Question Is my 13600k supposed to run this hot when rendering videos?

[wilczur]

Hello, I recently upgraded to a 13600k and was very satisfied with the performance. Games ran noticably better and the temps seemed normal (80C max). Then I tried rendering a video of mine and all hell broke loose; my 9700k would hit around 80C while rendering while this thing would fly up to mid 90s and at peak hit 99C after about 5mins, so I got scared and cancelled the rendering. As far as I know 100C is the max safe temperature and mid 90s seems like flying very close to the sun. I wanted to know if this was normal or should I be worried? What can I do?

My specs are:

Mobo: MSI PRO Z690-A
CPU: Intel i5-13600k
Cooler: Noctua NH-D15
RAM: Corsair Vengeance 32 GB (2 x 16 GB) DDR5-5200 CL40
GPU: MSI GAMING Z TRIO GeForce RTX 3080 10GB LHR
Case: Lancool 2 Mesh

 

[Darkbreeze]

Technically, 100°C IS the max temp you want to see and the system will throttle once it reaches that temp, however, you really don't want to see it hitting that temp if you can avoid it and that's for a couple of reasons.

One, you're literally at the threshold of TJunction/PROCHOT which is where throttling begins BECAUSE you are at the temperature where damage can start occurring, so while it should throttle and not continue on beyond that temperature, repeated or prolonged visits to that end of the thermal envelope are simply not recommended due to the probability of incremental cumulative thermal damage that may be incurred. With only a 13600k there is no way you ought to be seeing those temps with that cooler unless there is a problem with the installation of the cooler in some way, or a bad paste job, or the rendering you are doing is using AVX instructions in which case you might want to go into the BIOS and configure an offset for AVX so that applications using AVX instructions won't tend to unrealistically overload the capabilities of your CPU and cooling system.

What software are you using to render and do you know if it is employing AVX instructions or not?

 

[jnjnilson6]

Hello, I recently upgraded to a 13600k and was very satisfied with the performance. Games ran noticably better and the temps seemed normal (80C max). Then I tried rendering a video of mine and all hell broke loose; my 9700k would hit around 80C while rendering while this thing would fly up to mid 90s and at peak hit 99C after about 5mins, so I got scared and cancelled the rendering. As far as I know 100C is the max safe temperature and mid 90s seems like flying very close to the sun. I wanted to know if this was normal or should I be worried? What can I do?

My specs are:

Mobo: MSI PRO Z690-A
CPU: Intel i5-13600k
Cooler: Noctua NH-D15
RAM: Corsair Vengeance 32 GB (2 x 16 GB) DDR5-5200 CL40
GPU: MSI GAMING Z TRIO GeForce RTX 3080 10GB LHR
Case: Lancool 2 Mesh

Back in the day my Core i7-3770K used to hit 99 C on Cinebench while running at 5 GHz. However it was pretty stable and there was no throttling even in hardcore games. That was also on the best water cooler at the time - the Corsair H110.

You can always try getting a better cooler and making sure the air circulation in your box is good. There is also always the option of downclocking and lowering the voltage, although you'd need to know your way in the BIOS quite well in order to achieve this. (I had changed many settings to get the 3770K up to 5 GHz back in the day, however, I have completely forgotten them.)

 

[wilczur]

Technically, 100°C IS the max temp you want to see and the system will throttle once it reaches that temp, however, you really don't want to see it hitting that temp if you can avoid it and that's for a couple of reasons.

One, you're literally at the threshold of TJunction/PROCHOT which is where throttling begins BECAUSE you are at the temperature where damage can start occurring, so while it should throttle and not continue on beyond that temperature, repeated or prolonged visits to that end of the thermal envelope are simply not recommended due to the probability of incremental cumulative thermal damage that may be incurred. With only a 13600k there is no way you ought to be seeing those temps with that cooler unless there is a problem with the installation of the cooler in some way, or a bad paste job, or the rendering you are doing is using AVX instructions in which case you might want to go into the BIOS and configure an offset for AVX so that applications using AVX instructions won't tend to unrealistically overload the capabilities of your CPU and cooling system.

