25 degrees lower after suspending the PC

[guligan18]

Hi all,

I have a 14700K with a Be Quiet! Loop 2 AIO and I m using MSI MEG Z690 Unify 32Gb RAM 6200 Dominator, and I've been using this PC for about a year. Early on, I ran lots of tests and set up undervolting, since this generation runs hot. During gaming, CPU temps are usually 65–75°C, and I always monitor with Afterburner.


The other day, while gaming, I had to step away, so I suspended the PC for the first time. After about 10 minutes, I resumed, launched the same game (which always ran at ~70°C), and to my surprise, the CPU stayed at 45–55°C all afternoon, never exceeding 60°C. I thought maybe it was just because the system was "cold," but the low temps persisted.


The next day, I tested again: played for 30 minutes (temps normal at 70°C), then suspended, waited 10 seconds is enough, resumed, checked with HWinfo64, relaunched the game, and again saw temps at 45–55°C, with peaks at 55–60°C.


So my question is: what changes in the PC after suspend/resume that causes these lower CPU temperatures, even though performance seems unchanged?
Anyone can try this? and tell me if also temps are less, stable , and persist...

thanks in advance ,

 

[Richj44]

I suspect that the performance is not actually the same. Have you tested by comparing results from a benchmark program like Cinebench before and after suspending the pc?

 

[guligan18]

Hi Again ,

Looking with HWInfo and extracting files form before after suspending and in the middle of a game , I can see this :

Below is a comparison of the average values of the main variables measured in the (Before) and (After) :

Summary of Results​

• Core VIDs (avg) [V]: Significantly decreased after the change, from 1246.45 to 921.38.
• Total CPU Usage [%]: Increased from 270.72% to 305.08%1.
• CPU Temperature [°C]: The average CPU temperature dropped from 64.35°C to 47.79°C.
• Total CPU Power [W]: There was a notable reduction from 96,453 W to 38,427 W.
• Vcore [V]: Decreased from 1182.39 to 886.51.
• GPU Temperature [°C]: Remained almost the same, with a slight decrease.
• GPU Power [W]: Also decreased, from 337,923 W to 314,369 W.

These results show that, after the suspension, most parameters related to consumption and temperature decreased, suggesting an improvement in the system’s energy efficiency and cooling

Comparative Table of Averages​

Variable	Before (Average)	After (Average)	Difference (Before - After)
Core VIDs (avg) [V]	1246.45	921.38	325.06
Total CPU Usage [%]	270.72	305.08	-34.36
CPU Temperature [°C]	64.35	47.79	16.56
Total CPU Power [W]	96,453.28	38,427.45	58,025.83
Vcore [V]	1182.39	886.51	295.87
GPU Temperature [°C]	648.97	645.58	3.39
GPU Power [W]	337,923.30	314,368.94	23,554.37

Conclusion​

The comparison demonstrates a general improvement in system efficiency after suspension, especially in the reduction of both CPU and GPU power and temperature. This may translate into lower energy consumption and greater thermal stability for the equipment1.

Questions:​

looking at the results , which options in BIOS should I change to have "After" details?


Thanks!

 

[citay]

Man, those AI-generated summaries...

Richj44 is right, the performance is the big question mark here. Because if the VIDs are lower and the power draw is lower, it can only mean that the CPU also clocks lower / boosts less. Of course, if a game is firmly limited by GPU performance (which most of them are), and not by CPU performance, then in games, the performance impact may be cushioned to where you may not even notice it. In other words, even with an i5 instead of an i7 CPU, you'd still have the same performance in that game, at a lower CPU power draw.

However, you probably don't want to permanently turn your i7 into an i5 by prohibiting it from boosting higher for workloads that are more CPU-heavy, right? Anyway, the main "normal" way to optimize how the CPU is running is described in my Guide: How to set good power limits in the BIOS and reduce the CPU power draw.

But what's happening here surely has to do with limited CPU frequencies after suspending the PC. So you should take some baseline performance numbers with Cinebench R23, as shown in step 1) of the guide above, then repeat after having suspended the PC. Surely the score will be lower for that second run.

What's your BIOS version?

 

[guligan18]

Yep that summary was extracted after comparing before and after HWinfo Stats which I m absolutelly agree with this .
Anyway After followed that guide never had temps like I m getting after suspended, but I ll try to do some benchmark R23 , but not really interesting on this cause this PC is only for gaming perspectives.
BIOS is last one : 7D28v1N1(Beta version)

Thanks for reply.

 

[citay]

Yes, those lower temperatures after suspending the PC come from some kind of bug/issue where the CPU doesn't boost as high anymore (i'm almost certain). With the method from my guide, you won't restrict the boosting in any way, so it will clock much higher than with that bug. But you will still lower the power consumption (and thus temperatures) compared to default operation.

but I ll try to do some benchmark R23 , but not really interesting on this cause this PC is only for gaming perspectives.

