Question How to tell which component is failing CPU or GPU?

Gamefreaknet

Commendable
Mar 29, 2022
336
15
1,685
Hi. I've had an Alienware M17 with:
i7 8750H
RTX 2070MQ
16GB (2x8GB 2667mhz) RAM
(new) (Samsung 970 Evo Plus 1TB + 250GB)
However across all games there are moments where the frames go from consistent 200fps (or any consistent framerate) to really unstable under 30fps for a few seconds before returning back to a consistent 100+/200+ stable framerate. The drops are infrequent but happen at least once a day. I can tell it's not thermal throttling since my CPU on all cores peaks at 90c (most cores avg around 85c) and my GPU barely ever tops 60c and very rarely 65c.
The laptop itself however is old and I did swap out 2 previously failed drives with the Samsung drives I currently have installed. Is there any way I can figure out which component is underperforming/failing?
 

Ralston18

Titan
Moderator
@Gamefreaknet

And look in Reliability History and Event Viewer for error codes, warnings, and even informational events that occurred just before or at the time the framerates became unstable.

Also use Task Manager and Resource Monitor to observe system performance via an open window to the side. Use both tools but only one tool at a time.

Determine what the system is doing or trying to when the frame rates change.
 

Gamefreaknet

Commendable
Mar 29, 2022
336
15
1,685
Have you checked the power supply to the laptop? That could be overheating and causing your performance issues.
I do find the power supply block (which is rather big) gets quite hot... however is there a way to fix this/stop it getting so hot or is it just a sign that my laptop might be nearing its "end"?
 

Eximo

Titan
Ambassador
Replace the power supply, particularly with a larger one.

Assuming that is the actual problem.

You need to rule out software issues as well before pointing fingers at the hardware. If the temperatures of the internal components are fine as you say, then it is something else.
 
  • Like
Reactions: helper800
Run HWmonitor.
It will record the current, minimum and maximum temperatures on all cores.
What you are looking for is one core, usually the master core, that shows 100c. in red.
It means that that one core throttled at least for a little bit.
If it took too long to recover the temperature you would experience what you described.

Check your cooling airways and verify that the cpu cooling fans are working.
If this is a new phenomena, you might want to repaste the cpu cooler.
In time, some pastes will dry out and become less effective.

If all else fails, try
changing the windows balanced power profile advanced functions to a max of 90% instead of the default of 100%
You may not notice the reduced cpu performance.
 
  • Like
Reactions: helper800

Gamefreaknet

Commendable
Mar 29, 2022
336
15
1,685
@Gamefreaknet

And look in Reliability History and Event Viewer for error codes, warnings, and even informational events that occurred just before or at the time the framerates became unstable.

Also use Task Manager and Resource Monitor to observe system performance via an open window to the side. Use both tools but only one tool at a time.

Determine what the system is doing or trying to when the frame rates change.
what code should I look into on event viewer? I have found that consistently there has been a critical error when I've been on a game (Rocket League) at least twice each time however to "diagnose" what is going on I am not sure which code ID I should look up?
 

Ralston18

Titan
Moderator
I start with the code(s) that occur the most frequently. Usually I click to see the details but the details may or may not be helpful.

Could be multiple codes as well.

What critical error codes are listed?

Another thing that I do is look for error patterns in Reliability History's timeline. Some pattern in the occurances, some constant predecessor. May not be absolutely consistent but something that constantly occurs is a likely clue.

Generally I rely on Microsoft's website for error code meanings. Googling any given error code is likely to result in numerous hits - many being apps and tools with "fixes". Often those websites will show up no matter what code is being researched and many resort to registry edits.

Be very wary of any software "tools" that may be being offered. Especially downloads. Some websites may flash red warnings, etc. trying to panic you into calling someone. System may lock up in the process and a hard shutdown is needed to recover.

Registry edits are a last resort and should be attempted only after a full system backup including the registry itself.

Reading back, the power supply per @Eximo and @geofelt is a likely suspect.

Increasing numbers of errors and varying errors would be something I would look for.