What software are you using to render and do you know if it is employing AVX instructions or not?

I will have a look at the thermal paste tomorrow but I use Vegas Pro 13 and I have no idea about the AVX thing.

 

[jnjnilson6]

I will have a look at the thermal paste tomorrow but I use Vegas Pro 13 and I have no idea about the AVX thing.

Remember though, your CPU most likely will not let the temperatures go above 99 C and even if they do go up about 60 to 70% of the time you are using the machine it should still last a good 5 years.

However, if you can make it run at lower temperatures that would generally be very good for the CPU. As I've said, lowering the voltage and clock or getting a water cooling system and making sure the box has nice air circulation are some of the things you could think about.

 

[jnjnilson6]

If you can you can also manually set the fan speed to remain at 90 to 100% the entire time through BIOS. If there is any setting regarding the CPU speed like in the image below you should consider putting it on the mildest available preset.

 

[wilczur]

Remember though, your CPU most likely will not let the temperatures go above 99 C and even if they do go up about 60 to 70% of the time you are using the machine it should still last a good 5 years.

However, if you can make it run at lower temperatures that would generally be very good for the CPU. As I've said, lowering the voltage and clock or getting a water cooling system and making sure the box has nice air circulation are some of the things you could think about.

Well I usually ugrade every 3-5 years so that's fine, it's just that I never had a CPU reach these temps before so it scared me and if I can lower the temps to more comfortable ones without doing some leet hax bios stuff that's beyond my understanding then I'd like do that.

 

[Darkbreeze]

What is the CPU cooler fan curve set to in the BIOS? Is it on a preset or did you create a custom curve? At what temperature with the current fan curve is the cooler set to hit 100% PWM signal / full speed?

EXACTLY how many case fans are installed in that case, where is each of them installed and exactly which direction is EACH of them blowing?

Try pulling off the side panel, and running the same operations, to see if you get the same temperatures or if your max temps seem to drop somewhat with the side panel off. This will tell us if there is a case fan problem or if the case fans are not feeding the CPU cooling adequately.

Posting an image of the inside of your case would be helpful as well.

 

[wilczur]

What is the CPU cooler fan curve set to in the BIOS? Is it on a preset or did you create a custom curve? At what temperature with the current fan curve is the cooler set to hit 100% PWM signal / full speed?

EXACTLY how many case fans are installed in that case, where is each of them installed and exactly which direction is EACH of them blowing?

Try pulling off the side panel, and running the same operations, to see if you get the same temperatures or if your max temps seem to drop somewhat with the side panel off. This will tell us if there is a case fan problem or if the case fans are not feeding the CPU cooling adequately.

Posting an image of the inside of your case would be helpful as well.

I repasted the CPU and that did not help much, maybe 2 or 3C less on the top end. I tried again with the side panel off with the same result. When I built this PC I preemptively set the fan curve in BIOS to be more aggressive since my 9700k was already hitting 80C while rendering so I thought additional 6 cores would cause even more heat:

100C = 100%
80C = 90%
60C = 65%
40C = 40%

As for the fan set up, it's pretty standard; 120mm exhaust fan, 2x 140mm fans on the cooler, 2x 140mm fans on the front blowing air into the CPU and GPU; cool air in the front, warm air out the back.

 

[KyaraM]

Did you activate Creator Genie in the BIOS? It's a button next to XMP and what it does is essentially an auto-OC of the CPU; it's easy to find and deactivate, so I would check that first of all. Even 100MHz would be quite a bit of extra heat to dissipate. Also, if you don't want to mess around with BIOS settings too much, you can download XTU. That is Intel's overclocking software, but you can use it for more than that, too. You can, for example, set an undervolt far more easily, and set multiplicators for each core, and set an AVX offset. It's not quite as good as doing it in BIOS, but it should be a good way to experiment with settings. Bonus points for not having to reboot a million times. You can test while the system is running.

