Air Cooling -- Myths and setup tips for the novice performance / gaming builder

[Deleted member 185158]

It's good for a 13600kf. I wouldn't use it on a higher chip though. Under most brutal prime95 load I could throw at it, it hit 97 celsius... that was its max... I have it set to 102 celsius limit in bios.

That's extreme though... I mean I won't ever see more than 60 celsius in most games I play. so eh

Yes, sorry for the late reply.

You could just record the wattage in game, and come up with an average output yes. Most bigger coolers can handle average loads surely. But not entirely what my point was really.

You, you don't mind 97c and stretching past 100c Tjunction set by the manufacturer. But does the cpu?

Why is AMD for example, using 70c as a high temp alert (first 4 ryzen generations). This with a default system commands 100% cpu fan duty cycle.
Essentially, they can't count on the cooler because there is no awareness to ambient temperatures.

At one time, manufacturer would list a Tcase temp max. Intel would state 72c and this is an off die temperature, at the ihs plate.

Today we have package temp. For your Intel 13600K is the Tcase temperature, the temp at the IHS plate.

So if you are reading me the 97c core temp, it's usually cooler than the accumulated Tcase temp which seems to run a bit hotter. At least that's what I experience with 12th gen chips, I imagine 13th gen isn't much different in this temp reading aspect.

But really, 97c for Me is a little on the upper end warm side. And I mean package (Tcase) temp, not core temp. It throttles at 100c and that's uncomfortable. Once the package temp hits that 100c with it, the cpu throttles really really hard.

But none of this corresponds with the title of the thread "myths" though. So this reply might be off topic and please excuse me from that.

 

[Bill_Bright]

I have it set to 102 celsius limit in bios.

That's extreme though... I mean I won't ever see more than 60 celsius in most games I play. so eh

I think setting and leaving the limit set that high is a mistake. Temporarily setting it that high while you are benchmarking/stress testing is fine and makes sense because one has to assume you are sitting there, keeping a watchful eye on your temps.

But leaving it there while you go off and perform other tasks, and especially engrossing games, leads to you being distracted and no longer paying close attention to your temps. Should something go wrong, and your temps start climbing unnoticed well above your normal maximums of 60°C, and then suddenly spike into a dangerous/destructive range - not good.

Since you typically never see above 60°C, I would urge you to set the threshold setting in the BIOS setup menu to a "comfortable" (relatively speaking) and safe 70 or 80°C. This will surely give your most taxing games plenty of headroom, but still [hopefully] alert you before it is too late, and [again hopefully] prevent unwanted consequences should something really go awry.

 

[Space Lynx]

I think setting and leaving the limit set that high is a mistake. Temporarily setting it that high while you are benchmarking/stress testing is fine and makes sense because one has to assume you are sitting there, keeping a watchful eye on your temps.

But leaving it there while you go off and perform other tasks, and especially engrossing games, leads to you being distracted and no longer paying close attention to your temps. Should something go wrong, and your temps start climbing unnoticed well above your normal maximums of 60°C, and then suddenly spike into a dangerous/destructive range - not good.

Since you typically never see above 60°C, I would urge you to set the threshold setting in the BIOS setup menu to a "comfortable" (relatively speaking) and safe 70 or 80°C. This will surely give your most taxing games plenty of headroom, but still [hopefully] alert you before it is too late, and [again hopefully] prevent unwanted consequences should something really go awry.

pretty sure intel's default is 95 celsius? or even 100, not sure. this idea is not bad though, i may set it to 90. I don't think anything bad can happen, cause if I hear my fans ramping up when its idle or im just browsing the web, I will investigate the cause quickly. every game I have thrown at it doesn't even ramp up the fan yet. only synthetic benching, so the noise alone would alert me... also I am in my room all the time... so eh I am not too worried.

I'd be worried if I had water cooling with leaks and all, but since i am rocking air, it will be fine.

Yes, sorry for the late reply.

You could just record the wattage in game, and come up with an average output yes. Most bigger coolers can handle average loads surely. But not entirely what my point was really.

You, you don't mind 97c and stretching past 100c Tjunction set by the manufacturer. But does the cpu?

Why is AMD for example, using 70c as a high temp alert (first 4 ryzen generations). This with a default system commands 100% cpu fan duty cycle.
Essentially, they can't count on the cooler because there is no awareness to ambient temperatures.

At one time, manufacturer would list a Tcase temp max. Intel would state 72c and this is an off die temperature, at the ihs plate.

Today we have package temp. For your Intel 13600K is the Tcase temperature, the temp at the IHS plate.

