Auto Overclock/Boost Clock Help

[Erabior]

I have an i9 9900k on a ROG strix z390 EGaming Motherboard and a noctua NH-D15 with only one fan as a cooling solution. My question is:
Is there a way to have my base clock increase automatically while using my pc, as the games/processes i am running use more processing power? I ask this because i was recently given an Acer Nitro 5 with an 8th gen core i5 for setting a sales record at work and it seems to run Hearts of Iron 4 at INSANE speeds as well as, i noticed that the clock speed changes based on load while viewing the speed in the task manager. If i could get any clarity or a solution with how to get a similar effect on my desktop with the i9 that would be absolutely amazing

[Autumn_]

Every CPU that I know of turbo boosts, so your i9-9900k already does what you discribed, unless you've disabled it in the BIOS.
So you'll notice if you monitor the speeds of your i9-9900k it will jump from 3.6ghz to 5ghz.

[tacoshy]

There 2 things by default:

Turbo Boost - clocks cores higher on cost of other cores that are put into idle state and transferring their TDP to boost the other cores.

SpeedStep - clocks the entire CPU higher temporarily and raises it TDP for a short while burst wise as normal you don't have steady load but peak wise. During the peaks it will boost higher and will cool down later on during the low load phases. It is highly complex and also has something to do with your temperature.

[_I_]

looks like it will only turbo 2 cores at a time

https://en.wikichip.org/wiki/intel/core_i9/i9-9900k

Turbo Frequency
5,000 MHz (1 core),
5,000 MHz (2 cores)

[Erabior]

the thing is, i can oc it to 5GHz no problem, the thing is i ran Aidax64 with the cpu at 5GHz and my temps went to 80C at the highest. so should i just oc my cpu to 4.2/4.3/4.4/4.5 and see what my temps go to under normal load (not aida cuz its insane) and just go from there?

[Mad Scientist]

Be under 70C at all times to keep the lifespan of the CPU as high as possible (55C target), I don't care if it says the junction can handle 100C; your absolute highest safe temp should be treated as 70C. Seen people destroy CPUs by OCing and hitting the 80-90C range. Highly recommend water cooling for beefy CPUs especially when you're making it hit 80C.

[_I_]

if 3 or more cores are in use it will not turbo
it will only turbo 1-2 cores when the rest are idle and temps are low

if you want them all at a higher clock, disable turbo in bios and set the cpu multi higher

basic rules to oc
temp low, stable = raise cpu multi
temps low unstable = raise core voltage
temps high, stable = lower core voltage
temps high, unstable = go back to last stable and stop

[tacoshy]

引用自 Orion：

Be under 70C at all times to keep the lifespan of the CPU as high as possible (55C target), I don't care if it says the junction can handle 100C; your absolute highest safe temp should be treated as 70C. Seen people destroy CPUs by OCing and hitting the 80-90C range. Highly recommend water cooling for beefy CPUs especially when you're making it hit 80C.

You clearly have no idea what you talk about... Temperature doesn't destroy a CPU or damage it when run 24/7 at high Temps. Just stay below rated Temps preferably below throttle temps.

I overclock on daily basis for customers and for myself and have never seen a CPU die because of that the last 10 years.

If you say you should possible stay below 55C then you're dreaming. Most CPU's on the current market can't get that even in stock settings unless you pay more for cooling then the entire CPU+MoBo+RAM setup.

Water cooling is not necessarily better then Air cooling. Especially true if you talk about AiO coolers.


It would be nice if you provide proof for your theory. There is no evidence the overclocking or running it at high Temps for long durations have a real effect on lifespan. They easily live 15 years still.

[Mad Scientist]

引用自 tacoshy：

引用自 Orion：

Be under 70C at all times to keep the lifespan of the CPU as high as possible (55C target), I don't care if it says the junction can handle 100C; your absolute highest safe temp should be treated as 70C. Seen people destroy CPUs by OCing and hitting the 80-90C range. Highly recommend water cooling for beefy CPUs especially when you're making it hit 80C.