 

[wilczur]

Did you activate Creator Genie in the BIOS? It's a button next to XMP and what it does is essentially an auto-OC of the CPU; it's easy to find and deactivate, so I would check that first of all. Even 100MHz would be quite a bit of extra heat to dissipate. Also, if you don't want to mess around with BIOS settings too much, you can download XTU. That is Intel's overclocking software, but you can use it for more than that, too. You can, for example, set an undervolt far more easily, and set multiplicators for each core, and set an AVX offset. It's not quite as good as doing it in BIOS, but it should be a good way to experiment with settings. Bonus points for not having to reboot a million times. You can test while the system is running.

I do not have Creator Genie enabled. I tried using XTU but the undervolting stuff is disabled due to undervoltage protection. I don't know what multiplicators for cores or AVX is.

 

[KyaraM]

I do not have Creator Genie enabled. I tried using XTU but the undervolting stuff is disabled due to undervoltage protection. I don't know what multiplicators for cores or AVX is.

In the "Advanced Tuning"-tab, there should be an option called "AVX Ratio Offset". It is regularly set to -2. You can try decreasing it to -3 or even -4 and see what changes. On the same tab, further below, you can see settings for core ratios. You can decrease them by one and see what happens. That would essentially underclock your CPU by 100MHz each step, the AVX ratio setting only for tasks using AVX instructuins, the core ratio setting in all tasks. About the undervoltage protection being turned on... that's weird. I have the DDR4 version of that board, and I wasn't activated there. You might want to dig for that setting in your BIOS... it will lokely be in the OC tab. You would also configure your BIOS undervolt there.

 

[jnjnilson6]

In the "Advanced Tuning"-tab, there should be an option called "AVX Ratio Offset". It is regularly set to -2. You can try decreasing it to -3 or even -4 and see what changes. On the same tab, further below, you can see settings for core ratios. You can decrease them by one and see what happens. That would essentially underclock your CPU by 100MHz each step, the AVX ratio setting only for tasks using AVX instructuins, the core ratio setting in all tasks. About the undervoltage protection being turned on... that's weird. I have the DDR4 version of that board, and I wasn't activated there. You might want to dig for that setting in your BIOS... it will lokely be in the OC tab. You would also configure your BIOS undervolt there.

Remember some of the Pentium IIIs running only with heatsinks? No fans...
And then came the Pentium 4s and afterwards the Pentium 4 PresHots...
Despite people are saying the Pentium 4s were weak, I think they were regarded as revolutionary and quite powerful for quite some time. Despite the digression, it seems we're entering the days of fastly heating CPUs again. The long anticipated jump from 4 core / 8 thread was totally worthwhile, though; having been instigated most firmly by and past Alder Lake.

 

[wilczur]

In the "Advanced Tuning"-tab, there should be an option called "AVX Ratio Offset". It is regularly set to -2. You can try decreasing it to -3 or even -4 and see what changes. On the same tab, further below, you can see settings for core ratios. You can decrease them by one and see what happens. That would essentially underclock your CPU by 100MHz each step, the AVX ratio setting only for tasks using AVX instructuins, the core ratio setting in all tasks. About the undervoltage protection being turned on... that's weird. I have the DDR4 version of that board, and I wasn't activated there. You might want to dig for that setting in your BIOS... it will lokely be in the OC tab. You would also configure your BIOS undervolt there.

I only have AVX2 Ratio Offset and there is no option for values below 0.0x. As for core ratios, I changed the value from 51x to 50x and it did lower the stress temps by a few degrees. However, isn't this going to decrease the overall performance of the cpu? I don't want to hurt my gaming performance when I can just use task manager to set Vegas Pro's affinity to less cores like I've been doing so far. Also, I tried looking in the BIOS and I didn't see anything about undervoltage protection.

 

[jnjnilson6]

I only have AVX2 Ratio Offset and there is no option for values below 0.0x. As for core ratios, I changed the value from 51x to 50x and it did lower the stress temps by a few degrees. However, isn't this going to decrease the overall performance of the cpu? I don't want to hurt my gaming performance when I can just use task manager to set Vegas Pro's affinity to less cores like I've been doing so far. Also, I tried looking in the BIOS and I didn't see anything about undervoltage protection.

