Can the Cooling capacity (TDP) of a CPU cooler indicate potential CPU temps?

[login211]

I have a Ryzen 5 2400G which has a TDP rating of 65W. I am using the stock Wraith Stealth cooler.

I have manually overclocked it to run at 3.8Ghz but during prime95 the CPU temperature reaches close to 100* C. At regular use (youtube, 30 chrome tabs...) temps are at ~35* C.

I want to buy the Pure rock slim cpu cooler which is rated at 120W TDB.

What kind of temperatures can I expect?

 

[EarthDog]

No way to tell exact temperatures as there are too many variables (case, fans, ambient temps, TIM application, etc). It will be better though!

 

[Mr.Scott]

It won't be that much better.
You'd be better off with a cheap 212 Evo. At least that's rated for 150w.
Count the heatpipes. More is better.

 

[login211]

It won't be that much better.
You'd be better off with a cheap 212 Evo. At least that's rated for 150w.
Count the heat-pipes. More is better.

I'm restricted to a 140mm tall cpu cooler unfortunately.

Anyway, I decided to spend the extra money and get the Noctua NH-U9S instead. Many cheaper coolers out there but I wanted a tower heat-sink and my mini-ITX case limits my options.

Thanks for the replies.

 

[Mandrake4565]

Let us know how it performs, please.

 

[trents]

The stock wraith cooler should do better than you seem to be reporting at a very modest overclock of 3.8 ghz. How did you go about the overclocking? Did you enter manual values or use some auto overclocking genie in bios? What core voltage are you using?

Also, sounds like you might have a compact case. What is the make and model of your case? With an air cooler for the CPU, it is important to have good case ventilation to remove the hot air. Using hot air to do heat exchange is not very efficient. Tell us about your case fans: number, size and position in the case. Better yet, take some pics of the system with the side/top panels off so we can see where the different elements in the system are. Use the Go Advanced tool (lower right corner of any new post window) to attach pics with hour post rather than linking them on a webpage.

Also, it's not necessary to use Prime95 to demonstrate the stability of an overclock. Many people are abandoning Prime95 as a stress tester because it generates such excessively high temps. Try AIDA64 Extreme (run for 5 hrs.) and Realbench (3-4 hr.).

 

[login211]

Let us know how it performs, please.

Will do.

The stock wraith cooler should do better than you seem to be reporting at a very modest overclock of 3.8 ghz. How did you go about the overclocking? Did you enter manual values or use some auto overclocking genie in bios? What core voltage are you using?

Also, sounds like you might have a compact case. What is the make and model of your case? With an air cooler for the CPU, it is important to have good case ventilation to remove the hot air. Using hot air to do heat exchange is not very efficient. Tell us about your case fans: number, size and position in the case. Better yet, take some pics of the system with the side/top panels off so we can see where the different elements in the system are. Use the Go Advanced tool (lower right corner of any new post window) to attach pics with hour post rather than linking them on a webpage.

Also, it's not necessary to use Prime95 to demonstrate the stability of an overclock. Many people are abandoning Prime95 as a stress tester because it generates such excessively high temps. Try AIDA64 Extreme (run for 5 hrs.) and Realbench (3-4 hr.).

For the overclock, I simply changed the multiplier from 36 to 38. No other adjustments were made

I have a Thermaltake Core V1 case. Although airflow is not restricted at all in the case itself, I only have the 200mm stock intake fan at present (the exhausts should be here tomorrow). However what I can tell you is that it's quite cool in my appartment (22 C; it's -7 C outside) and the air that is coming out of the sides of my case (side panels both have vents) is quite cool. I think I might be able to drop CPU temps by probably 10 C when I install the exhaust fans however that still is a bit too high for me. During 3DMark my CPU temps got to 85 C or a bit higher as well.

The temperatures that I'm getting are higher but still not far off from what others are getting with the stock Wraith Stealth cooler. I am also confident in my thermal paste (Arctic silver 5) application. (They will be inline when I install the exhausts.)

I will definitely try the benchmarks that you have suggested but I expect the same temperatures.

 

[trents]

Even though you say you are confident of your TIM application it certainly would not hurt to pull the cooler off and check the spread pattern for thorough contact across the surface. Also AS5 is no longer a great choice for TIM. Not many around this forum use it anymore. It's a rather dated product and there are better choices for the same or less money. One would be Arctic Cooling MX4. Not saying changing to a different TIM product will make a huge difference but may allow you to shave a couple degrees off the temps.

