CPU Throttling on 15 inch Surface Book 2 after 1709?

[wynand32]

^ Do voltages stay at that point for a while. I'd worry if that is sustained voltage though since it's still high. Normally, @ 1.351 V, a good overclocking i7 desktop chip would've hit at least 5 GHz. No reason for a 3.8 - 4 GHz overclock to reach 1.351 volts

Also, if you don't mind, what would be your idea of a proper max voltage?

 

[theveterans]

^ A good voltage would be about 1.15 volts maximum sustained and about 1.2 volts peak IMO.

You can check out some of the voltages here: Top 15 Highest frequencies for Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz - CPU-Z VALIDATOR

 

[wynand32]

^ A good voltage would be about 1.15 volts maximum sustained and about 1.2 volts peak IMO.

You can check out some of the voltages here: Top 15 Highest frequencies for Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz - CPU-Z VALIDATOR

Okay, so, I've had CPUID results that look a lot like in the screenshot that's in Seneleron's post (CPU Throttling on 15 inch Surface Book 2 after 1709?). There seems to be a spike to 1.3v+ (sometimes) right as I slide the power setting to high performance, and then it drops back down.

While it's under load (e.g., Intel XTU), it's running close to what you describe, around 1.15 +/-, and then it peaks at most around 1.25v. For example, I just ran an XTU stress test after restarting CPUID (with the slider already to the right), and it peaked at 1.239v.

Mine doesn't ever get hotter than 90 deg. C, and it passes the IPDT. So I'm not sure what to think at this point.

 

[theveterans]

Sustained load is what matters. 1.239 peak max and 1.15% load is 100% OK to me. BTW, even gaming doesn't subject the CPU to extreme test like XTU or IPDT so you're perfectly fine.

 

[wynand32]

Sustained load is what matters. 1.239 peak max and 1.15% load is 100% OK to me. BTW, even gaming doesn't subject the CPU to extreme test like XTU or IPDT so you're perfectly fine.

I appreciate all of your insights. One more note: I ran a couple of other tests, and there were a couple of occasions where it peaked at 1.3v+ and I saw it happen. But again, sustained loads were closer to around 1.15v. So, sounds like mine is okay even if it peaks that high?

 

[wynand32]

Considering that it remained stable at that load through the test is a good sign. I really wouldn't worry about it considering that you'll never reach those levels in real world use. Outside of stability, longevity of the chip is the only other concern and there doesn't appear to be anything abnormal to report.

If you're really worried, amazon 4 year extended warranty with accidental damage is 150 bucks on 2000-2500 dollar laptop. Return yours and get it through them. I got it just because I'm going to be editing a lot of video on this thing and it's in no way user serviceable.

Again, thanks for all of your help on this one! It's a huge investment for me and naturally I want my $3400 notebook to be perfect. I'll give that Amazon idea some thought...

 

[theveterans]

I appreciate all of your insights. One more note: I ran a couple of other tests, and there were a couple of occasions where it peaked at 1.3v+ and I saw it happen. But again, sustained loads were closer to around 1.15v. So, sounds like mine is okay even if it peaks that high?

Yep. Those peaks are about a millisecond long right? Anyways, I believe it's still a firmware issue where if all of the power savings are unlocked the UEFI/BIOS have a tendency to overclock it to its limits for a single millisecond. As long as the sustained load is at 1.15 to 1.2 volts at maximum turbo boost, it's perfectly fine since that's what I get too with the Surface Pro i7-7660U. The difference is that the Surface Pro's UEFI doesn't have an issue where the voltage spike up to 1.3+ volts when the slider is moved to best performance. Also, if I recall correctly, the OP had high sustained voltages and not just peak voltage (which causes instability) during the stress tests so that's the reason his SB2 keeps failing those test.

 

[Seneleron]

Yea, the CPU is a little weird. I can't prove this, but I SUSPECT the 1.3+ v temps were from single core turbo clock. Mine doesn't pull that much voltage during XTU either [around 1.25 +/-]. I only ever see 1.3+ during single core gaming [HoTS mostly]. I meant to go more into depth earlier about that, but the bat phone was ringing off the hook all day ; )

 

[theveterans]

I only ever see 1.3+ during single core gaming [HoTS mostly]

If you can run prime 95 torture test on 1 thread to see if voltages does spike to 1.3+

 

[wynand32]

Yep. Those peaks are about a millisecond long right?

Exactly, they happen so quickly that unless I'm literally focusing on a test for the entire run I tend to miss the spike. It's never been the case that it's hit that level and stayed there.

 

[Mbernert]

I purchased the 15 inch model yesterday and has this exact same problem. I took it back to the store and they exchanged it without any problems, except after getting home I see that this unit has the exact same issue. Either there is a bad batch of them, or I think every 15 inch model has the same problem.

Has anyone with the 15 inch model actually tested theirs to see what cpu temps are like? I am able to use the machine, in fact I’ve tried running multiple stress tests just to see if it would crash or catch fire, and it kept running, even though hwinfo and Intel xtu said all 4 cores were pegged at 99-100c.

 

[theveterans]

I purchased the 15 inch model yesterday and has this exact same problem. I took it back to the store and they exchanged it without any problems, except after getting home I see that this unit has the exact same issue. Either there is a bad batch of them, or I think every 15 inch model has the same problem.

Has anyone with the 15 inch model actually tested theirs to see what cpu temps are like? I am able to use the machine, in fact I’ve tried running multiple stress tests just to see if it would crash or catch fire, and it kept running, even though hwinfo and Intel xtu said all 4 cores were pegged at 99-100c.

Not normal since it should withstand a full CPU stress test without reaching 100 oC. This means you can't video edit a full 2 hour length video without thermal throttling your CPU.