You clearly have no idea what you talk about... Temperature doesn't destroy a CPU or damage it when run 24/7 at high Temps. Just stay below rated Temps preferably below throttle temps.

I overclock on daily basis for customers and for myself and have never seen a CPU die because of that the last 10 years.

If you say you should possible stay below 55C then you're dreaming. Most CPU's on the current market can't get that even in stock settings unless you pay more for cooling then the entire CPU+MoBo+RAM setup.

Water cooling is not necessarily better then Air cooling. Especially true if you talk about AiO coolers.


It would be nice if you provide proof for your theory. There is no evidence the overclocking or running it at high Temps for long durations have a real effect on lifespan. They easily live 15 years still.

Thats funny because where I worked people killed cpus all the time via temperature. What kind of idiot thinks cpus won't die from temperature? Next time you drive take the oil and coolant out of your engine, tell me it won't seize, blow a part or otherwise from heat. It'll just "reduce the lifespan" so clearly it's OK.

Also I run 55c with a stock Intel cooler in average unless the summer, which brings it to 61c under load with an i5. Even my basic cpu water cooler keeps my hotter running cpu at 55c absolute max with oc.

Really this is, also like someone claiming gpus don't fail under high Temps. Guess what? They do didn't believe a guy when he said he blew his but he sent in a video from when his, stream died: right before the gpu blew and it was visual.

Seen plenty of people fry cpus, gpus, on very rare occasions ram, motherboard components from lack of cooling (oc related) , dual cpu motherboard failures from overheating the cpu and socket of Socket 0 (yes, socket 0 is the factory way of saying "primary"). Happens, will happen. Just because it doesn't happen to you doesn't mean it doesn't happen to others.

Not sure how many "customers" you had, or what your oc is like. But do let me know when you supply kingpin and watch him destroy gpus. Wasn't different for cpus either.

[_I_]

temp can kill, but modern cpus will throttle or shutdown when near their max working temps

the old cpus you are refering to did not have thermal sensors built in them, the boards either did not monitor temp or use a socket sensor which is slow to respond and reads way below the cpu core temps

[tacoshy]

If you fry something when OCing it is not because of lack of cooling but because ppl disable security features and push to many volt through it.

My current OC's are 5.5FHz for a 7700K with 1.465V and a 7280X with 5.1GHz on 1.390V. They still run fine without issue 24/7.

But I always tend to overclock high. Lifespan is not an issue because of security features. They will throttle or shut down way befor harm is caused.

If you have a life expectancy of only 15 instead of 16 years, nobody cares when you won't keep the CPU for longer then 5 years anyways.

Also if OC would always kill a CPU so fast, then why you have 1st-4th Gen CPU's still running over locked in such high numbers?

Also you won't get 55C under load on an i7 even with a 360 AiO. Especially not with a weaker one that is outperformed by air coolers.
But I'm happy to see a video from you running an i7 or i9 at 55C under load with a stock cooler. I call massive BS on that. That would be a temp record even when drlidded.

[Mad Scientist]

You can try calling what you want, but I'm not running these systems inside of a volcano so managing Temps is quite easy especially with a better fan curve to start with. Seems you just don't like anything counter to your own minimal experience. Just because you can't achieve something or haven't experienced doesn't make it impossible. I've had people fry parts by failing to listen when other people tell them "it's OK don't worry about it" always telling them if they exceed a temp to end the test. Some systems will still remain operational even when at unsafe max temps - this happens even without users disabling something. These things happen rarely but they do happen, nothing new to anyone that's worked with me either.

Regardless of what you mention, seen plenty of people kill cpus in high temp and anyone that's done high amounts of user support in the hardware section of the industry knows its possible. Feel free to take a 30 minute drive at the very least with the conditions I mentioned. You'd still have some air cooling, it'll be fine by your logic.

But if you don't feel like it, guess we know who's really full of bs.

