Question How do i get my CPU temps down ?

[BossSiggy]

Hello. My old pc died because it got to hot, so recently i bought a new pc and one of the first things i did was checking what temperatures it would reach, because i dont want that to happen again and just by doing a simple task like running Windows full virus scan it was reaching +90 degrees celcius and ive been researching and from what i undertand your CPU shouldnt reach higher that 80 degrees celcius under maximum load, so i contacted the shop where i bought it and Intel about it and Intel told me to update the BIOS (i did that and it didnt fix the problem) they then said that if it didnt fix the problem, both them and the shop told me to change the CPU with a new one of same model (i7-12700F) and get a better CPU cooler instead of the stock one and no matter what i will change the CPU, because i obviously dont want a CPU that there is something wrong with, but id like to save the money for a new CPU cooler for something else, if its possible.

So my question is if there is any way for me to get my CPU down in temperature without changing the cooler, without loosing performance and my pc being to loud, i have set my fans to go to 100 at 50 degrees celcius in BIOS, but i still hit +81 degrees celcius doing virus scan, i can then set my pc in power saver mode and it will only 64 degrees celcius doing a scan, but that takes off way to much performance i think ? id like to have my pc on either balanced or high performance. Also Intel is telling me that i need a cooler with at least 180 watt, but the shop where i bought the pc is saying that BE QUIET PURE ROCK 2 is more than enough (its less than 180 watt though) and im thinking Intel is right, since they made the CPU, so they should know best, but i want to hear other peoples opinion on that aswell.

 

[Phaaze88]

Is this a joke or ? But lets say thats possible, what happens for it to not hit 79 then ? Its slows performance down ? when it hits 79 C then it shuts down or ?

It's not a joke, and it's entirely possible. If seeing 80C core temperature is that serious... if even seeing so much as a spike to that is too much, then just limit it to the value right before 80C.

It doesn't behave any differently - just throttles back 21C lower than usual, and boost right back up once there's even a little temperature headroom.
Unlike Ryzen, the current Intel chips do not care what the operating temperature is, besides the thermal limit.
If it's 30/50/80C, you still get the max turbo boost, whatever that value is in regards to how many cores are active. Even at an alarming 98C, those cores will still boost to their max. That doesn't happen with Ryzen.

There is an unspoken timer on thermal shutdown. If the throttle protections work, you never see shutdown, because the timer gets reset if temperature successfully drops.

 

[Darkbreeze]

You have a 12700F with stock cooler. So what you are getting is temps appropriate for a stock cooler. At PL1, that cpu can hit 165w, at PL2 max turbo it can hit 240w. That's Big Air / 240mm AIO minimum territory and it'd still be warm under an all core max turbo. To get temps down to the 'old Standby safe temp' of @ 70°C, you'd be looking at a 280mm/360mm AIO.

For good max temps, it's advised to do a minimum of 1.5x the wattage. For best temps under those kinds of loads, 2x wattage. Stock coolers are barely over 1.0x, if that much, so uber heavy loads like stress tests, renders, compiles, cinebench or Prime95 can generally have no issues putting a stock cooler to 100°C and throttling.

Stock coolers are designed for manageable temps at nominal use, Photoshop or light gaming or web surfing. Not stress tests or rendering or other excessive loads for excessive time periods.

Your temps are normal for what you have, to get better will require better cooling.

I don't know what you are looking at K, but that's wrong. He has a 12700F. Base TDP for that CPU is 65w. MAX turbo TDP is 180w. That is per the Intel data sheets. In any case, I'd guarantee this is a case and CPU cooler problem. Undervolting isn't a solution, it's just a good way to introduce instability unless you are a fairly well experienced tweaker/overclocker and know exactly what you are doing AND are willing to take the time to go through the highly involved process of verifying stability. And anybody who says you don't need to do that (Yes, I know it's not you) is absolutely somebody you want to immediately put on the "do not listen to" list. If you have to undervolt a locked CPU just to get it to avoid throttling then you have some other issue, and voltage is the least of your concerns.

Also, just because somebody has the same case (Or other hardware) as what is in question in any given situation does not automatically mean they "know" what they are talking about or that they are correct in their line of thinking. It also doesn't automatically mean they don't either. You don't have to own a case, or cooler, or other piece of hardware, to have a very good idea about it's capabilities and quirks if you are somebody with a good grasp of the relevant fundamentals and at least a moderately good amount of experience over the years with a variety of hardware. Having some knowledge in crossover fields doesn't hurt either.

I realize I'm kind of late to this thread, but since I was invited to it by the OP I feel compelled to offer my own interpretation of things. The very first thing I'd do is to remove the side panel completely and see if it makes any difference in CPU temperatures at all. If it does, then you have a case or case fan configuration problem. If it does not, then your case cooling is probably sufficient and you most likely have a CPU cooler problem. And using the stock cooler on any 12th or 13th Gen Intel CPU above an i3 is probably not a wise choice if you plan to do more with it than light to middle of the road tasks. It's not realistically any better than the stock coolers that were on past generation locked SKUs yet the CPUs themselves have significantly higher TDPs than past equivalent tiered models.

 

[Karadjgne]

Base TDP for that CPU is 65w. MAX turbo TDP is 180w. That is per the Intel data sheets.

