[RESPONDED] Sustained all-core boost clocks on Framework with 1280P lower than I expected

Framework Laptop 13 Linux
1

I got a Framework 12th with Intel i7-1280P. I’m primarily running Fedora 37 but i got a dual boot setup with Windows 11.

When hitting the CPU with an all-core stress test (prime95, compiling stuff), it starts off with around 3.5GHz on all cores (according to htop). After max. 10 seconds, it starts throttling down to 2.2GHz on all cores and stays there. At this point, temps hover around 90°C.

It seems to me that there should be more thermal budget, especially since the fan noise is still relatively reasonable. This is of course nice for working in an office or in public, but I would like to be able to get as much power as possible out of my machine for situations where noise is not a concern. Most of the time, I’m working by myself at a desk with the charger connected (official charger).

I’ve been searching the forums/internet and collected some bits and pieces to try and improve the situation.

echo performance | sudo tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

echo 1 > /sys/class/powercap/intel-rapl:0/enabled
#PL1
echo 64000000 > /sys/class/powercap/intel-rapl:0/constraint_0_power_limit_uw

#PL2
echo 10000000000 > /sys/class/powercap/intel-rapl:0/constraint_1_time_window_us
echo 64000000 > /sys/class/powercap/intel-rapl:0/constraint_1_power_limit_uw

It seems to be mostly placebo. During compile tasks, it does clock a bit higher (I’ve seen 2.7GHz on all cores), but for prime95 torture testing, it seems to do very little.

I’ve tried prime95 on my Windows 11 install, where it actually seems to be clocking even lower.

Is this the expected level of performance? Are the fans not capable of running faster? Am I being limited by power delivery? I have no peripherals connected.

If it is not the expected level of performance, are there any steps I can take to get the most of my CPU? My primary concern is performance in Fedora.

Thank you.

2

Thats about as good as it gets. Yours is throttling a little early, probably a bad factory paste job…not unusual. You still won’t see a much higher average, the various tools for monitoring the frequency and temps don’t do a really good job of showing how much of this is affected by the ecores. Gotta remember those top out at 3.5 boost and their base is 1.3 and in your case you have 8 of those…over half, yet they are figured into the overall average. What are your idle temps like?

1 Like
3

Hey @nadb, thanks for your insights.

I’m of course monitoring the frequency & temps of the cores seperately.

After 1min of prime95:

image
image1286×1101 202 KB

After 1min of idle:

image
image1286×1101 200 KB

Let me know what you think!

4

Yeah a repaste would probably help you out. I am sitting a 42C with an I7-1260p using a Honeywell PTM 7950 thermal pad. Previously with Thermal Grizzly Kryonaut it would sit at 44C, and with the factory paste it would sit at around 48-49C. If you can clear some thermal headroom you should at least be able to extend the duration of the initial boost and squeeze out a little extra on the average freq for the sustained.

Just for refrence these are my numbers:

coretemp-isa-0000
Adapter: ISA adapter
Package id 0:  +41.0°C  (high = +100.0°C, crit = +100.0°C)
Core 0:        +41.0°C  (high = +100.0°C, crit = +100.0°C)
Core 4:        +37.0°C  (high = +100.0°C, crit = +100.0°C)
Core 8:        +36.0°C  (high = +100.0°C, crit = +100.0°C)
Core 12:       +39.0°C  (high = +100.0°C, crit = +100.0°C)
Core 16:       +37.0°C  (high = +100.0°C, crit = +100.0°C)
Core 17:       +37.0°C  (high = +100.0°C, crit = +100.0°C)
Core 18:       +37.0°C  (high = +100.0°C, crit = +100.0°C)
Core 19:       +37.0°C  (high = +100.0°C, crit = +100.0°C)
Core 20:       +36.0°C  (high = +100.0°C, crit = +100.0°C)
Core 21:       +36.0°C  (high = +100.0°C, crit = +100.0°C)
Core 22:       +36.0°C  (high = +100.0°C, crit = +100.0°C)
Core 23:       +36.0°C  (high = +100.0°C, crit = +100.0°C)

BAT1-acpi-0
Adapter: ACPI interface
in0:          17.31 V  
curr1:         0.00 A  

iwlwifi_1-virtual-0
Adapter: Virtual device
temp1:            N/A  

nvme-pci-0100
Adapter: PCI adapter
Composite:    +39.9°C  (low  =  -0.1°C, high = +85.8°C)
                       (crit = +86.8°C)
Sensor 1:     +33.9°C  (low  = -273.1°C, high = +65261.8°C)
Sensor 2:     +36.9°C  (low  = -273.1°C, high = +65261.8°C)

acpitz-acpi-0
Adapter: ACPI interface
temp1:        +42.8°C  (crit = +210.0°C)
temp2:        +42.8°C  (crit = +210.0°C)
temp3:        +42.8°C  (crit = +210.0°C)
temp4:        +39.8°C  (crit = +210.0°C)
temp5:        +29.8°C  (crit = +210.0°C)

As you can see your core temps are a good 10C above mine. I would repaste for sure and I can sincerely recommend those Honeywell Thermal Pads. Absolutely fantastic mess free solution. If you search the forums there is a very detailed thread about it, very informative, it’s what made me buy it, and I am more than satisfied with it.

5

I see. I will definitely look into that. I guess i was hoping it was just a fan control issue, but that should at least provide some improvement.

Thanks for providing some reference and taking the time to help out :slight_smile:

1 Like
6

After your re-paste as recommended above, you can use s-tui to bang on things a bit with some stress testing. Good for some comparable test results post-pasting.

1 Like