Also, what tool are you using to measure temps? I think Ryzen Master is what people were recommending as other tools may not be reporting temps correctly on these CPUs.

Please create a Signature with a more complete list of your system components. You can refer to mine or to those of other regular contributors for guidance as to what we like to see. The sig tool can be accessed by clicking on Settings in the upper right corner of the page and scrolling down the left side until you see Edit Signature.

 

[Silver Surfer]

Even though you say you are confident of your TIM application it certainly would not hurt to pull the cooler off and check the spread pattern for thorough contact across the surface. Also AS5 is no longer a great choice for TIM. Not many around this forum use it anymore. It's a rather dated product and there are better choices for the same or less money. One would be Arctic Cooling MX4. Not saying changing to a different TIM product will make a huge difference but may allow you to shave a couple degrees off the temps.

Also, what tool are you using to measure temps? I think Ryzen Master is what people were recommending as other tools may not be reporting temps correctly on these CPUs.

Please create a Signature with a more complete list of your system components. You can refer to mine or to those of other regular contributors for guidance as to what we like to see. The sig tool can be accessed by clicking on Settings in the upper right corner of the page and scrolling down the left side until you see Edit Signature.

I agree AS5s matrix requiring the cure in time period is just too old school in this day and time that the best cooling comes from the thinnest TIM application possible to acquire, so you do not need a TIM that actually forms a layer, you literally need the TIM to squish out to nothing, or as close to that as possible.

TIMs true purpose is simply to fill the microscopic imperfections between the 2 mating surfaces, and it does not require much TIM for that.

Finding a TIM to actually do that is what you should look for as trents has suggested Artic Cooling MX-4, I also like Geilid GC Extreme as well, I have both on hand.

I also agree with checking that Thermal Footprint to know 100% for sure what the contact patch actually looks like!

I also like the "quote" in your sig, God Bless You Brother! SS

 

[login211]

Even though you say you are confident of your TIM application it certainly would not hurt to pull the cooler off and check the spread pattern for thorough contact across the surface. Also AS5 is no longer a great choice for TIM. Not many around this forum use it anymore. It's a rather dated product and there are better choices for the same or less money. One would be Arctic Cooling MX4. Not saying changing to a different TIM product will make a huge difference but may allow you to shave a couple degrees off the temps.

Also, what tool are you using to measure temps? I think Ryzen Master is what people were recommending as other tools may not be reporting temps correctly on these CPUs.

Please create a Signature with a more complete list of your system components. You can refer to mine or to those of other regular contributors for guidance as to what we like to see. The sig tool can be accessed by clicking on Settings in the upper right corner of the page and scrolling down the left side until you see Edit Signature.

All the utilities that I used are reporting the same temperatures (Ryzen Master, hwmonitor, gigabyte hardware monitor...).

The amount of thermal paste used, unless it is so much that it is silly, plays little role in temperatures. There are many videos out there to watch about this topic (Linus tech tips, gamersnexus...).

Anyway, I installed 2 80mm exhaust fans. System temps went down by 5-7* C. CPU temps after 10 minutes of prime95 were at 89ish but they were climbing. I think maybe CPU temps would have been 1-2* C lower after a longer prime95 run.

So, the exhaust fans helped stave off high CPU temperatures for a little while.

Only someone who has never held the AMD Wraith Stealth cooler in their hands would think that better CPU temperatures are achievable with it. My temperatures are now perfectly inline with the temperatures I see others getting with this stock cooler which is actually perfectly fine at stock speeds.

I'll be posting a more detailed review of the Noctua NH-U9S which I now have installed but my max CPU temp is now 77* C. This is at 3.9Ghz after 10hr50mins of prime95. The same thermal paste application method was used however I did switch to the Noctua paste as opposed to the AS5 (perhaps it made a difference of +/- 1 or 2* C?).

 

[freeagent]

I would check out Thermalrights True Spirit 140 Power.. Its rated for 360w tdp right out of the box with its quiet TY147A, you can step it up a bit with the addition of a TY143 or two for push/pull. You can find them on amazon for about 50 usd.

 

[ehume]

Manufacturer ratings are what the manufacturers say. They can lie, especially __ (they have yet to take down their "erroneous" specs for a certain fan). You should look at reviews. Reviewers tend to use some of the same CPU's you do -- tough to find a non-Intel CPU mated to a heatsink, though.