Exactly. That's what the cpu is spec'd for by Intel. But as I'm quite sure you are well aware of, motherboard vendors don't always agree or even listen to Intel, instead they'll bump power limits, change Tau or even override it completely with 99,999 seconds. Which is why I said 'it can hit' not 'it will hit'.

In its motherboards, Asus’ automatic BIOS-based power-boosting settings for the i7-12700 set PL1 to 165 W and PL2 to 241 W, which was good enough to max out the CPU’s performance in our test.

Intel’s Core i7-12700 tested: Top speeds or power efficiency—pick one

This CPU performs well, but its peak speeds mean high power draw and extra heat.

arstechnica.com

(and you should know by now, I'm personally always happy when you show up to the party, late or not, you generally have insights, ideas or just plain common sense that many, including myself, fall short on at times💡👍😁)

 

[Darkbreeze]

Right, but the bottom line is it really doesn't matter what board manufacturers do or do not do in regard to enforcing or eliminating power limits on a given board. If a CPU (Especially a LOCKED model) can only do 4.9Ghz, and if the CPU is locked so that you cannot change core frequency behaviors (Especially on a non-Z chipset), then whatever the maximum boost TDP is, is what it is. Sure, the board manufacturer can allow you to make changes to power but unless you were willing to increase the core voltage HIGHER than what the default core voltage is by default (And why would you ever do that since clearly it's already erring somewhat to the heavy side in an abundance of caution aimed at increased stability) then you wouldn't be able to exceed what they've already outlined.

In other words, if you can only do 120mph because there is a drivetrain governor installed (Max boost limit, not max power limit) then it doesn't matter if you can increase the available horsepower (Power limit) you are still only going to be able to go 120mph even if there is enough power available to go faster than that. And if you can only go X speed, then the only way you are going to increase the maximum TDP is by increasing the maximum voltage, but there would never be any reason to do that that I can logically think of. The only reason, ever, to increase voltage, is if you are overclocking. And in this case, since overclocking isn't even a possible ingredient in the cake batter, you have to think that there is only one or two things that could even remotely be causing this problem.

CPU cooler or bad paste job. Or they live somewhere with an extremely high ambient temperature, PLUS not having a good enough CPU cooler. So if you are at a full all-core maximum load, it should not matter whether you are looking at Intel spec or some other spec (???What other spec IS there), if it's a locked CPU on a locked chipset, max is max. Period. It can't be overclocked and it can't go past the predetermined maximum turbo boost, so unless you increase the core voltage it is what it is. Unless something has dramatically changed on 13th gen parts that I am unaware of and I guess that's always possible but I think I'd have heard or read about it by now. Then again, maybe I'm completely overlooking something but I don't think so.

 

[Darkbreeze]

Plus, there is also this, which is directly relevant to the fact that they are using the stock cooler but do not have the turbo boost features disabled.

Another new component we're testing in this review is the Intel Laminar RM1 box cooler. There's also a larger RH1 version that comes with the locked Core i9 parts, and then the more basic RS1 version that comes bundled with Pentium and Celeron parts. They're all technically 65 watt coolers, so the RM1 is designed to work with the 12700 (F)only when the 65w spec is enforced.

Overall this is a much better product, but it's only rated for 65 watts so don't expect anything amazing.

Which means PLAINLY that even Intel says the stock cooler is not enough for that CPU unless you are disengaged from boost profiles.

I realize that the PL1 and PL2 limits are usually ignored by the board manufacturers and that the revised PL limits DO result in a higher power draw, it just doesn't really make any sense to me WHY you would do that. More power does not equal faster compute. You could add 5v to the core voltage (IF it could withstand it, which of course it can't) and it would never result in higher core performance without an equivalent increase in core frequency, so what then is the point? Just to run hotter? Higher power draw equals more heat which generally equals "will throttle" which then equals "Less performance" so none of it really makes much sense. Unless you're running an unlocked SKU and overclocking, just run your CPU within either the Intel or AMD spec, give it the best cooling you can afford and let the profiles that engineers MUCH smarter than 99.999% of the "several people told me so" crowd or even most of us, will ever come close to being, work the way it was intended to.

Certainly there are exceptions for those who are exceptionally well versed in making things work the way they want but were not intended to, but those are not ever the same people coming to the forums to ask for advice and usually also aren't most of the people giving or offering it.

 

[Karadjgne]

I realize that the PL1 and PL2 limits are usually ignored by the board manufacturers and that the revised PL limits DO result in a higher power draw, it just doesn't really make any sense to me WHY you would do that.

I, personally, wouldn't. And I know you wouldn't either, excepting test scenarios. But a lot of ppl ain't us. That's why they install MSI Dragon Center, set mobo bios options to Advanced Performance, use Performance power plans, BCLK OC, all in the name of trying to eeek out that last 5 fps. Like you, I don't understand Why they do it, but they do. And mobo vendors do what they can to help that happen, within limits.

 

[BossSiggy]

It's not a joke, and it's entirely possible. If seeing 80C core temperature is that serious... if even seeing so much as a spike to that is too much, then just limit it to the value right before 80C.

