XPS 9530 overheating (?) issue while gaming

Hi,

My XPS 9530 acts weird while gaming. After launching CPU/GPU intensive games (like Fallout 4, Skyrim from Steam store in my case) a game runs smoothly as expected for a while but then FPS drop to around 10 for a while (a minute or so) and then it comes back to normal. I've been researching for a while about this and the most common issue was laptop overheating, so I started monitoring CPU / GPU temperatures. CPU maxed at around 81C and GPU around 82C. Notice the graph below (red - GPU, blue - CPU) - around 17 minutes of smooth gameplay and stable temperatures and then FPS drop:



Do you have any idea what is happening? What else can I check? Are those CPU / GPU temps too high? I've been reading people having > 90C GPU temp and it's normal while gaming.

 

It's called throttling, you will find threads on here for the 9550/9560 and they apply to the 9530.

 

I believe it's power throttling due to the VRMs of the CPU or/and GPU getting overly toasty.

 

May be - I'm trying to figure out what it is. My GPU temp max is 82C (with temperature limit reported by GPU-Z at 97C). The only perfcap I see is "therm" but it doesn't match game slowdowns. I've tried Throttle Stop, disabled turbo and tried undervolting (under -50mV system wasn't stable) but without success.

What is the most interesting is that I can see power limit throtling in XTU while stress testing but I can't see it while gaming but still having those slowdowns ...

Stress test:



In game (see temperature drop? This is when slowdown occurred, but no throttle applied):



I'll be researching some more, but if you have any suggestions - I'll be grateful.

 

You're not showing CPU & GPU clocks.
Check also for another program running in the background using Task manager or Process explorer.

 

Ok - I'll retest with the clocks. XTU shows only iGPU clock right?

 

It isn't just the CPU/GPU that cause the throttle, after gaming for a while it overheats the ambient sensors and throttle everything back a you saw.

 

Use HWinfo64, shows everything.
> It isn't just the CPU/GPU that cause the throttle, after gaming for a while it overheats the ambient sensors and throttle everything back a you saw.
I guess XTU would show power limit throttling, even if one has no clue what's the cause.

 

Makes sense - but how to check that? How to check what is the reason of the throttle? System event, XTU, any other tool ?

 

Power Limit throttling occurs due to the VRMs overheating. It's a common thing on the 9550.

 

I thought so too, but can't see any Anyway - I'll retest everything again with XTU throttles, cpu/gpu clocks, HWinfo64 and background processes and get back to you with my findings. Thanks!

 

Power limit throttle can happen due to one of many sensors going over the temp limit or a combination of many getting close will cause it to lock it to the lower multiplier.

You cannot check, this is hardware based throttle via the BIOS and the EC doesn't communicate this to software.

Not much more to say on this apart from take the back off and clean out the dust bunnies, repaste the CPU/GPU. If you want to consider that you will see the thermal threads dotted about this forum section for the 9550/9560 as a guide.

This will never be a gaming spec laptop just like the 9550/9560

 

I know it's not a gaming laptop therefore as Sunday gamer I'm fine with lower fps and lower quality, but not with total fps drops without any good reason.

Some more research. CPU clocks an voltage seem to be pretty stable:



Those freq jumps started right after I alt-tab to desktop, not when in game slowdowns occurred. Next step - the GPU and ...



And voila! GPU clock dropped from 941MHz to 135MHz. GPU temp before it happend was constant 80-81 and afaik it's not that hot ... so why it happens? :O I tried to do some NVIDIA tuning using NVIDIA Inspector but Performance Level 2 (P0) setting "priorize temperature" is fixed to 80C and I can't increase that ... Is it hardcoded in vBIOS? Is there any way to change it? Am I wrong assuming it's too low?

 

I repasted both cpu and gpu. Everything came back to normal. No lags, no slowdowns. Thanks for your help. One thing that still interests me a lot is that temps readings didn't change that much. I still can see GPU going up till 81C (no that often as it used to) but this time no throttling ... Anyway - case solved.

 

Because the heat is escaping via the heatsink instead of into the board causing other sensors to hit the throttle, you will see more sensors using other apps like hwinfo64 or aida64