As was said above, counting heatpipes is a good indicator of how much heat a heatsink can carry away from a CPU. That is a reliable indicator.

Noctua makes good heatsinks; but why choose an NH-U9S? Why not an NH-C14S? I remain impressed by the sample unit I looked at, back in 2015.

 

[freeagent]

I suppose so. Any manufacturer can say what they want..

Well, 6x8mm is what TS140P has, I think it performs better than Le Grand Macho RT that has 7x6mm, which performs on par with D15, according to reviews.

 

[login211]

Manufacturer ratings are what the manufacturers say. They can lie, especially __ (they have yet to take down their "erroneous" specs for a certain fan). You should look at reviews. Reviewers tend to use some of the same CPU's you do -- tough to find a non-Intel CPU mated to a heatsink, though.

As was said above, counting heatpipes is a good indicator of how much heat a heatsink can carry away from a CPU. That is a reliable indicator.

Noctua makes good heatsinks; but why choose an NH-U9S? Why not an NH-C14S? I remain impressed by the sample unit I looked at, back in 2015.

I have a mini itx case so I'm limited as to my choices.

 

[ehume]

I have a mini itx case so I'm limited as to my choices.

I get that, but the NH-U9S is tall. If an NH-C14S won't fit, perhaps an NH-C12S would fit? But those two would cool better than an NH-U9S.

Review of the U9S here: https://www.overclockers.com/noctua-92mm-heatsink-review

 

[Woomack]

I have something like NH-U9S on TR 1920X (almost the same cooler just different mounting and a bit smaller base) and it works without issues. CPU can OC up to 4.1GHz on this cooler and keeps "safe" temps while CPU itself can reach 300W under load. At stock CPU is rated at 180W where cooler has the same so there is no way it wouldn't be enough for 2400G.
I had my 2200G on Cryorig C7 in ITX case and it was enough after OC up to 3.8GHz (CPU couldn't make much more stable anyway). When I say enough, it doesn't mean it's silent but keeps safe temps. However even U9S should run quiet but if you have enough space in the case then as ehume said, better get a cooler with 120mm fan.

 

[login211]

I get that, but the NH-U9S is tall. If an NH-C14S won't fit, perhaps an NH-C12S would fit? But those two would cool better than an NH-U9S.

Review of the U9S here: https://www.overclockers.com/noctua-92mm-heatsink-review

I considered a variation of the NH-C12 and it would have fit however I was unwilling to spend CAD$103.99 + 13% taxes on a cooler.

The NH-U9S is performing great though: it's barely audible and keeps prime95 torture temps <=75 C. I'm currently in the process of manually setting a lower vcore coupled with LLC=medium in order to reduce temps further. I'll post results at some point.

 

[trents]

All the utilities that I used are reporting the same temperatures (Ryzen Master, hwmonitor, gigabyte hardware monitor...).

The amount of thermal paste used, unless it is so much that it is silly, plays little role in temperatures. There are many videos out there to watch about this topic (Linus tech tips, gamersnexus...).

Anyway, I installed 2 80mm exhaust fans. System temps went down by 5-7* C. CPU temps after 10 minutes of prime95 were at 89ish but they were climbing. I think maybe CPU temps would have been 1-2* C lower after a longer prime95 run.

So, the exhaust fans helped stave off high CPU temperatures for a little while.

Only someone who has never held the AMD Wraith Stealth cooler in their hands would think that better CPU temperatures are achievable with it. My temperatures are now perfectly inline with the temperatures I see others getting with this stock cooler which is actually perfectly fine at stock speeds.

I'll be posting a more detailed review of the Noctua NH-U9S which I now have installed but my max CPU temp is now 77* C. This is at 3.9Ghz after 10hr50mins of prime95. The same thermal paste application method was used however I did switch to the Noctua paste as opposed to the AS5 (perhaps it made a difference of +/- 1 or 2* C?).

I never said anything about "the amount of thermal paste used." I was referring to the completeness of the coverage from front to back and from side to side. The goal is the get even coverage across the contact plane which sometimes doesn't happen for various reasons. As far as the amount of thermal paste used not mattering that is true as long as enough is applied to give full coverage. When very viscous thermal paste products are used you can also apply too much. In such cases, if the clamping pressure is insufficient to cause the TIM to spread, the TIM layer will not thin out sufficiently and become a significant insulating factor. This is not likely to happen with AS5 however, as it is not a viscous product.