It doesn't behave any differently - just throttles back 21C lower than usual, and boost right back up once there's even a little temperature headroom.
Unlike Ryzen, the current Intel chips do not care what the operating temperature is, besides the thermal limit.
If it's 30/50/80C, you still get the max turbo boost, whatever that value is in regards to how many cores are active. Even at an alarming 98C, those cores will still boost to their max. That doesn't happen with Ryzen.

There is an unspoken timer on thermal shutdown. If the throttle protections work, you never see shutdown, because the timer gets reset if temperature successfully drops.

So i would lose performance, since it would start throttling earlier right ? I mean if this would solve temperature problems no one would do anything else than putting a limit on how much degrees they hit ?

 

[BossSiggy]

I don't know what you are looking at K, but that's wrong. He has a 12700F. Base TDP for that CPU is 65w. MAX turbo TDP is 180w. That is per the Intel data sheets. In any case, I'd guarantee this is a case and CPU cooler problem. Undervolting isn't a solution, it's just a good way to introduce instability unless you are a fairly well experienced tweaker/overclocker and know exactly what you are doing AND are willing to take the time to go through the highly involved process of verifying stability. And anybody who says you don't need to do that (Yes, I know it's not you) is absolutely somebody you want to immediately put on the "do not listen to" list. If you have to undervolt a locked CPU just to get it to avoid throttling then you have some other issue, and voltage is the least of your concerns.

Also, just because somebody has the same case (Or other hardware) as what is in question in any given situation does not automatically mean they "know" what they are talking about or that they are correct in their line of thinking. It also doesn't automatically mean they don't either. You don't have to own a case, or cooler, or other piece of hardware, to have a very good idea about it's capabilities and quirks if you are somebody with a good grasp of the relevant fundamentals and at least a moderately good amount of experience over the years with a variety of hardware. Having some knowledge in crossover fields doesn't hurt either.

I realize I'm kind of late to this thread, but since I was invited to it by the OP I feel compelled to offer my own interpretation of things. The very first thing I'd do is to remove the side panel completely and see if it makes any difference in CPU temperatures at all. If it does, then you have a case or case fan configuration problem. If it does not, then your case cooling is probably sufficient and you most likely have a CPU cooler problem. And using the stock cooler on any 12th or 13th Gen Intel CPU above an i3 is probably not a wise choice if you plan to do more with it than light to middle of the road tasks. It's not realistically any better than the stock coolers that were on past generation locked SKUs yet the CPUs themselves have significantly higher TDPs than past equivalent tiered models.

Ive spent so many hours on trying to undervolt now, that not going through with it would almost feel like a waste of time and from what i understand i just need to run tests in Cinebench and look at HWInfo or CPU-Z and check scores and temperatures ? Ive mostly looked into undervolting, not how to find the right undervolt yet, so i only got an idea, since i just wanted to see if i even could undervolt to begin with. But yeah i think my biggest problem is my CPU cooler, but my case is <Mod Edit> aswell, so i will try taking the side off to see if it makes any difference and to get more knowledge about what needs to be done.

yeah i wish i had known how bad stock coolers were before buying my pc, so i could have made some changes to it before getting it, but the shop where i bought it from said that the stock cooler would be more than good enough for my use, so i trusted that, but i disagree with that after reading on the internet about temperatures and my old pc dying from bad maintenance including to high temperatures.

Also crazy that Intel are using the same coolers for better CPU's, but i guess they only care about the CPU's.

 

[BossSiggy]

Exactly. That's what the cpu is spec'd for by Intel. But as I'm quite sure you are well aware of, motherboard vendors don't always agree or even listen to Intel, instead they'll bump power limits, change Tau or even override it completely with 99,999 seconds. Which is why I said 'it can hit' not 'it will hit'.

Intel’s Core i7-12700 tested: Top speeds or power efficiency—pick one

This CPU performs well, but its peak speeds mean high power draw and extra heat.

arstechnica.com

(and you should know by now, I'm personally always happy when you show up to the party, late or not, you generally have insights, ideas or just plain common sense that many, including myself, fall short on at times💡👍😁)

''change Tau or even override it completely with 99,999 seconds'' Is this something i can do or is this only something the BIOS manufacturer can do ? Does this lower CPU temperatures or give you more performance ?

 

[BossSiggy]

Exactly. That's what the cpu is spec'd for by Intel. But as I'm quite sure you are well aware of, motherboard vendors don't always agree or even listen to Intel, instead they'll bump power limits, change Tau or even override it completely with 99,999 seconds. Which is why I said 'it can hit' not 'it will hit'.

Intel’s Core i7-12700 tested: Top speeds or power efficiency—pick one

This CPU performs well, but its peak speeds mean high power draw and extra heat.

arstechnica.com

(and you should know by now, I'm personally always happy when you show up to the party, late or not, you generally have insights, ideas or just plain common sense that many, including myself, fall short on at times💡👍😁)

I just read that whole article, that was alot of reading holy .. But putting P1 and P2 up gives better performance from what i understood ? Would lowering those giving me worse performance then but lower temperatures ? Because i dont want lower performance, either higher performance and lower temperatures or just lower temperatures without loosing performance.

 

[Phaaze88]

So i would lose performance, since it would start throttling earlier right?

Lose performance in one of the following, which you posted in post #24?