With your CPU a decrease of about 500 MHz should not result to a falling of the framerate by increments higher than 3-4 FPS at a frame rate of 60 FPS, which is negligible and worth it if you can get those temperatures down.

 

[KyaraM]

I only have AVX2 Ratio Offset and there is no option for values below 0.0x. As for core ratios, I changed the value from 51x to 50x and it did lower the stress temps by a few degrees. However, isn't this going to decrease the overall performance of the cpu? I don't want to hurt my gaming performance when I can just use task manager to set Vegas Pro's affinity to less cores like I've been doing so far. Also, I tried looking in the BIOS and I didn't see anything about undervoltage protection.

You can set profiles in that program and switch between them. You could, for example, make one for gaming and one for productivity. As stated above, though, the hit to gaming shouldn't be too big, if it is even noticeable at all. Also, it seems I misremembered how the AVX Offset setting looks. It shows positive numbers (x2.0, x3.0 etc), but applies a negative offset. Apologies for the confusion. Hovering over the "i" behind the setting name should provide some information about the setting, too, and if the offset is negative. Sadly I'm at work right now for a couple more hours, so I cannot look at it myself.

 

[Darkbreeze]

I repasted the CPU and that did not help much, maybe 2 or 3C less on the top end. I tried again with the side panel off with the same result. When I built this PC I preemptively set the fan curve in BIOS to be more aggressive since my 9700k was already hitting 80C while rendering so I thought additional 6 cores would cause even more heat:

100C = 100%
80C = 90%
60C = 65%
40C = 40%

As for the fan set up, it's pretty standard; 120mm exhaust fan, 2x 140mm fans on the cooler, 2x 140mm fans on the front blowing air into the CPU and GPU; cool air in the front, warm air out the back.

That's definitely not how I'd set it, and really doesn't seem aggressive at all. You don't want the fan to only achieve full speed once you've already reached Tjuncture/PROCHOT temps. That will just leave you in a constant state of remaining exactly where you are and pretty much guarantee full time throttling. On CPU air coolers I like to set the lowest point at about 45°C with a fan speed of about 30-35% PWM signal. The second one I like to set at about 55°C with about a 45% PWM signal. The third one at 65°C with a 55% PWM signal and for 12th and 13th Gen Intel CPUs I'm setting the fan to hit full speed at 95°C. For 1st through 11th Gen models I want the fan hitting full speed at 80°C.

It makes no sense to me that you can't keep that 13600k cool with a Noctua NH-D15 because I have a Noctua NH-U14S and mine never even comes close to being a problem even running full stress and heat using Prime95, Realbench, etc. My 12700k has more performance cores than your 13600k and the same total number of cores, with a slightly higher max turbo power TDP and the same base power TDP, so if I can keep mine cool with what amounts essentially to exactly HALF the CPU cooler you have (Which is what the U14S is, is pretty much half of a D14 or D15) you should not be having a problem.

Have you modified the cooler in any way? Are you running both fans that came with it? You are not using the stupid LNA (Low noise adapter) cable extensions that came with the cooler are you? Have you CHANGED the fans that came with the cooler?

Would still like to see a picture of the inside of your case to see how things are configured? Have asked twice for that now.

You can post images as follows:

How to post images on the Tom's hardware forum​

 

[GeneticWeapon]

My MSI motherboard by default power unlocked my 13600k, and let it feed off of 288W. Also, the CPU Lite Load was set to 12 in my bios, by default, when Intel recommends only Lite Load 9. I was seeing temps in the low 90's while stressing my CPU until I changed things.

Go into the overclocking section of your bios, and click on Advanced CPU Configuration, then set the wattage available to a lower amount. I set mine to 250W for short term and long term boost. After that, change your Lite Load settings to mode 9(or lower). If your cooling solution works, your CPU should max out in the 70's from now on.

 

[Darkbreeze]

That's an option, but then you're potentially hampering the expected boost performance when running other things that aren't using AVX instructions, unnecessarily, as setting that limit is going to affect ALL operations, not just the fact that AVX always runs hotter than other types of instructions. If one thing is all that causes problems, then it makes sense to JUST apply a solution that addresses THAT problem, which setting an AVX offset should do. Otherwise, you're using a sledgehammer to fix a problem that could have been fixed with a ball peen.

