[SOLVED] Is my CPU supposed to reach its advertised TDP?

[Qattos]

Hello everyone

I have a couple of questions, I have a Lenovo Y520 laptop with an I5-7300HQ with an advertised TDP of 45W and its configurable TDP-down is 35W, I undervolted it using ThrottleStop and it's working fine (unless I just got used to its hiccups) but as I'm monitoring its temps and wattage using HWMonitor, I found that it's only using around 20-ish W at max, I looked up TDP and have a general idea of its meaning and people say that as long as the CPU is using less than the advertised TDP it's fine but I'm afraid my CPU is using way less than it's supposed to and I can't help but wonder if there are any performance benefits if it reached its full power capacity?My max temps under load are around 60-ish and I believe if it used more wattage it will produce more heat but I'd love to know if it's possible to feed it more power?

Thank you in advance.

 

[Barty1884]

TDP is only one aspect. In a lot of cases, CPUs can actually draw more than it's advertised TDP -- see almost any modernish desktop CPU review.

In this instance, the max clock speeds are 3.5GHz, so if you can do so with a lesser power draw, that's not a bad thing.
Increasing the power draw is not going to allow your CPU to boost higher, it's just going to generate more heat.

As an aside, HWMonitor is not always 100% reliable for a lot of aspects. Gives you a good idea, sure.... but when things don't add up, I'd sanity check those results against programs like HWInfo64.

I believe the true "power draw" in HWMonitor is a sum of a few aspects, not strictly just the 'package', but I could be wrong.
In this instance, if your 100% load was a sum of package + IA Cores, you'd be drawing ~39W. That seems more reasonable to me, for a chip with a configured TDP at 35W.

 

[CompuTronix]

... I found that it's only using around 20-ish W "at max" ...

"at max" is a HUGE variable, among many others. Please narrow that down for us a bit.

Power consumption (Watts) is driven by workload, which, depending on what software you're running, varies wildly, so it's highly important to be very specific. TDP can only be reached on certain processors under specific conditions.

What exactly is "max"?

CT

 

[Qattos]

TDP is only one aspect. In a lot of cases, CPUs can actually draw more than it's advertised TDP -- see almost any modernish desktop CPU review.

In this instance, the max clock speeds are 3.5GHz, so if you can do so with a lesser power draw, that's not a bad thing.
Increasing the power draw is not going to allow your CPU to boost higher, it's just going to generate more heat.

As an aside, HWMonitor is not always 100% reliable for a lot of aspects. Gives you a good idea, sure.... but when things don't add up, I'd sanity check those results against programs like HWInfo64.

I believe the true "power draw" in HWMonitor is a sum of a few aspects, not strictly just the 'package', but I could be wrong.
In this instance, if your 100% load was a sum of package + IA Cores, you'd be drawing ~39W. That seems more reasonable to me, for a chip with a configured TDP at 35W.

Thank you so much, your remark about the sum of package + IA Cores answered my question.

 

[Barty1884]

Thank you so much, your remark about the sum of package + IA Cores answered my question.

As I stated

I believe the true "power draw" in HWMonitor is a sum of a few aspects, not strictly just the 'package', but I could be wrong.

I'm not 100% confident in that and, in all honesty, @CompuTronix knows a heck of a lot more than I do. If you share, as requested, specifics regarding the actual workload being performed, you'll get a more definitive answer.

 

[Qattos]

"at max" is a HUGE variable, among many others. Please narrow that down for us a bit.

Power consumption (Watts) is driven by workload, which, depending on what software you're running, varies wildly, so it's highly important to be very specific. TDP can only be reached on certain processors under specific conditions.

What exactly is "max"?

CT

I used Cinebench and I usually play CSGO, I even used to test with Prime95 but I forgot how much power my CPU was drawing at that time, I guess I'll let it run tonight and see how much it will draw.
Thanks for your contribution.

 

[Qattos]

As I stated


I'm not 100% confident in that and, in all honesty, @CompuTronix knows a heck of a lot more than I do. If you share, as requested, specifics regarding the actual workload being performed, you'll get a more definitive answer.

Yes he's right, I will run Prime95 overnight and report back.

 

[CompuTronix]

Qattos,

Prime95 is the correct utility for running a thermal test, which only takes 10 minutes, so there's no need to run Prime95 overnight, unless you're using "Blend" for a stability test. Also, with respect to being specific ...

Which version of Prime95?

Which torture test?

Small FFT's?

With or without AVX?

Games, apps, streaming, rendering, transcoding and most utilities have partial, fluctuating workloads with fluctuating Power consumption and fluctuating Core temperatures that are not well suited for testing thermal performance.

“Stress” tests vary widely and can be characterized into two categories; stability tests which are fluctuating workloads, and thermal tests which are steady workloads. Prime95 v29.8 Small FFT's (AVX disabled) is ideally suited for testing thermal performance, because it conforms to Intel's Datasheets as a steady-state 100% TDP workload with steady Core temperatures. No other utility can so closely replicate Intel's thermal test workload.

However, although your i5-7300HQ shares the same microarchitecture design and TDP specifications as it's i7-7920HQ counterpart with Hyper-Threading, only the flagship processor should reach its rated TDP value.

For more detailed information, please read Section 11 - Thermal Test Basics:

Intel Temperature Guide - https://forums.tomshardware.com/threads/intel-temperature-guide.1488337/

CT