I dont use my pc for gaming, my normal workload is: PrivadoVPN, Spotify, ZoneAlarm, CoreTemp and Intel Driver Support Assistant open down in the right corner and a few documents...

A VPN hammers a cpu? No way...
Spotify? No.
So a virus scan takes a few seconds longer with a lower thermal limit... where's the harm in that?
CoreTemp? That's a monitoring tool...
Can't imagine IDSA doing anything hard on the cpu.
A few still documents won't do anything.
... or are you referring to benchmarks, which don't reflect/represent any of the above, thus are not relevant to how you use your PC? I hope that's not it...

These cpus have default thermal limits of 100C, but just about no one wants to see core temperatures get anywhere close to that, due to fans and personal feelings.
What's being missed out on by lowering the thermal limit to a value you're comfortable with?
What's gained by leaving the thermal limit at 100C, when you're not comfortable seeing 80C?

I mean if this would solve temperature problems no one would do anything else than putting a limit on how much degrees they hit ?

Because 'everyone' seems to be caught up in all the buzz about undervolting, or whatever is trending across Youtube. Thinking and second-guessing are drifting away.
One big ol' hivemind misses 2 functions that do almost the same thing, and it doesn't require hours of Cinebench, Prime, OCCT, Linpack, or whatever, to be sure it's stable, unlike with UV/OCs.
I admit to playing around with cpu/gpu UVing and OCing for a bit, due to curiosity and the buzz, but some posts on here and others got me thinking just how much of this is really necessary.
Currently got a thermal limit of 85C and a power long + power short of 210w set for my 7820X. Didn't need to set both a thermal and power limit, but I figure, why not? Doesn't cost anything to do.
I guess I'm one of the odd ones out there...

If you UV/OC, you will NOT want to update the bios anymore. Some updates contain new, usually improved, voltage curves. These will break UV/OCs, forcing you to find new settings.
Temperature and power limits? Doesn't break - but clearing CMOS resets them.

 

[Karadjgne]

The reason undervolting exists is very simple. Intel/Amd Guarantee that in a stock, default pc, that the cpu will be 100% stable.

That said, stability has a Lot do do with voltages used by the cpu. However, cpus are made of silicon and silicon has variences. It's not all identical, doesn't react to voltages in the same way etc. It's a Bell curve, there's going to be a majority that are extremely similar responses but some that are going to deal very well with voltages and some that do not.

In order to be able to claim that Guarantee, the voltages used by the cpu must cover ALL of the cpus, great, middling and poor.

So for instance, if the median cpu requires 1.208v to maintain stability, and the great only require 1.116v, there will be some that require at least 1.300v. So Intel/Amd will put the working voltages for default cpus at 1.400v, just to cover their Guarantee claim.

With that, you get a cpu, and it's showing a 1.400v default vcore. Intel/Amd isn't about to voltage test every single cpu produced and tailor its voltage use. That's on you, if you decide to do such. So undervolting exists as a possible or potential way to reduce vcore, without affecting performance to the negative. But whether that's a 1.300v poor cpu or a 1.116 great cpu or anything in between is anyone's guess.

You'll need to run multiple tests at differing undervolt levels, the exact same test each time, to determine what is going to be the limit to how far down the voltage scale you can go before losing performance or suffering instability. Which btw, is exactly what happens with a manual OC, and is technically OverClocking the cpu to run it at higher speeds than the voltages are set for regardless of what that speed actually is.

Oh, and undervolting does not reduce performance. Undervolting too far does. Setting a thermal limit also does not reduce performance, it limits the maximum performance possible.

A Ferrari will still drive, accelerate, brake, corner or sound like a Ferrari regardless of whether the speed limit is 55mph or 155mph. Just because it has external limitations doesn't mean it's going to suddenly behave like Mom's mini-van.

 

[Darkbreeze]

Lose performance in one of the following, which you posted in post #24?

A VPN hammers a cpu? No way...
Spotify? No.
So a virus scan takes a few seconds longer with a lower thermal limit... where's the harm in that?
CoreTemp? That's a monitoring tool...
Can't imagine IDSA doing anything hard on the cpu.
A few still documents won't do anything.
... or are you referring to benchmarks, which don't reflect/represent any of the above, thus are not relevant to how you use your PC? I hope that's not it...

These cpus have default thermal limits of 100C, but just about no one wants to see core temperatures get anywhere close to that, due to fans and personal feelings.
What's being missed out on by lowering the thermal limit to a value you're comfortable with?
What's gained by leaving the thermal limit at 100C, when you're not comfortable seeing 80C?




Because 'everyone' seems to be caught up in all the buzz about undervolting, or whatever is trending across Youtube. Thinking and second-guessing are drifting away.
One big ol' hivemind misses 2 functions that do almost the same thing, and it doesn't require hours of Cinebench, Prime, OCCT, Linpack, or whatever, to be sure it's stable, unlike with UV/OCs.
I admit to playing around with cpu/gpu UVing and OCing for a bit, due to curiosity and the buzz, but some posts on here and others got me thinking just how much of this is really necessary.
Currently got a thermal limit of 85C and a power long + power short of 210w set for my 7820X. Didn't need to set both a thermal and power limit, but I figure, why not? Doesn't cost anything to do.
I guess I'm one of the odd ones out there...