[Bad 💀 Motha]

80*C is not hot just disable SpeedStep and Turbo and lock it at 5ghz

[tacoshy]

引用自 Orion：

You can try calling what you want, but I'm not running these systems inside of a volcano so managing Temps is quite easy especially with a better fan curve to start with. Seems you just don't like anything counter to your own minimal experience. Just because you can't achieve something or haven't experienced doesn't make it impossible. I've had people fry parts by failing to listen when other people tell them "it's OK don't worry about it" always telling them if they exceed a temp to end the test. Some systems will still remain operational even when at unsafe max temps - this happens even without users disabling something. These things happen rarely but they do happen, nothing new to anyone that's worked with me either.

Regardless of what you mention, seen plenty of people kill cpus in high temp and anyone that's done high amounts of user support in the hardware section of the industry knows its possible. Feel free to take a 30 minute drive at the very least with the conditions I mentioned. You'd still have some air cooling, it'll be fine by your logic.

But if you don't feel like it, guess we know who's really full of bs.

Which is a stupid analogy as the engine will exceed the rated temperature. A CPU rated for 100C is fine to 100C not only to 70C.
Why would Intel or AMD set a safety feature to 93C for CPU's rated for 100C when they break at 70C....


Like I said, to actually break a CPU to temperature you have disable all security features.

Last but not least, you'll find nobody even with delidded CPU that will get 55C under load with a stock cooler that isn't included for a reason at the 9900K.

[Mad Scientist]

引用自 tacoshy：

Which is a stupid analogy as the engine will exceed the rated temperature. A CPU rated for 100C is fine to 100C not only to 70C.
Why would Intel or AMD set a safety feature to 93C for CPU's rated for 100C when they break at 70C....


Like I said, to actually break a CPU to temperature you have disable all security features.

Last but not least, you'll find nobody even with delidded CPU that will get 55C under load with a stock cooler that isn't included for a reason at the 9900K.

Analogy is fine, you wouldn't risk your engine by running it too hot in the first place because exceeding certain levels - but still below the max limit - vastly increases the chances of something going wrong.

Max at die to 100C doesn't mean it's going to survive to said temperature, much like running a GPU at 90C is "ok" to a lot of people (it's not) but hitting 91-95C can blow VRM or other various components that compromise the GPU. It's simply unwise to have higher temps if you want the CPU or socket to remain undamaged. Seen far too many roasted CPU's & GPU's running high temps but below max threshold. Technology is not infallible; there are times when the user does nothing, everything is untouched, and these features fail to do their job - again rarely, but it happens. It's much like when you run a system far too long especially a server; even with ECC ram you can get a "2" instead of a 1 or 0 and it wreaks havoc on some tasks until rebooted or the error corrected.

Though with the pure numbers of individuals I've worked with and supported; of course I'm going to see far more unlikely things including death-by-temp CPUs, roasted GPU, parts on fire/exploding parts, etc - especially with the types of users associated with it that even make rookie mistakes.

Also those temps, doesn't have to be a 9900k - not everyone is going to use your desired cpu and some people live in far cooler areas than other people. Not achieving that on a 9900k again; doesn't mean it's overall impossible. Just like death by temps isn't impossible.

[tacoshy]

A CPU is fine beyond 100C if rated for that. So 80 or 90 C and not even at a concerning level.

A Car engine without cooland has no way to protect itself against overheating unlike a CPU...

So your anathalogy is terrible. And when literally nobody will be able to achieve your recommended temperature despite with some 800 USD custom loops then it barely fit the normal way of usage. Don't have to be the 9900K. Literally all K CPU-s the past 7-8 years won't be able to achieve that.

I have seen a lot of damaged stuff too, not because of overclocking itself but ppl trying to do it while they have no idea what they do.
Also literally every prebuilds highly exeed your limits and does not fail and end up broken. Evidence is clearly against you.
Fried stuff and failing stuff these days not because of missing cooling or running to hot but ppl trying to be smart and pushing it settings they have no idea what they do but believe it will give them higher fps...