So if you are reading me the 97c core temp, it's usually cooler than the accumulated Tcase temp which seems to run a bit hotter. At least that's what I experience with 12th gen chips, I imagine 13th gen isn't much different in this temp reading aspect.

But really, 97c for Me is a little on the upper end warm side. And I mean package (Tcase) temp, not core temp. It throttles at 100c and that's uncomfortable. Once the package temp hits that 100c with it, the cpu throttles really really hard.

But none of this corresponds with the title of the thread "myths" though. So this reply might be off topic and please excuse me from that.

I mean this was only with prime95... I will never see those temps in anything else... so eh... not going to worry about it.

 

[Bill_Bright]

I don't think anything bad can happen, cause if I hear my fans ramping up

This assumes everything is, and always will work properly. I don't assume that. Also, temps can skyrocket long before fans have time to ramp up to speed.

since i am rocking air, it will be fine.

Right. Because fan bearings never seize - until they do. And when they do, BIOS monitoring circuits for fan RPMs and CPU over-temps never fail - until they do.

I hope you have current backups.

 

[Space Lynx]

ya I got backups, and I have 7 fans... not all 7 will fail at once. I will hear the ramp up if anything weird happens. thanks for your concern though

 

[Deleted member 185158]

That's how myths are

I'd be worried if I had water cooling with leaks and all, but since i am rocking air, it will be fine.

I mean this was only with prime95... I will never see those temps in anything else... so eh... not going to worry about it.

Let the myths be born!
Not worried about your personal preference. It's OK and valid because that belongs to you!!

________________________________________________________________________
I seriously could rip this thread to a lot of different aspects.
Just the original post is easy to tear up, just view the first few responses.

Step 6 is good. Watch package temps, no mention of core temps though.
What about VRMs, board temps or I/O (chipset), memory modules??

The rest of it is to adjust fan rpm. "some fans 600-900 rpm, some 1600 rpm.
No mention of ambient temps, all the possible cases and fan layouts.
Cpu fan rpm adjustments only..... I'm sure there are fans between above and below stated RPMs.
The RPM means nothing to me. What about fan pitch, depth, placement, count ect ect...

"No air cooler can handle 300w...."
This statement depends on a fist full of missing data.

MYTH #1 – “My processor requires liquid cooling”
"This simply is not true in 99.99% of cases"

I'll call this one right out, manufacturers have RECOMMENDED liquid cooling for their products.

"I never see those temps in anything else, so eh....."

lol, no. There's just a lot of wrong or perhaps misleading information here. And missing information too.

 

[Space Lynx]

Step 6 is good. Watch package temps, no mention of core temps though.
What about VRMs, board temps or I/O (chipset), memory modules??

my mobo has so much metal on it, the VRM's need a sweater, maybe I will knit them one someday.

 

[Deleted member 185158]

my mobo has so much metal on it, the VRM's need a sweater, maybe I will knit them one someday.

Mine too actually.
Because the ambient temp in my work space is roughly 5c.
Look some actual useful data! This would take a 250w cooler and probably make it 300w capable!

 

[Mussels]

This assumes everything is, and always will work properly. I don't assume that. Also, temps can skyrocket long before fans have time to ramp up to speed.

Yep heat can literally go up in 1 millisecond on ryzen (cant be too much slower on intel) which how fast they power on/off cores - and the default fan curves on my asus boards change fan speeds every 3 seconds

In 3 seconds the case air wont heat up too much, but if that was controlling a water pump or the fan on a smaller heatsink it's simply too slow

"No air cooler can handle 300w...."

I like this quote, because noctua have something super useful to say on this front
summarised its "fuck TDP ratings, get something that works well with your CPU design"

Noctua’s Standardised Performance Rating (NSPR) and compatibility classification for CPU coolers

 

[dgianstefani]

In 3 seconds the case air wont heat up too much, but if that was controlling a water pump or the fan on a smaller heatsink it's simply too slow

It doesn't matter, with water you have a thermal buffer to absorb changes. The fan or pump speed rarely or never needs to change.

 

[Deleted member 185158]

Yep heat can literally go up in 1 millisecond on ryzen (cant be too much slower on intel) which how fast they power on/off cores - and the default fan curves on my asus boards change fan speeds every 3 seconds

In 3 seconds the case air wont heat up too much, but if that was controlling a water pump or the fan on a smaller heatsink it's simply too slow


I like this quote, because noctua have something super useful to say on this front
summarised its "fuck TDP ratings, get something that works well with your CPU design"

Noctua’s Standardised Performance Rating (NSPR) and compatibility classification for CPU coolers
View attachment 272572

They left out some critical details, but I get the point.