If you UV/OC, you will NOT want to update the bios anymore. Some updates contain new, usually improved, voltage curves. These will break UV/OCs, forcing you to find new settings.
Temperature and power limits? Doesn't break - but clearing CMOS resets them.

Totally agree on the VPN. It ABSOLUTELY does NOT have any effect in that regard. I've been using NordVPN for a while now, and it is simply what it presents itself to be. Nothing more. Nothing less. If you are using some other VPN I can't say but I CAN say that if you are using Nord, you are wrong if you think it is doing anything that significantly affects system performance. It just doesn't. Period.

 

[KyaraM]

Totally agree on the VPN. It ABSOLUTELY does NOT have any effect in that regard. I've been using NordVPN for a while now, and it is simply what it presents itself to be. Nothing more. Nothing less. If you are using some other VPN I can't say but I CAN say that if you are using Nord, you are wrong if you think it is doing anything that significantly affects system performance. It just doesn't. Period.

I used, like... three different VPNs not named nordVPN now at this point? None of these noticeably impacted system performance or even got my CPU up in temperatures. They did internet network connection, obviously, but that's a different story.

 

[BossSiggy]

Right, but the bottom line is it really doesn't matter what board manufacturers do or do not do in regard to enforcing or eliminating power limits on a given board. If a CPU (Especially a LOCKED model) can only do 4.9Ghz, and if the CPU is locked so that you cannot change core frequency behaviors (Especially on a non-Z chipset), then whatever the maximum boost TDP is, is what it is. Sure, the board manufacturer can allow you to make changes to power but unless you were willing to increase the core voltage HIGHER than what the default core voltage is by default (And why would you ever do that since clearly it's already erring somewhat to the heavy side in an abundance of caution aimed at increased stability) then you wouldn't be able to exceed what they've already outlined.

In other words, if you can only do 120mph because there is a drivetrain governor installed (Max boost limit, not max power limit) then it doesn't matter if you can increase the available horsepower (Power limit) you are still only going to be able to go 120mph even if there is enough power available to go faster than that. And if you can only go X speed, then the only way you are going to increase the maximum TDP is by increasing the maximum voltage, but there would never be any reason to do that that I can logically think of. The only reason, ever, to increase voltage, is if you are overclocking. And in this case, since overclocking isn't even a possible ingredient in the cake batter, you have to think that there is only one or two things that could even remotely be causing this problem.

CPU cooler or bad paste job. Or they live somewhere with an extremely high ambient temperature, PLUS not having a good enough CPU cooler. So if you are at a full all-core maximum load, it should not matter whether you are looking at Intel spec or some other spec (???What other spec IS there), if it's a locked CPU on a locked chipset, max is max. Period. It can't be overclocked and it can't go past the predetermined maximum turbo boost, so unless you increase the core voltage it is what it is. Unless something has dramatically changed on 13th gen parts that I am unaware of and I guess that's always possible but I think I'd have heard or read about it by now. Then again, maybe I'm completely overlooking something but I don't think so.

its not the ambient temperature, im not sure what degrees its in my room and inside my case, but i live in Denmark, our summers go from 15-30 degrees C, usually between 20-25, so it never gets really hot here and i have windows open atleast 3 times a day in every room with doors open and right now its winter with and we maximum get 6 degrees celcius at the moment.

 

[BossSiggy]

Plus, there is also this, which is directly relevant to the fact that they are using the stock cooler but do not have the turbo boost features disabled.



Which means PLAINLY that even Intel says the stock cooler is not enough for that CPU unless you are disengaged from boost profiles.

I realize that the PL1 and PL2 limits are usually ignored by the board manufacturers and that the revised PL limits DO result in a higher power draw, it just doesn't really make any sense to me WHY you would do that. More power does not equal faster compute. You could add 5v to the core voltage (IF it could withstand it, which of course it can't) and it would never result in higher core performance without an equivalent increase in core frequency, so what then is the point? Just to run hotter? Higher power draw equals more heat which generally equals "will throttle" which then equals "Less performance" so none of it really makes much sense. Unless you're running an unlocked SKU and overclocking, just run your CPU within either the Intel or AMD spec, give it the best cooling you can afford and let the profiles that engineers MUCH smarter than 99.999% of the "several people told me so" crowd or even most of us, will ever come close to being, work the way it was intended to.

Certainly there are exceptions for those who are exceptionally well versed in making things work the way they want but were not intended to, but those are not ever the same people coming to the forums to ask for advice and usually also aren't most of the people giving or offering it.

I wish they would make coolers good enough for their CPU's, but i guess they dont think thats worth it.

 

[BossSiggy]

Lose performance in one of the following, which you posted in post #24?

A VPN hammers a cpu? No way...
Spotify? No.
So a virus scan takes a few seconds longer with a lower thermal limit... where's the harm in that?
CoreTemp? That's a monitoring tool...
Can't imagine IDSA doing anything hard on the cpu.
A few still documents won't do anything.
... or are you referring to benchmarks, which don't reflect/represent any of the above, thus are not relevant to how you use your PC? I hope that's not it...