Well, if you are not interested in the CPU's performance, you could've gotten an i3 or i5. With an i7, you want to get the performance you paid for, no? It is highly likely that, with the bug after suspending the PC, the performance will not measure up anymore, because the CPU cannot "magically" run that much cooler than before without also clocking lower. If the CPU is requesting under 1V all of a sudden, it means the clocks will be considerably lower, there is no other explanation. And lower clocks mean less performance. As i said, with that game, you might not notice it.

 

[Richj44]

Hi Again ,

Looking with HWInfo and extracting files form before after suspending and in the middle of a game , I can see this :

Below is a comparison of the average values of the main variables measured in the (Before) and (After) :

Summary of Results​

• Core VIDs (avg) [V]: Significantly decreased after the change, from 1246.45 to 921.38.
• Total CPU Usage [%]: Increased from 270.72% to 305.08%1.
• CPU Temperature [°C]: The average CPU temperature dropped from 64.35°C to 47.79°C.
• Total CPU Power [W]: There was a notable reduction from 96,453 W to 38,427 W.
• Vcore [V]: Decreased from 1182.39 to 886.51.
• GPU Temperature [°C]: Remained almost the same, with a slight decrease.
• GPU Power [W]: Also decreased, from 337,923 W to 314,369 W.

These results show that, after the suspension, most parameters related to consumption and temperature decreased, suggesting an improvement in the system’s energy efficiency and cooling

Comparative Table of Averages​

Variable	Before (Average)	After (Average)	Difference (Before - After)
Core VIDs (avg) [V]	1246.45	921.38	325.06
Total CPU Usage [%]	270.72	305.08	-34.36
CPU Temperature [°C]	64.35	47.79	16.56
Total CPU Power [W]	96,453.28	38,427.45	58,025.83
Vcore [V]	1182.39	886.51	295.87
GPU Temperature [°C]	648.97	645.58	3.39
GPU Power [W]	337,923.30	314,368.94	23,554.37

Conclusion​

The comparison demonstrates a general improvement in system efficiency after suspension, especially in the reduction of both CPU and GPU power and temperature. This may translate into lower energy consumption and greater thermal stability for the equipment1.

Questions:​

looking at the results , which options in BIOS should I change to have "After" details?


Thanks!

I'm sorry I didn't make myself clearer about the performance monitoring. Did you happen to look at the boost clocks for the cores before and after? I'd look there to make sure all the temp and vid improvement isn't because of lower clock speeds. Is anything showing as throttling for any reason? I wouldn't think so, but take a look. Any change to PLL 1 or PLL 2 settings?

The reason I'm asking is there's obviously a huge difference pre and post suspending your system, but some setting(s) has to be changing, it isn't just magically cooling better. If you can find that setting you should be able to fix it in BIOS permanently. Hopefully, anyway. Nothing should be changing just from sleeping or hibernating the system in the first place.

Is your power plan in Windows set to balanced? If not, set that using defaults for the balanced plan.

Edit: posted this reply before I read Citay's, sorry to duplicate everything.

 

[guligan18]

Well, if you are not interested in the CPU's performance, you could've gotten an i3 or i5. With an i7, you want to get the performance you paid for, no? It is highly likely that, with the bug after suspending the PC, the performance will not measure up anymore, because the CPU cannot "magically" run that much cooler than before without also clocking lower. If the CPU is requesting under 1V all of a sudden, it means the clocks will be considerably lower, there is no other explanation. And lower clocks mean less performance. As i said, with that game, you might not notice it.

Well, I paid the same price as an i5 or less, so I couldn't resist. On the other hand, I understand your commutative and logical explanation: more volts more frequency more consumption and temperature, but right now I prefer less temperature and graphics card to do its job and give me the maximum FPS.

I'm sorry I didn't make myself clearer about the performance monitoring. Did you happen to look at the boost clocks for the cores before and after? I'd look there to make sure all the temp and vid improvement isn't because of lower clock speeds. Is anything showing as throttling for any reason? I wouldn't think so, but take a look. Any change to PLL 1 or PLL 2 settings?

I ve already had some settings in BIOS , basically offset negative and PLL1 and 2, and never had low temps , so when this happened my first thought was.. so maybe this is really now applied after suspending PC?

I have attached the HWinfo report, there are 2 tabs before supending and after , if it can help .

I really appreciate your both help.
thanks

 

[citay]

That report doesn't give enough information, since it doesn't seem to list the clock speeds and some other important things. Also, the spreadsheet, as weird as it sounds, is not as good as a comparison method as screenshots. So you should do it like i describe in step 1) of my guide: Set HWinfo to English, and expand the sensors with the <-> arrows on the bottom left of the window, click it twice, so all the sensors are shown side-by-side. Arrange them properly so they can all be read.

Then, run Cinebench R23, let it run for the full ten minutes from a normal boot. Now bring the HWinfo Sensors to the front and take a screenshot of it. We don't need to see the Cinebench window, just the score as text is enough.

Repeat after waking the PC up and triggering the "bug". If HWinfo was kept open in between, reset the sensors first. Then again a full ten-minute run, take note of the score, and show the HWinfo window afterwards.