One of the details that is left out is that by design, air coolers are generally more efficient when already warmed up vs going from cold to warm.
Another is the failed mention that everyone has a different ambient temp, and different type of case or flow, perhaps lack there of.

Either way, X amount of BTU must be moved. Another angle I take when performing talks about coolers. converting the wattage to BTU rather than repeating watts this and watts that. Can the cooler move 1200BTU/hr?

Thermal density.
This one is interesting to throw into the mix. Die size and solder or pasted plates.
Either way, you are dissipating X amount of BTU at X given point.

Well the plate and coolers are 2D in BTU movement. Straight up. And down, but the focus is mainly shedding the BTU up, through solder, a plate (of any given size) then through some paste, the cooler and so forth. We cannot cool the cpu from any other direction. One direction only.

And one of the last things, I often noticed through cooler talks is the lack to mention time.
As mentioned above, BTU/hr.
OK, thermal dissipation by time. How long does it take for your 250w to travel from die to air??
I suppose this depends on a lot of variables.

But I do agree TDP ratings have always been a skew since the birth of "BOOST".

CPU manufacturer goal is to make a fast CPU.
Inherently, processors have gained wattage through the years.
The cooler designs have remained similar because the surface area has remained similar.

The idea in the past for overclocking is to REDUCE temps, not run the chip at it's peak output wattage.
This would help reduce leakage and accelerate stability. Unfortunately, those days are gone for most people.
I for one still try and follow this tradition as best as possible. Simply because I have specific goals.

 

[Count von Schwalbe]

And one of the last things, I often noticed through cooler talks is the lack to mention time.
As mentioned above, BTU/hr.
OK, thermal dissipation by time. How long does it take for your 250w to travel from die to air??
I suppose this depends on a lot of variables.

The thing is, BTU/hr and W are the same type of unit. Being as they are a rate rather than an amount, time does not come into play - they are by default assumed to be constant as a worst-case scenario.

 

[Deleted member 185158]

The thing is, BTU/hr and W are the same type of unit. Being as they are a rate rather than an amount, time does not come into play - they are by default assumed to be constant as a worst-case scenario.

Converted, you can get an idea of the rate of dissipation.

300w = 1025 BTU in one HOUR. (roughly)

Or we all just go around and say, my CPU dissipates 300w. This figure means nothing to me unless I can find a cooler than can move that much thermal energy in a short amount of time.

That's why copper, not aluminum.

Coolers aren't usually "rated" over copper's capability of (roughly) 235 BTU/hr.

Just can't make the metal "copper" dissipate heat faster than that. It's just not possible.

I guess in reality, we battle the material's capability more so than the design in which the material is implemented.

See below the quote? T = time.

In imperial units, thermal conductivity is measured in BTU/(h⋅ft⋅°F). The dimension of thermal conductivity is M1L1T−3Θ−1, expressed in terms of the dimensions mass (M), length (L), time (T), and temperature (Θ).

 

[Count von Schwalbe]

Otherwise known as:

398 W/m•K

Watts per meter per degree Kelvin delta.

copper's capability of (roughly) 235 BTU/hr.

I consider this to leave out several pertinent pieces of information.

That assumes 1 degree (F) difference and 1 foot thickness.

Most coolers operate in much different conditions.

 

[Deleted member 185158]

Otherwise known as:

Watts per meter per degree Kelvin delta.

I consider this to leave out several pertinent pieces of information.

That assumes 1 degree (F) difference and 1 foot thickness.

Most coolers operate in much different conditions.

Well I am just generalizing the idea that you can't magically make copper faster at dissipating heat.

Thus a cooler's capability could never be over it's material rated thermal movement through time.

And we haven't even talked about the thermal pastes and the effect this has increasing the thermal movement time. Now you're at the mercy of filling a space.

Of course there's a handful of data missing. I'm not going to measure eveyone's cooler, the amount of mass of the cooler and then ambient temps, then measure all the fuckin fans out there and there's CFM rated capability and so forth. It'd be one hell of a long list.

But either way, at least you are aware and acknowledge TIME is part of the equation, the original reason why I quoted you.

 

[Mussels]

you can't magically make copper faster at dissipating heat.

sure you can! by using it in thin layers with slight imperfections, stacked together and blowing air over them

 

[Deleted member 185158]

sure you can! by using it in thin layers with slight imperfections, stacked together and blowing air over them

Or design the IHS plate and cooler with Silver instead. But that would be one expensive cooler.

