Question 7950x should you purposefully make it go to 95c?

[skaz5]

I have a few questions about my 7950x since I recently just got it. I've understood that it's supposed to run at 95c under load. I currently have it cooled by my peerless assassin, but my temps under load stay around 70-80c and around that range is when my cooler fans start ramping up and getting louder. My question is should i purposefully configure the fans to only kick in when its around 90-95c? and will that have any affect on the longevity whether it's 75c under load or 95c? Also i've seen that ECO mode on 105w only decreases performance by roughly 5%, is it worth switching?

 

[helpstar]

@95°C lifetime should be shorter, but none can say how much or if really
cooler will be better to achieve higher frequencies while under load

 

[Misgar]

My question is should i purposefully configure the fans to only kick in when its around 90-95c?

No, leave the fans as they are. I've set my NH-D15 fans to ramp up to full speed at 60C on my 7950X (I'm not bothered by the noise).

On long video renders using Topaz Video AI, I've seen all cores hovering around 90C with a few hitting 95C. A lot depends on the algorithm used.

If you want to give your CPU a work out, try a Handbrake H265 video conversion or use WinZIP/WinRAR to compress a large folder full of files (several GB). These are realistic loads which should push your CPU hard, or you could run a stress test if you want to watch all cores hit 95C and stay there.

 

[skaz5]

No, leave the fans as they are. I've set my NH-D15 fans to ramp up to full speed at 60C on my 7950X (I'm not bothered by the noise).

On long video renders using Topaz Video AI, I've seen all cores hovering around 90C with a few hitting 95C. A lot depends on the algorithm used.

If you want to give your CPU a work out, try a Handbrake H265 video conversion or use WinZIP/WinRAR to compress a large folder full of files (several GB). These are realistic loads which should push your CPU hard, or you could run a stress test if you want to watch all cores hit 95C and stay there.

Okay, I did run some benchmarks like cinebench etc, and it does go up to 95c and stay there, I work a lot with after effects, does the CPU work better under 90-95c? or is it the same up until that point?

 

[skaz5]

@95°C lifetime should be shorter, but none can say how much or if really
cooler will be better to achieve higher frequencies while under load

even though its intended to run at 90-95c?

 

[helpstar]

if you look into the manufacturer´s head, the intended lifetime is till the warranty ends 😉

the maximum temp. is 95°C, after that, it will throttle itself.

other components will heat up as well, like motherboard or GPU or M.2 and other components soldered on these
electronics: the cooler the better

 

[Misgar]

I work a lot with after effects, does the CPU work better under 90-95c? or is it the same up until that point?

I work a lot with Topaz Video AI and occasionally with Adobe Premiere Pro. When rendering, programs like these can push both the CPU and GPU up to 100%.

In the case of the 7950X, the nominal TDP is quoted as 170W and thermal throttling commences at 95C.

Similarly, if CUDA is tasked by your app during rendering, your GPU may also be pushed up to 100% and its own throttle temperature.

You get the maximum amount of work out of a CPU and a GPU when they're pushed up to their respective thermal limits. In the case of the 7950X, that's when it's running all cores at 95C. For an Intel 14900K it's 100C.

If you want to extract a bit more processing power out of a modern CPU, you need to find some way to make it boost harder, whilst sitting at the same throttle temperature. Modern CPUs are designed to boost as hard as possible, until they reach their maximum allowable temperature. That's when you get the most work out of them. To achieve higher boosts/faster clock frequencies, you need a more efficient cooler to extract more heat and gain marginally shorter render times.

On a big air cooler, the 7950X might dissipate 170W at 4.5GHz all cores. On a big 420mm triple fan AIO/custom loop, the same 7950X might dissipate 200W at 4.65GHz all cores. I've pulled these figures out of thin air, so don't expect them to be 100% accurate, but they serve to illustrate the point. As you increase cooling performance, the maximum boost frequency increases slightly too.

I've performed 36-hour long renders with my 7950X sitting between 90 and 95C, dissipating roughly 170W (Noctua NH-D15). At the same time, my RTX 3060 was running at 100% and 170W continuously. I built the PC in 2022 and have no qualms about using its full capabilities. It's a tool and there to be used. I consider it wasting money if performance is left untapped, e.g. by restricting the CPU's power output, to achieve lower temperatures. If my system survives another 10 years that's fine. If it dies in 4 years, so be it. By 2027 I'll have built a faster system.