These cpus have default thermal limits of 100C, but just about no one wants to see core temperatures get anywhere close to that, due to fans and personal feelings.
What's being missed out on by lowering the thermal limit to a value you're comfortable with?
What's gained by leaving the thermal limit at 100C, when you're not comfortable seeing 80C?




Because 'everyone' seems to be caught up in all the buzz about undervolting, or whatever is trending across Youtube. Thinking and second-guessing are drifting away.
One big ol' hivemind misses 2 functions that do almost the same thing, and it doesn't require hours of Cinebench, Prime, OCCT, Linpack, or whatever, to be sure it's stable, unlike with UV/OCs.
I admit to playing around with cpu/gpu UVing and OCing for a bit, due to curiosity and the buzz, but some posts on here and others got me thinking just how much of this is really necessary.
Currently got a thermal limit of 85C and a power long + power short of 210w set for my 7820X. Didn't need to set both a thermal and power limit, but I figure, why not? Doesn't cost anything to do.
I guess I'm one of the odd ones out there...

If you UV/OC, you will NOT want to update the bios anymore. Some updates contain new, usually improved, voltage curves. These will break UV/OCs, forcing you to find new settings.
Temperature and power limits? Doesn't break - but clearing CMOS resets them.

What i mean is if i set the maximum temperatures my pc can hit to 80, wont it run slower when its gets close to the maximum temperature ive set it to ? So now it would run slower getting close to 100, if i put the maximum degrees to 80, it would run slower getting close to 80 ?

i mean my virus program with the settings i have now makes my pc jump in 8x degrees, so what would happen it the maximum i allowed my pc to hit was 80 degrees ? And i want to change my settings back aswell, so that my pc runs faster, but then i would hit 9x degrees doing a virus scan, so im waiting with changing those settings until it doesnt get this hot anymore.

ZoneAlarm is probably not good for temperatures though, there comes alot of unnecessary stuff with that firewall.

no im not talking about Benchmarks, i dont run those at all, i would only do that if i was going to UV, but im gonna wait with that since im gonna change my CPU exactly same model just a new one and from what i understand every chip is different, so i would have to UV again after changing the CPU

 

[BossSiggy]

The reason undervolting exists is very simple. Intel/Amd Guarantee that in a stock, default pc, that the cpu will be 100% stable.

That said, stability has a Lot do do with voltages used by the cpu. However, cpus are made of silicon and silicon has variences. It's not all identical, doesn't react to voltages in the same way etc. It's a Bell curve, there's going to be a majority that are extremely similar responses but some that are going to deal very well with voltages and some that do not.

In order to be able to claim that Guarantee, the voltages used by the cpu must cover ALL of the cpus, great, middling and poor.

So for instance, if the median cpu requires 1.208v to maintain stability, and the great only require 1.116v, there will be some that require at least 1.300v. So Intel/Amd will put the working voltages for default cpus at 1.400v, just to cover their Guarantee claim.

With that, you get a cpu, and it's showing a 1.400v default vcore. Intel/Amd isn't about to voltage test every single cpu produced and tailor its voltage use. That's on you, if you decide to do such. So undervolting exists as a possible or potential way to reduce vcore, without affecting performance to the negative. But whether that's a 1.300v poor cpu or a 1.116 great cpu or anything in between is anyone's guess.

You'll need to run multiple tests at differing undervolt levels, the exact same test each time, to determine what is going to be the limit to how far down the voltage scale you can go before losing performance or suffering instability. Which btw, is exactly what happens with a manual OC, and is technically OverClocking the cpu to run it at higher speeds than the voltages are set for regardless of what that speed actually is.

Oh, and undervolting does not reduce performance. Undervolting too far does. Setting a thermal limit also does not reduce performance, it limits the maximum performance possible.

A Ferrari will still drive, accelerate, brake, corner or sound like a Ferrari regardless of whether the speed limit is 55mph or 155mph. Just because it has external limitations doesn't mean it's going to suddenly behave like Mom's mini-van.

Makes sense. I guess it would be way to much work for Intel and AMD to test every CPU's preferable voltage and that would increase the prices way to much for the CPU's compared to what you get, so thats why they are not doing it ?

Can you increase performance with UV ? I read that somewhere, i dont know if its true though, because it sounds like you just lower the temperatures to the point where you dont lose performance and get instability ?

but there is not point in me putting my thermal limit down then, i dont want my pc to run slower, i want it to run as fast as possible without getting higher than 80 C.

 

[BossSiggy]

Totally agree on the VPN. It ABSOLUTELY does NOT have any effect in that regard. I've been using NordVPN for a while now, and it is simply what it presents itself to be. Nothing more. Nothing less. If you are using some other VPN I can't say but I CAN say that if you are using Nord, you are wrong if you think it is doing anything that significantly affects system performance. It just doesn't. Period.

Yeah VPN doesnt do that, i have set up CoreTemp so that i can see all 12 cores temperatures all the time down in the taskbar and the only thing that makes my pc get to hot is when i run anti virus full scan.

 

[Karadjgne]

I keep thinking that you are seeing performance loss as a definitive difference. It isn't. If an 8 core cpu is running 5.0GHz on 7 cores and 4.95GHz on 1 core, that'd be like driving at 54.9mph in a 55mph zone vs doing exactly 55.0mph.

