Heating and CPU 100%

Hello,
I have a question about CPU 100% usage. I know that 100% CPU usage can cause heat. However, I read some arguments that heat can also cause 100% CPU usage. The reason is when a computer gets dirty, the CPU can overheat before reaching its potential maximum power, and hence throttle itself to, say 50%. In this case, if a program requires 60% of the CPU power, it will use all of the available CPU's power, reaching 100% usage in the Task Manager.
It makes sense to me, but I'm not sure if it really does. Sorry if it's a basic question. I'm totally noob to computers.

Thank you.

 

No, that is not how it works.

The CPU needs to actually draw that power (amps) before it can heat up to it's rated throttling temps. If it is only drawing the amps for 50% CPU utilization, it will only heat to that degree (and note that it is not linear... 50% CPU utilization will not be drawing 50% of the amps drawn at 100% CPU utilization... it will be closer to 40% or less).

When dirt/dust and clogged vents and fans make the CPU run hotter at anything less than it's design limit, it will simply provide that less than 100% performance but at the heat output as if CPU utilization is closer to 100%.

If and when it surpasses it's throttling temps, then it will throttle itself even at a 50% work load to lower the temp to a safe level.

Task Manager is not smart enough (today) to know any of the above things going on with the chassis/fan or the CPU (except for what it is told by that same CPU). Therefore, what it will report is what it can. Namely (Windows 10) 50% Load @ 3.4GHz (for eg.). If it throttles, it may show something like 90% Load @ 1.2GHz, but that is hardly using all (or even 90%) of the CPU's performance at that point.

Hope this helps?

 

Thank you for a fast and brilliant answer. Just for a clarification, I think I understand the first and second paragraph. For the last paragraph, when I look at the Task Manager -> Performance (tab), it said, for example, i7-4700mq @ 2.4 GHz. Then if the graph shows 100% CPU utilization, it means the computer is using all 2.4 GHz. Now suppose there's a heat problem in my computer, does it say i7-4700mq @ 1.5 GHz to mean a thermal throttle, and the graph in this case shows the percentage of 1.5 GHz? Am I understanding it correctly?

Also, can I look at Clock Multiplier in the CPU tab of CPU-Z to know if it's throttling?

Thank you again.

 

In Task Manager/Performance/CPU (Windows 10), 100% @ 2.4GHz doesn't mean anything by itself.

What load is put on the computer to get that indicated CPU utilization?

In my workflows, I know that certain work loads take the system to an indicated 50% and some other work loads take the system to an indicated 80% or higher. What determines (to me) if the CPU is giving me all it can for the specific work load, is how long it takes to actually finish the task.

In your example, if TM was showing 100% @ 1.5GHz it could mean a cooling problem. Or, it could simply be the specific work load presented to the system too. I wouldn't know how to differentiate simply by looking at the graphs.

When TM shows us temperatures along with the rest of the info it does already, we might have a better chance of taking a better guess of if or why we have an issue.

If you really want to know if your system is throttling; use TS.

See:
http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/

 

PS: I got it now. Thank you very much.

PS: I want to give each of your answers reputations, but the forum says I should spread them.

 

I didn't look at the graph. Fair warning:

Heat is proportional to voltage. You run Prime95, CPU voltage increases to it's set maximum. Voltage = heat, heat gets transfered to the cooling solution [heat sync] and is expelled.

When your heat sync and fans are clogged with dust, and you start Prime95, the CPU voltage increases to it's set maximum. Because the cooling solution cannot properly transfer the heat away from the CPU, the CPU hits it's thermal throttling temperature, at which point it drops the voltage to the CPU.

Removing that voltage from the CPU causes the CPU to generate less heat. Less heat means less workload for the cooling solution, resulting in lower [or in this case within thermal specification] temperatures.

 

Get HWinfo64 and show the temps... Do a normal session of what you do and the take a screenshot of what the sensor shows..

 

Thank you, but it's not happening on my computer. My laptop is good. I asked the question because I read an explanation as stated in my #1 post. I'm not sure if it's correct because it made sense to me. I asked for my desire to learn and understand.

I'm good with the answer in post #4 now.

Thank you again for your time.

 

Pretty sure heat isn't proportional to voltage? It is more exponential or logarithmic instead.

If the max voltage is 1.5V (eg.) and the processor is at it's max TDP temp (eg. 105C) at a specific load (L), then when it is at idle load and 'sipping' power at 1/1000th it's max performance, it is not then using 1/1000th of 1.5V. It will be around the 1V mark or maybe at 0.8 or so.

Exponential/logarithmic. Definitely not proportional (to voltage, frequency, work performed or any other relevant metric).

 

It is definitely NOT a linear relationship.

 

This can be true in extreme circumstances. A real world example is a Q9200 quad core in my P79xx. Running at stock vCore the CPU under full load on 4 threads will throttle the CPU to 1.2 GHz. At that lowered speed and load it will report 100% CPU load even at the diminished performance level. Even though dirt is not causing the heat the system just does not have enough to cool the CPU.

The same system at a lowered vCore will allow 2.6 GHZ under full load without topping 85C with 4 thread 100% load. So cooling can matter, in this circumstance the cooling being the lowered vCore. Just remember this is not a common circumstance.

 

The 'work' the CPU can do at 100% @ 1.2GHz is not remotely close to the 'work' it can do at 100% @ 2.6GHz.

While TM may show 100% CPU utilization for both scenarios, that 100% is on a movable scale...

The need for lower vCore voltage on a Q9200 is an anomaly (in design). Most modern CPU's are not that much over voltage to begin with.

As you said; not a common circumstance.

 

Even under a q9100 the system with prime95 on all four cores overheated causing thermal slowdown. While slowed down it would still show 100% load at the TM while agreed doing much less work then before the thermal slowdown. It is more an anomaly of the system not being able to cool the quad cores. I could run a x9100 at 4.0 GHz with no slowdown though.

I just wanted to point out the author while being correct needed to point out the unusual circumstances for it to show. A blanket statement that is always the case is not only wrong but confusing to say the least.

 

In post two and later I have already alluded to that fact.

I think your post is adding to the confusion? Maybe it's just me?

No problem, I can see we are saying the same thing in the end.

 

Thank you for your answers. I understand that it is a rare condition for heat to play a role in CPU 100% load. I just wondered if it is possible or not. Now I have the answer.

 

You'd get a better answer if you install Hwinfo64 and see the temps... You obviously want to ignore my good advice to do this so good luck..

 

I think the question is not is this his issue, at least that is how I took it. The question is more that is this possible. The short answer is yes but it is a rare occurrence. You aer correct in that no matter what software to monitor the system is needed to diagnose the issue on a specific system.