Framework 12th Gen i5-1240p Benchmarking

[Joseph_Iacobbo]

Hello Everyone,

I was asked in a separate thread to run some benchmarks on the new 12 gen Framework so I’m starting this thread to post my results. @Second_Coming

Ambient = 76F

Specs:

Geekbench:

Framework Laptop (12th Gen Intel Core) - Geekbench Browser

HWINFO csv:

geekbench_cpu_1.CSV

Cinebench:

HWINFO csv:

cinebench_multi_1.CSV

Let me know if there are any other benchmarks that you all would like to see.

[Second_Coming]

Thanks for the benchmark…

Looks like your system didn’t go into boost while benchmarking. (Max CPU Package Power was only at 30W)

Is it plugged into the wall, and set to Performance power profile?

[Joseph_Iacobbo]

@Second_Coming

Is it plugged into the wall, and set to Performance power profile?

This was run on battery and I believe the power profile was Balanced.

[Second_Coming]

With no boost (only max out at 30w), the score is pretty impressive!

Seems to be inline with the 2 samples from Notebookcheck:

The Yoga Slim in the screenshot is really killing it in benchmarks with the 50W non-boost TDP.

Can you benchmark again with the laptop plugged in, and with Performance profile (when you have time, that is)?

Much appreciated.

[Usernames]

@Second_Coming

Can you benchmark again with the laptop plugged in, and with Performance profile (when you have time, that is)?

Would be interested to se this also, thanks.

[Joseph_Iacobbo]

@Second_Coming

Can you benchmark again with the laptop plugged in, and with Performance profile (when you have time, that is)?

Plugged into my laptop dock and set profile to performance.

Ambient = 79 F

Geekbench:
Results:

Framework Laptop (12th Gen Intel Core) - Geekbench

Compared to previous run:
https://browser.geekbench.com/v5/cpu/compare/16255572?baseline=16257672

HWINFO csv:

geekbench_cpu_2.CSV

Thermal throttled pretty quickly. I don’t really want to run a cinebench at those temps.

[Second_Coming]

Thanks for the update.

So it only boosted to 50w in the Geekbench case. But the score is very respectable. About 10% increase in score from battery (non-boost) to plugged-in (boosted).

[greytrimble]

Very cool! Can you try running Cinebench multithreaded with the 10 minute “throttle” mode (you have to turn on Advanced settings first)? That way we can see the sustained non-turbo performance.

To compare, my 1135G7 scored 5021 multithread on Cinebench R23 10 minute mode with 76F ambient. With it throttling to baseline, that meant it stuck at 28W for that score. You say you don’t really want to run at those temps but FWIW, mine only hit 99-100C briefly during the turbo at the beginning. It hovered around 90C when sustaining 28W. Curious what the 1240P’s temps would be for this too!

[lessthanjoey]

@greytrimble

Very cool! Can you try running Cinebench multithreaded with the 10 minute “throttle” mode (you have to turn on Advanced settings first)? That way we can see the sustained non-turbo performance

Edit: disregard the link below, wrong processor!

Notebookcheck have it covered, looks very good:

Framework Laptop 13.5 Intel 12th gen review: Like the Microsoft Surface, but...

[Second_Coming]

@lessthanjoey

Notebookcheck have it covered, looks very good:

There are mainly two reasons for this thread:

1. This is about the i5-1240p (Notebookcheck has the i7-1260p).
2. Real end-user received unit (as opposed to review unit).

[Usernames]

@Joseph_Iacobbo

I don’t really want to run a cinebench at those temps.

If you want to avoid these temperatures in the future you should disable boost. If the workload is demanding enough the CPU in this system will always hit 100*C and start to downclock to protect itself. I would not worry about it though quite safe. I understand if makes you uncomfortable and you don’t want to do it. Thanks for sharing your tests.

[lessthanjoey]

@Second_Coming

This is about the i5-1240p (Notebookcheck has the i7-1260p).

My bad, totally glossed over that in my haste!

[Joseph_Iacobbo]

@greytrimble

Very cool! Can you try running Cinebench multithreaded with the 10 minute “throttle” mode (you have to turn on Advanced settings first)? That way we can see the sustained non-turbo performance.

What power profile would this be on, Balanced or Performance?

EDIT
I ran it on Balanced

Ambient = 75 F

cinebench_multi_10throttle_1.CSV

[Usernames]

@greytrimble

Very cool! Can you try running Cinebench multithreaded with the 10 minute “throttle” mode (you have to turn on Advanced settings first)? That way we can see the sustained non-turbo performance.