I calculated that if I used an RTX 4090, render times would complete in roughly one third the time, i.e. 12 hours instead of 36 hours with my RTX 3060.

https://www.pugetsystems.com/soluti...e-after-effects/hardware-recommendations/#cpu

 

[wyliec2]

All things being equal (ie. clock speed), cooler is better.

I've done thousands (literally) of Handbrake encodes on 3950X, 5950X and now 7950X.

At times, I have had a queue of encodes that kept my system > 90% CPU utilization continuously for several weeks on end.

My target is to keep temps under 80C. As some have noted, the high temps over a long period will increase mobo temps, VRM, etc. Heat and power/voltage are ultimately destructive to electronics.

For my Handbrake CPU encoding, particularly 4K, an all-core OC + under-volt produces good results and typically under 75C. My current default is 4.85 GHz all-core for the 75C target. I could go to 5.2 GHz all-core but then I'd be in to the 85-90C range.

I'm much more comfortable leaving my system running unattended for long periods of time at 75C and taking a few % performance reduction versus pushing the 90-95C range.

FWIW - I have a 420mm AIO cooler and this system is in my basement utility room which is a fairly cool environment.

 

[JayGau]

A CPU doesn't work better because it reaches its max temp. It's actually the opposite. If it reaches its max temp it will start throttling down to not exceed it. The ideal is to have some headroom, like if you stay 5-10° below the max temp under full load you are sure you never throttle down.

If the fan noise is bothering you, you could get a 360mm AIO and set the fan curve with the liquid temperature.

 

[Misgar]

The ideal is to have some headroom, like if you stay 5-10° below the max temp under full load you are sure you never throttle down.

This implies that when a CPU starts to throttle, the BIOS reduces the CPUs speed by a large amount, which would significantly increases render times. It's my understanding that throttling is a more subtle effect, designed stop the CPU from exceeding its rated maximum 90/95C AMD or 100C Intel. Throttling maintains the temperature at 90/95C or 100C.

The result (I believe) is you get the most performance out of a system if you hit throttling. That's why modern CPUs automatically boost up to the limit (when cooling allows) durng heavy computing tasks. The better the cooling, the faster the boosting. If I were to fit a 420mm AIO in place of my NH-D15, I'd expect marginally shorter render times, due to increased cooling at the same throttle temperature.

I liken CPU throttling to the red line on a car/bike rev counter. When you pin the throttle wide open, the engine hits the red line, but does not exceed the maximum specified RPM.

CPU throttling/engine rev imiting is a safety feature designed to make sure you don't exceed the manufacturers' specified design limits. If they were worried you might damage the CPU/engine within the warranty period, they'd reduce the throttle temperature or max rev setting.

For anyone worried about electromigration, this article might help to allay some fears (or not):
https://www.anandtech.com/show/15839/electromigration-amd-ryzen-current-boosting-wont-kill-your-cpu

"As processor manufacturers and semiconductor fabs have iterated through the design of logic gates in CMOS and FinFET processors, there have been active countermeasures put in place to reduce the levels of electromigration (or reduce the effect of the levels of electromigration).

As we shrink process nodes, and voltages decrease, it also becomes less of an issue – the fact that wires also decrease in area has the opposite effect. But as mentioned, the manufacturers now actively take steps to reduce the effect of electromigration inside a processor.

Electromigration has not been an issue for most consumer semiconductor products for a substantial time. The only time I personally have been affected by electromigration issues is when I owned a Sandy Bridge-based 2011 Core i7-2600K, that I used to use for overclocking competitions at 5.1 GHz under some extreme cooling scenarios. It eventually got to a point, after a couple of years, where it needed more voltage to run at stock."

Note: This Anandtech article deals with normal CPU boosting up to the throttle temperature, not deliberately overclocking CPUs to extract the last ounce (28.35g) of performance, e.g. using PBO on an AMD Ryzen. I think I'm correct in saying an overclocked CPU will still throttle at exactly the same temperature as a non-overclocked CPU.

If you're thinking of keeping your new CPU until 2045, by all means limit the temperature to 75C if it it inspires more confidence, but you'll miss out on shorter render times or potentially higher FPS.