But, if all else fails, it's certainly an option that should be kept open as a last resort.

 

[GeneticWeapon]

you're potentially hampering the expected boost performance when running other things that aren't using AVX instructions

He isn't getting "expected boosts" though, he's getting much more. His motherboard is by default unlocking power limits and overclocking his CPU. It's an MSI Z690 Pro, that's what they do to bring their boards up high on the performance charts.

Setting CPU Lite Load to mode 9, like I recommended, will put him into "expected boost performance".

Lowering his Lite Load number is literally the easiest and quickest way to bring his temps down and put his CPU back into Intel spec.

https://forum-en.msi.com/faq/article/printer/cpu-lite-load

 

[Darkbreeze]

So "lite load" is just MSI's current "Automatic overclocking" BIOS utility then. No different than ASUS AI overclocking or any other. Figures. They just aren't calling it that, but that's exactly what it is and is technically no different than any previous mechanism for automatic overclocking profiles. They even SAY right in that article that for some models it may not offer a stable enough configuration and BSOD. Lame.

 

[GeneticWeapon]

They even SAY right in that article that for some models it may not offer a stable enough configuration and BSOD. Lame.

Like any overclocking, Lite Load has limits that can be pushed which can eventually cause instability, and crash your system. Like I said previously, Intel recommends Lite Load 9, but my MSI board had it set to 12 by default.

If OP's situation is similar to mine, he can decrease his temps by lowering his Lite Load.

 

[Darkbreeze]

Like any overclocking, Lite Load has limits that can be pushed which can eventually cause instability

So, as I said, there is no difference between THIS and overclocking. Or underclocking for that matter.

 

[wilczur]

That's definitely not how I'd set it, and really doesn't seem aggressive at all. You don't want the fan to only achieve full speed once you've already reached Tjuncture/PROCHOT temps. That will just leave you in a constant state of remaining exactly where you are and pretty much guarantee full time throttling. On CPU air coolers I like to set the lowest point at about 45°C with a fan speed of about 30-35% PWM signal. The second one I like to set at about 55°C with about a 45% PWM signal. The third one at 65°C with a 55% PWM signal and for 12th and 13th Gen Intel CPUs I'm setting the fan to hit full speed at 95°C. For 1st through 11th Gen models I want the fan hitting full speed at 80°C.

It makes no sense to me that you can't keep that 13600k cool with a Noctua NH-D15 because I have a Noctua NH-U14S and mine never even comes close to being a problem even running full stress and heat using Prime95, Realbench, etc. My 12700k has more performance cores than your 13600k and the same total number of cores, with a slightly higher max turbo power TDP and the same base power TDP, so if I can keep mine cool with what amounts essentially to exactly HALF the CPU cooler you have (Which is what the U14S is, is pretty much half of a D14 or D15) you should not be having a problem.

Have you modified the cooler in any way? Are you running both fans that came with it? You are not using the stupid LNA (Low noise adapter) cable extensions that came with the cooler are you? Have you CHANGED the fans that came with the cooler?

Would still like to see a picture of the inside of your case to see how things are configured? Have asked twice for that now.

You can post images as follows:

How to post images on the Tom's hardware forum​

Well the default fan curve was less aggressive than the one I set, and no I am not using those low noise adaptors for my CPU fans since last time I tried the PC would not boot into windows and give me a CPU fan error. As for the cooler, all I did was install those NA-HC4 chromax covers, but I did testeing with and without those and there was no difference in temps.

However, I read somewhere that Vegas Pro 13 (which I was using) has terrible performance, so I switched to VP19 and now the overheating seems to be gone. I rendered a 50 minute video and my CPU temps never exceeded 75C. And since rendering is the most intense thing I do on this PC then I guess I'm good?

 

[Darkbreeze]

I would say so. By all known metrics you're well below any temperature that would be problematic. The newer version probably uses a different AVX instruction set than the older version as there are several different versions of the AVX framework.