Looking at cinebench or other numbers and freaking out because of a couple points loss isn't doing you any good.

It's a very simple fix. If you personally do not like the temp number, and want it lowered for your own personal tastes, you'll need a better cooler. The cpu really doesn't care about the temps, they are inside specs.

 

[Phaaze88]

Spoiler

Totally agree on the VPN. It ABSOLUTELY does NOT have any effect in that regard. I've been using NordVPN for a while now, and it is simply what it presents itself to be. Nothing more. Nothing less. If you are using some other VPN I can't say but I CAN say that if you are using Nord, you are wrong if you think it is doing anything that significantly affects system performance. It just doesn't. Period.

Not just the VPN, but none of what they stated the PC is used for really screams 'performance lost' if they were to set a lower thermal limit.
Would you notice a few seconds longer virus scan?

its not the ambient temperature, im not sure what degrees its in my room and inside my case, but i live in Denmark, our summers go from 15-30 degrees C, usually between 20-25, so it never gets really hot here and i have windows open atleast 3 times a day in every room with doors open and right now its winter with and we maximum get 6 degrees celcius at the moment.

Then buy a thermometer that you can dangle the sensor in front of the cpu cooler - or perhaps you already have a thermometer, just need to improvise in suspending it in front of the cooler.
Then you'll know how warm/cool the air going into the cooler is.
Inside the case is typically warmer than outside and in your room. When the gpu is active, the case ambient will be warmer.
Just because it's 6 or 25C outside, doesn't mean it's the same going in the case.

What i mean is if i set the maximum temperatures my pc can hit to 80, wont it run slower when its gets close to the maximum temperature ive set it to ? So now it would run slower getting close to 100, if i put the maximum degrees to 80, it would run slower getting close to 80 ?

I guess you missed where I posted that Intel cpus don't care about their operating thermals sans the thermal limit. Understandable, as I have a bad habit of long posts.
They do not run any slower, unlike Ryzen cpus, Geforce and Radeon gpus. [Can't recall if Intel ARC functions the same as the other gpus.]
Cores are at 20C? It's 5.0ghz or whatever the boost is.
50C? Still 5.0ghz.
85C? Yep, still 5.0ghz.
98C? You guessed it - 5.0ghz.
At 100C? It'll dial back a few hundred mhz, and if successful in lowering temperature, they're right back up to 5.0ghz until 100C is reached... and this is repeated over and over and over, and over - it does not stay at lower clocks unless the throttle protections aren't working.

so what would happen it the maximum i allowed my pc to hit was 80 degrees ?

Take the above and change 100C to 80C.

And i want to change my settings back aswell, so that my pc runs faster...
...
i have set up CoreTemp so that i can see all 12 cores temperatures all the time down in the taskbar and the only thing that makes my pc get to hot is when i run anti virus full scan.

So what you're saying is, the PC doesn't run slower. Cpu gets too hot for you during virus scans, but everything else is fine. One application is making everything else look bad?
What was wrong with Windows Defender? Or maybe you're using Linux.

Get a bigger cooler then - despite my responses to your query about no cost solutions. That's the main reason I chimed in.
Undervolting will only go so far; it has limits. If it doesn't do enough to lower core thermals, you gotta get a bigger cooler.

 

[DSzymborski]

I can't believe we're in page four of this thread.

Really, the premise is the same: wattage means heat. If you don't want to ever have your CPU go over 80 degrees, that's fine, but you have to take the very real performance hit as a tradeoff. OP, it sounds in places like you want for us to make the laws of physics to not apply to your PC. We can't do that. You get to determine what tradeoffs work best for your own situation, but you don't get to not have tradeoffs.

 

[GarrettL]

Spoiler

Not just the VPN, but none of what they stated the PC is used for really screams 'performance lost' if they were to set a lower thermal limit.
Would you notice a few seconds longer virus scan?


Then buy a thermometer that you can dangle the sensor in front of the cpu cooler - or perhaps you already have a thermometer, just need to improvise in suspending it in front of the cooler.
Then you'll know how warm/cool the air going into the cooler is.
Inside the case is typically warmer than outside and in your room. When the gpu is active, the case ambient will be warmer.
Just because it's 6 or 25C outside, doesn't mean it's the same going in the case.


I guess you missed where I posted that Intel cpus don't care about their operating thermals sans the thermal limit. Understandable, as I have a bad habit of long posts.
They do not run any slower, unlike Ryzen cpus, Geforce and Radeon gpus. [Can't recall if Intel ARC functions the same as the other gpus.]
Cores are at 20C? It's 5.0ghz or whatever the boost is.
50C? Still 5.0ghz.
85C? Yep, still 5.0ghz.
98C? You guessed it - 5.0ghz.
At 100C? It'll dial back a few hundred mhz, and if successful in lowering temperature, they're right back up to 5.0ghz until 100C is reached... and this is repeated over and over and over, and over - it does not stay at lower clocks unless the throttle protections aren't working.


Take the above and change 100C to 80C.


So what you're saying is, the PC doesn't run slower. Cpu gets too hot for you during virus scans, but everything else is fine. One application is making everything else look bad?
What was wrong with Windows Defender? Or maybe you're using Linux.