I don't run PBO on my 7950X and 3800X, but I do run them up to 95C. I overclock two i7-4770K and a G3258 Pentium quite hard, plus an Athlon 955 and 965 bought in 2006. I still have working CPUs from before the year 2000. It's the motherboards that often die first (burst caps).

When all said and done, how you run your computer is a personal choice and long may it be so.

 

[skaz5]

This implies that when a CPU starts to throttle, the BIOS reduces the CPUs speed by a large amount, which would significantly increases render times. It's my understanding that throttling is a more subtle effect, designed stop the CPU from exceeding its rated maximum 90/95C AMD or 100C Intel. Throttling maintains the temperature at 90/95C or 100C.

The result (I believe) is you get the most performance out of a system if you hit throttling. That's why modern CPUs automatically boost up to the limit (when cooling allows) durng heavy computing tasks. The better the cooling, the faster the boosting. If I were to fit a 420mm AIO in place of my NH-D15, I'd expect marginally shorter render times, due to increased cooling at the same throttle temperature.

I liken CPU throttling to the red line on a car/bike rev counter. When you pin the throttle wide open, the engine hits the red line, but does not exceed the maximum specified RPM.

CPU throttling/engine rev imiting is a safety feature designed to make sure you don't exceed the manufacturers' specified design limits. If they were worried you might damage the CPU/engine within the warranty period, they'd reduce the throttle temperature or max rev setting.

For anyone worried about electromigration, this article might help to allay some fears (or not):
https://www.anandtech.com/show/15839/electromigration-amd-ryzen-current-boosting-wont-kill-your-cpu

"As processor manufacturers and semiconductor fabs have iterated through the design of logic gates in CMOS and FinFET processors, there have been active countermeasures put in place to reduce the levels of electromigration (or reduce the effect of the levels of electromigration).

As we shrink process nodes, and voltages decrease, it also becomes less of an issue – the fact that wires also decrease in area has the opposite effect. But as mentioned, the manufacturers now actively take steps to reduce the effect of electromigration inside a processor.

Electromigration has not been an issue for most consumer semiconductor products for a substantial time. The only time I personally have been affected by electromigration issues is when I owned a Sandy Bridge-based 2011 Core i7-2600K, that I used to use for overclocking competitions at 5.1 GHz under some extreme cooling scenarios. It eventually got to a point, after a couple of years, where it needed more voltage to run at stock."

Note: This Anandtech article deals with normal CPU boosting up to the throttle temperature, not deliberately overclocking CPUs to extract the last ounce (28.35g) of performance, e.g. using PBO on an AMD Ryzen. I think I'm correct in saying an overclocked CPU will still throttle at exactly the same temperature as a non-overclocked CPU.

If you're thinking of keeping your new CPU until 2045, by all means limit the temperature to 75C if it it inspires more confidence, but you'll miss out on shorter render times or potentially higher FPS.

I don't run PBO on my 7950X and 3800X, but I do run them up to 95C. I overclock two i7-4770K and a G3258 Pentium quite hard, plus an Athlon 955 and 965 bought in 2006. I still have working CPUs from before the year 2000. It's the motherboards that often die first (burst caps).

When all said and done, how you run your computer is a personal choice and long may it be so.

I see, the cpu has been used for a year before i bought it, and im thinking about keeping it for 3 years maybe so i'm guessing it won't matter whether it's on 95c under load or 70c. Do you think I should configure my cooler fans to let it go to 90-95c and then make them kick in (under load)?

 

[wyliec2]

I see, the cpu has been used for a year before i bought it, and im thinking about keeping it for 3 years maybe so i'm guessing it won't matter whether it's on 95c under load or 70c. Do you think I should configure my cooler fans to let it go to 90-95c and then make them kick in (under load)?

NO!!!

There is no reason not to let your cooling do all that it can - this will let the CPU boost as high as possible within its design parameters.

Limiting your cooling would only give a greater chance of performance loss due to throttling.

 

[skaz5]

NO!!!

There is no reason not to let your cooling do all that it can - this will let the CPU boost as high as possible within its design parameters.

Limiting your cooling would only give a greater chance of performance loss due to throttling.

okay got it thanks