Thinkpads Forum

Anyone using tpfc on their X100e?

I have figured out that the only active sensor "aps" is measuring reliably about 14C lower than actual cpu, so you can edit the ini to set an offset of -14 and it works quite well. I also edited the fan settings/temp thresholds for better cooling on AC and longer life on battery (running a little hotter, but I use this with low cpu).

I also figured out that the X100 has four fan speeds. When choosing thresholds as I did, you have fan setting:

0=off
1=low
2=med
3=high
7=max
4,5, and 6 duplicate 1,2,3 for some reason.

One scary note: When testing the limits with Prime95 (cpu stress test), the default BIOS fan speed allows the CPU to go way too hot for my tastes. I stopped the stress test when it climbed above 95C! ...and it was still not running fan at full speed. ...scary. I would not suggest running at max cpu load for extended periods on this little guy.

Anyone here using another fan control tool for the X100?

think2wice wrote:Anyone using tpfc on their X100e?

I have figured out that the only active sensor "aps" is measuring reliably about 14C lower than actual cpu, so you can edit the ini to set an offset of -14 and it works quite well. I also edited the fan settings/temp thresholds for better cooling on AC and longer life on battery (running a little hotter, but I use this with low cpu).

I also figured out that the X100 has four fan speeds. When choosing thresholds as I did, you have fan setting:

0=off
1=low
2=med
3=high
7=max
4,5, and 6 duplicate 1,2,3 for some reason.

One scary note: When testing the limits with Prime95 (cpu stress test), the default BIOS fan speed allows the CPU to go way too hot for my tastes. I stopped the stress test when it climbed above 95C! ...and it was still not running fan at full speed. ...scary. I would not suggest running at max cpu load for extended periods on this little guy.

Anyone here using another fan control tool for the X100?

You are probably loocking at the wrong sensor, the internal cpu core temp is reliable, and the acpi warning threshold is set @85C over that temp the x100e simlpy goes in ibernation.

One thing that has also to be noticed is that there can be a lot of difference between the core0 and the core1 even more than 5C. This is just normal for the amd cpus.

Just use speedfan or coretemp or sensors under linux to figure that.

Btw speedfan is not able to change the fan's speed as happen in many other MB.

Maybe the next version will.

Okay, tried a few software tools and not sure which to believe:

Processor idle, full speed (not adaptive), default fan

Using the terminology from each application, I measured:

CoreTemp: Core#0 = 46, Core#1 = 42

SpeedFan: Core = 42, Temp1 = 52

CPU Thermometer: CPU = 52

Everest: CPU = 55, Core#1 = 68, Core#2 = 65

Speccy: Core0 = 64, Core1 = 67

A little nervous about believing the best case (CoreTemp) when their site does not claim support for Neo X2.

I'm confident I can make tpfc work if I have the right offset, but what is the real temp???

think2wice wrote:Okay, tried a few software tools and not sure which to believe:

Processor idle, full speed (not adaptive), default fan

Using the terminology from each application, I measured:

CoreTemp: Core#0 = 46, Core#1 = 42

SpeedFan: Core = 42, Temp1 = 52

CPU Thermometer: CPU = 52

Everest: CPU = 55, Core#1 = 68, Core#2 = 65

Speccy: Core0 = 64, Core1 = 67

A little nervous about believing the best case (CoreTemp) when their site does not claim support for Neo X2.

I'm confident I can make tpfc work if I have the right offset, but what is the real temp???

I know, it's confusing... the temperatures are exposed in three different ways, and each of this way expose some different values side by side with the correct ones ( which are the core temperatures and the mb sensor).

You can get a more precise idea using linux you will see all this values exposed and you can judge what are the correct ones.

The windows programs often are simpli peeking the values coming from the wrong source, but speedfan is working correctly exposing both the core temp + the mb cpu sensor ant those values are perfectly reasonable: about the same reading with different thermal inertia.

Coretemp look only at the core temp and so can't be wrong.

BTW is not possible that correct value is the higher one, exposed by some windows programs and linux.

Compiling a kernel with the j2 param (dual thread load) will raise the reading well over 100 C, the processor should be melted then, and is not.