Get a bigger cooler then - despite my responses to your query about no cost solutions. That's the main reason I chimed in.
Undervolting will only go so far; it has limits. If it doesn't do enough to lower core thermals, you gotta get a bigger cooler.

Agreed, this was obvious on page 1 or 2 of this thread.

 

[BossSiggy]

I keep thinking that you are seeing performance loss as a definitive difference. It isn't. If an 8 core cpu is running 5.0GHz on 7 cores and 4.95GHz on 1 core, that'd be like driving at 54.9mph in a 55mph zone vs doing exactly 55.0mph.

Looking at cinebench or other numbers and freaking out because of a couple points loss isn't doing you any good.

It's a very simple fix. If you personally do not like the temp number, and want it lowered for your own personal tastes, you'll need a better cooler. The cpu really doesn't care about the temps, they are inside specs.

I think i understand, but it sounds like you dont think i understand and maybe i dont understand. But if my CPU is throttling running virus scan now where it can hit maximum 100 degrees and i put the temperature down to 80 degrees as maximum, then it should throttle more, because its gonna hit 80 way faster than 100 right or does it mean that now im hitting 8x-9x degrees doing a virsun scan where it can hit maximum 100 degrees, so if i put it down to being able to hit maximum 80 is it going to hit 6x-7x degrees instead and just run slower in general ?

im not referring to Cinebench at all in this thread i think, except for if im gonna test for instability and performance before UV'ing ?

simple fix ? I asked this question in multiple forums and ive been writing down peoples suggestions to fix this and people came with atleast 21 different suggestions, like some is saying i need both a better case and cpu cooler, some are saying that a new cpu cooler is enough, some is saying take the side off, some is saying get Intel ME, some is saying UV, some is saying get a contact frame, some is saying your pc has to be this many cm's off the ground, some is saying watercooling, some is saying air cooling, some is saying get air condtion and thats just some of the solutions ive heard, im probably just gonna get a better cpu cooler, because that is what most people are saying and what makes the most sense, but im still gathering information atm. Maybe someone has some really good idea no one else thought about and when im done with that i'll do what makes most sense, i mean yeah i could just buy a better case, get 2 front intake fans, 2 intake fans under the pc, another top fan, buy the best CPU cooler, but i already spend the limit i wanna spend on this pc for now, so im also trying to do it as cheap as possible and from the shop i buy from. I should probably just have went with what Intel said to begin with, because all these suggestions is actually just making me more confused, but the solution in the end will probably be better than just going with what Intel said.

 

[BossSiggy]

Spoiler

Not just the VPN, but none of what they stated the PC is used for really screams 'performance lost' if they were to set a lower thermal limit.
Would you notice a few seconds longer virus scan?


Then buy a thermometer that you can dangle the sensor in front of the cpu cooler - or perhaps you already have a thermometer, just need to improvise in suspending it in front of the cooler.
Then you'll know how warm/cool the air going into the cooler is.
Inside the case is typically warmer than outside and in your room. When the gpu is active, the case ambient will be warmer.
Just because it's 6 or 25C outside, doesn't mean it's the same going in the case.


I guess you missed where I posted that Intel cpus don't care about their operating thermals sans the thermal limit. Understandable, as I have a bad habit of long posts.
They do not run any slower, unlike Ryzen cpus, Geforce and Radeon gpus. [Can't recall if Intel ARC functions the same as the other gpus.]
Cores are at 20C? It's 5.0ghz or whatever the boost is.
50C? Still 5.0ghz.
85C? Yep, still 5.0ghz.
98C? You guessed it - 5.0ghz.
At 100C? It'll dial back a few hundred mhz, and if successful in lowering temperature, they're right back up to 5.0ghz until 100C is reached... and this is repeated over and over and over, and over - it does not stay at lower clocks unless the throttle protections aren't working.


Take the above and change 100C to 80C.


So what you're saying is, the PC doesn't run slower. Cpu gets too hot for you during virus scans, but everything else is fine. One application is making everything else look bad?
What was wrong with Windows Defender? Or maybe you're using Linux.

Get a bigger cooler then - despite my responses to your query about no cost solutions. That's the main reason I chimed in.
Undervolting will only go so far; it has limits. If it doesn't do enough to lower core thermals, you gotta get a bigger cooler.

Hmm.. Okay.. No i wouldnt care about the virus scan, the problem right now is when im doing a full virus scan and listening to Spotify at the same time its starts lagging in the sound, so i guess thats because its throttling ? Wouldnt that be worse with lower the thermal limit ? I mean how do i lower the thermal limit in BIOS ? I guess i could try it out to see if it makes any difference in performance, lag, throtting, etc.

what am i going to do with the temperature of inside the case ? And yeah i know its obviously hotter inside the case, it was just to say its not that hot that i need to get a air conditioner like someone said (no one uses air conditioners in Denmark)

I understand, it just doesnt make sense to me, because whats the point of having a cooler then ? I could just put my pc to maximum 40 degrees, then after run virus scans all day long, but it wont get higher than that in temperatures, but also it wont lose performance, it wont lag, it wont throttle, there is no consequence to this ? But i mean i could try it and see what i think for myself.

it does run slower, Spotify starts lagging in sound. When i say virus scan i mean Windows Defender, since thats what im using as virus scanner.