Is this just the same as normal multicore test unless your CPU finished the 5 renders in less than 10 minutes? My CPU still “boosts” (36w) when I tried this.

By the way I have the same CPU as you and have improved my scores from ~5100 points to ~5750 by bending heatsink clamps numbered 1-3 a little.

@Joseph_Iacobbo

What power profile would this be on, Balanced or Performance?

I think the interest from most will likely be “best performance” but your CPU it at 60W PL2 in that screenshot which is the main thing I believe. You also got quite close to peak wattage at 56.2 nice to see. Thanks again for these.

[greytrimble]

@Usernames

Is this just the same as normal multicore test unless your CPU finished the 5 renders in less than 10 minutes? My CPU still “boosts” (36w) when I tried this.

I observe it boosting for a fraction of a second, basically a blip, each time the test restarts. Basically negligible in my opinion as far as the score goes.
What are the 5 renders? I’m only aware of it doing one render again and again until the 10 minutes are up, and then it finishes the render it was doing.

@Usernames

By the way I have the same CPU as you and have improved my scores from ~5100 points to ~5750 by bending heatsink clamps numbered 1-3 a little.

Neat! Have you tried any thermal paste changes?

@Joseph_Iacobbo

What power profile would this be on, Balanced or Performance?

Performance profile please! Although if the PL1 and PL2 are 28/60W on Balanced then I imagine the score won’t be particularly different. Worth a test to see though, if you have a chance!

[Usernames]

@greytrimble

I observe it boosting for a fraction of a second, basically a blip, each time the test restarts. Basically negligible in my opinion as far as the score goes.
What are the 5 renders? I’m only aware of it doing one render again and again until the 10 minutes are up, and then it finishes the render it was doing.

It should boost for up to 28s primarily given it is cool enough and by that I mean anything above 28 watt. Here is a graph I had once planed to send support (only image I currently have available on my phone)

You can see there should be some seconds of boosting.

I meant that Cinebench runs the same render 5 times during the run.

@greytrimble

Neat! Have you tried any thermal paste changes?

Actually the thermal paste provided is very good, on a previous board I lowered my peak boost a little by using MX4. On this board I just bent the brackets a little and it seems to have been effective!

@greytrimble

Although if the PL1 and PL2 are 28/60W

By the way 12th gen has 30/60W PL1 and PL2.

[Jieren_Zheng]

Does anyone know if this chip will continue to boost as long there is sufficient thermal headroom, I know Intel previous generations have a fixed boosting time (less the manufacturer BIOS increased limits).

[greytrimble]

@Usernames

It should boost for up to 28s primarily given it is cool enough and by that I mean anything above 28 watt. Here is a graph I had once planed to send support (only image I currently have available on my phone)

Yeah I observe the same thing as you, including that my 1135G7 slowly climbs and has never been above ~39C.
EDIT: I mean ~39W above.

“given it is cool enough” is why I want the 10 minute throttle mode, because it gives the score of the last run when the system is already not “cool enough” anymore and won’t turbo.

[Usernames]

@greytrimble

~39C

Did you mean watts?

@greytrimble

“given it is cool enough” is why I want the 10 minute throttle mode, because it gives the score of the last run when the system is already not “cool enough” anymore and won’t turbo.

I will have to look into this feature to understand what it does that is different from a normal run but you could disable turbo in the BIOS or start a run and restart it just after boost window is over on the initial run if you want to test without boost.

[greytrimble]

@Usernames

Did you mean watts?

Oh yes watts, haha

@Usernames

I will have to look into this feature to understand what it does that is different from a normal run but you could disable turbo in the BIOS or start a run and restart it just after boost window is over on the initial run if you want to test without boost.

I ultimately want to see what kind of performance this CPU can put out in a constant, long-running workload.

Disabling turbo might be equivalent if a constant workload is assumed to result in the same behavior. It is possible for other scenarios though:

1. Brief and periodic turbo boosts during constant load whenever thermals allow it. In this case, disabling turbo would be leaving a bit of performance on the table.
2. Cooling isn’t actually adequate enough and the CPU has to run at even lower than PL1. Marginally possible here considering there doesn’t seem to be any change to the cooling fan setup since 11th gen. They rate their cooler to be 28W TDP but as you say, the 12th gen uses a 30W PL1. In this case, running the render just once with turbo disabled would not cause the actual throttle issue to appear. If anything, the benchmark should be run even longer than 10 minutes.

These possibilities can only be found out by doing actual tests. Mocking the scenario by disabling turbo, or whatever other tweaks, isn’t quite the same.