 

[freeagent]

Or design the IHS plate and cooler with Silver instead. But that would be one expensive cooler

A little OT sorry... I had an old Harmon/Kardon with a Champagne face (not silver lol) from the late 70s or early 80s when I was in my 20s. Thing weighed liked 65lbs no shit. Anywhoo, the cables in the output stage were silver. The amp had soo much muscle. I brought it to my friends once because his dad had some big Cerwins.. and his dad was gone for a few days. That amp was the perfect match! We were in the basement and pictures were falling off the walls upstairs because of the seismic bass being produced, Quite violent, had to take the glass in the coffee table out because it was trying to flop around

His mom was screaming at the top of her lungs and we had no idea, good times

She told me to take that fucking amp home and never bring it back lol. Sorry you mentioned silver, and I flashed back.

My Friday ramble sorry

 

[Shrek]

sure you can! by using it in thin layers with slight imperfections, stacked together and blowing air over them

I've always wondered about the influence of surface roughness in inducting turbulent flow and so enhancing heat transfer.

 

[freeagent]

I like turbulence, it's good for scrubbing heat away

But now I can't really OC, so I am enjoying the quiet

 

[Mussels]

I've always wondered about the influence of surface roughness in inducting turbulent flow and so enhancing heat transfer.

Any and all increase to surface area helps cooling, but you cant have it so rough that it stops air flowing to the next bits after it. Every bump has an area of reduced airflow behind it, so they gotta be small - golfball style, but even smaller. If you look at the popular coolers that work well, they do various tricks like this, even just angling the fins slightly so the airflow ricochets back and forth in the channel its in

The TRUE120 shows this well, with its "no you didnt drop it, and no that corner isn't melting" look

Their frost tower 120 shows this in an evolution that ups manufacturing costs, but absolutely makes the best of that airflow getting it bouncing all over the place, without excess resistance or noise

 

[Shrek]

I was thinking sandblasting, but don't know how effective that would be.

 

[Count von Schwalbe]

If you inspect a higher-quality (read: commercial -grade) HVAC condenser, you will see that the coil fins often have little sections bent back and forth to reduce laminar flow without dead zones.

 

[Deleted member 185158]

A little OT sorry... I had an old Harmon/Kardon with a Champagne face (not silver lol) from the late 70s or early 80s when I was in my 20s. Thing weighed liked 65lbs no shit. Anywhoo, the cables in the output stage were silver. The amp had soo much muscle. I brought it to my friends once because his dad had some big Cerwins.. and his dad was gone for a few days. That amp was the perfect match! We were in the basement and pictures were falling off the walls upstairs because of the seismic bass being produced, Quite violent, had to take the glass in the coffee table out because it was trying to flop around

His mom was screaming at the top of her lungs and we had no idea, good times

She told me to take that fucking amp home and never bring it back lol. Sorry you mentioned silver, and I flashed back.

My Friday ramble sorry

Nice! Had Technics back in the day, but rocking Sherwood these days pushing some AVIDs. Decent little set up, rocks loud enough I think.
Had a studio amp at one time. Was also a Harmon Kardon. 16 channels of muscles. The heat sinks, which I still have are bigger than an ITX case.

Back OT, I actually have a few silver plates I made from Silver coins. Can't tell because thermal paste is weak transfer therms. This is probably one of the main downfalls of any cooling system unless those like liquid metal for use.

Any and all increase to surface area helps cooling, but you cant have it so rough that it stops air flowing to the next bits after it. Every bump has an area of reduced airflow behind it, so they gotta be small - golfball style, but even smaller. If you look at the popular coolers that work well, they do various tricks like this, even just angling the fins slightly so the airflow ricochets back and forth in the channel its in

The TRUE120 shows this well, with its "no you didnt drop it, and no that corner isn't melting" look
View attachment 272779

Their frost tower 120 shows this in an evolution that ups manufacturing costs, but absolutely makes the best of that airflow getting it bouncing all over the place, without excess resistance or noise
View attachment 272780

Just like the little fins in water blocks. There's been several designs for extra surface area, and that's right where we want to be. One of the reasons I like full copper water blocks. Sure, you can dissipate heat inside, but throw a spot fan on the outside for additional thermal release on the outer surface area.

But none of that changes thermal properties of the material. We've just mastered using it in design we see fit.

Like coating all that copper with nickel plating so it looks good. lol. With 1/3 (roughly) the thermal conductivity of the copper under it..... lol

 

[Count von Schwalbe]

Like coating all that copper with nickel plating so it looks good. lol. With 1/3 (roughly) the thermal conductivity of the copper under it..... lol

Also helps with corrosion.

Anyways, plating runs around .002 thick, so I doubt it is significant.