heat pipe and the copper fins on the other end. And there are two heatpipes per assembly)

Try checking temps. That seems like the most likely culprit.

Do you have any example on what are some reasonable temps?
I do notice the fan on my laptop is often very loud for just web browsing and discord.
For my usage, the CPU utility is around 20% but the temp is always above 60C.

I also re-ran Geekbench on a Ubuntu live USB.
The score is just about the same as on Windows 11.

1 Like

You have a very uneven thermal dissipation between the CPU and the heatsink. Try re-seating the heatsink with new paste. (From your screenshot: core max 99c, but average is only 65c) It’s thermal throttling…

If you’re Geekbench numbers match (short benchmark) but your Cinebench (longer benchmark), then your chip is likely fine, but maybe you have a badly pasted/mounted heatsink.

I’d recommend searching for repaste in the forums and you’ll find a bunch of threads w/ people posting their before and after temps.

Results from a 1260p from Linux Mint.
Single: 1859
Multi: 9726

Results from 2x8G i7 1260P running Arch Linux

Single: 1955
Multi: 9449

Hum…looks like the 1260p is about 800-1200 points lower than the leading pack (at 10600+ points):

well, i think there’s something wrong with my 1260p, in Cinebench R23, I score ~6600ish points plugged in (HWMonitor shows a peak CPU power of ~64 watts, usb-c power monitor on the input shows a peak of ~90 watts), on ‘best performance’ and ~8800ish points on battery using balanced (HWMonitor shows peak cpu power of 35 watts). I repasted the CPU today (before the repaste, plugged in was 6300ish points), so at this point i suspect it may be a bad heatsink or something?