Yes yes I know, there are a lot of CPUs and they may all have different ideal temperatures and heat tolerences. If that's the case let's talk about AMD cpus, specifically Ryzen CPUs.


I have a Ryzen 5 3600 in a desktop. The specs on their website say it's max operating temp is 95C. it's being cooled by a 240mm radiator with pull fans (although soon I'm going to add push fans to it as well). I have a supply duct coming from my AC unit that supplies cold air to a fan on the top of the tower that ushers in cool air (keep in mind that the supply duct is not attached to the top of the tower but hovers vertically above it about 6 inches to a foot).

My operating temps sometimes rise into the 60s (celsius) but are more commonly in the 30s and 40s and sometimes, usually when I first start up my Desktop in the mornings or I'm not running and CPU-demanding processes, it is running in the high 20s. If I turn on the AC when the system is running in the 50-60s I can get it down to the 30s and high 20s in a moderate amount of time I'd say (5-10mins or so).

Just as a note I've never seen my CPU temp below about 22 or 21C

With all that in mind here are my questions:

1. What is the true ideal temperature for AMD cpus (generally speaking of course)?

2. Is it important that a CPU get hot and stay hot? Like am I damaging my CPU in some way by letting it run more cool?

3. What would happen in an extreme situation where a CPU reached a temperature that was far too cold? OR can that even happen?

4. With what I described above, am I in danger of damaging my CPU in any way?


I've looked online for this information specifically but was unfortunately unable to find such information as would be sufficiently satisfying to my curiousity.


My humble gratitude,

You are over thinking it. All of those temperatures you mention are perfectly acceptable to a PC. The highs of the 60 are great when I encode video with mine it hits the high 60s, low 70s and will run for hour or days at a time like this. The lows you mentions getting are really good I usually have high 30s for a low only using a good air cooler. The never seeing below the 21 or 22 would be normal unless you allow the room temperature to get below that, it has to be higher when running. In short stop worrying you are doing fine with your temperatures if the heat gets excess there are builtin protections on the chips that will throttle the activity on them to cool them down and prevent damage. It gets hot enough, it will just trip the circuits in it and shut the machine down.

1 members found this post helpful.

Thanks! My chief concern is not that my system is getting TOO hot but rather TOO cold; I just wanted to make sure that I'm not up to anything stupid by keeping my temps this cool.

Hi,

My biggest concern when temperature regulation is the delta T or swing in temperatures. If you can keep temperatures constant at or around any manufactures specs then things would be good to go.

You just do not want large swings/changes in temp. Abide by manufactures specs not arbitrary specs from a public query.
I use water cooling for some of my systems and laptop cooler to keep temps constant by providing constant flow.

For laptop coolers I have found by spacing the laptop by about 1/2 inch above cooler I have a better exchange thus cooler.

Hope this helps.
Have fun & enjoy a stable system!

1 members found this post helpful.

I agree with onebuck. In addition I would not use both push and pull fans together. You could actually decrease versus increase the amount of air flow which would cause internal temperatures to rise not fall. On some cases there is a screen on the front which helps to prevent dust from accumulating inside. Excessive dust on fan blades and internal components cause temperatures to rise.

Serious gamers use liquid cooling for CPUs. It amazes me, the lengths to which they will go, and the time they spend on analyzing the efficiency of the cooling pumps, worried about cavitation. Heat is the enemy, and cooler is better, for any electronic device. But I don't believe there is any reason to go overboard, especially for a regular desktop general purpose computer.

All of AMDs current Ryzen CPUs seem to have that 95C max temp spec. I have a Ryzen 2 2200 with Vega graphics and I typically see it running right around 40C. If I throw something computationally heavy at it, I've seen the core temps get up to the low 60s. I don't do gaming on that system so I haven't seen what happens when all the CPU cores and the GPU are running at or near 100%. I suppose that could cause temperatures to approach that stated maximum.

I wouldn't worry about running the CPU at too low of a temperature. Unless you're using a chilled water cooler you'll probably be lucky to see it get much lower than the upper 30s.

I'd be more worried about the internal case temperature getting too high. Disk drives are definitely not happy running when the temperatures get too high---shortens their lifetimes. I lean toward cases with fans that pull in air to blow across the disk drives and other fans to push the hot out. And checking the filters frequently -- our current digs get very dusty for some reason -- to ensure that the airflow is not impeded by blocked filters. Back when I was running 15Krpm SCSI disks, I had to be careful using them inside a cramped computer case lest they cook themselves. I wound up opting for external enclosures that were designed to handle them via higher speed fans.

Aside:

95C is right in the middle of the operating temperature range that one sees for MIL-SPEC grade components. I can recall when someone would be looked at funny for specifying that sort of part for a piece of consumer hardware. :^D


HTH...

There is no ideal temperature, there is a Safe Operating Area.From what I've read, the Ryzen keeps you in it. The thing is, cooling keeps the Ryzen from slowing or turning off cores to stay within the SOA.

You can't have a "too cold" CPU... but of course you may get condensation problems, if running below 0 degree C.

Have been running water cooled here for the past decade or so and never see temps above 35-38 degree C.

1 members found this post helpful.

Commercial temp spec if 0-70ºC; Industrial is -40 to +85ºC; Mil spec is -55 to +125ºC. If you're going below 0ºC you may want to upgrade your temperature range. Most of what you buy is Commercial spec.

in the normal range the actual temperature is irrelevant, you will not be able to detect any kind of problem (including speed and operational issues).
Above/below it the CPU (system) may misbehave, which means it may fail during the execution of something (like accessing RAM/hdd/video or calculation or ...). That will make the system unstable (this is somehow similar to overclocking).
Extrem low or high temperature may stop the CPU completely. Additionally it may cause irreversible damage.

Yes, you can have a CPU too cold to run properly. Most manufacturers do not recommend operation below 45F (around 7C) but milspec devices have a wider range. Go by the mfg recommendation.


I once kept a server room at 63F. I had a MFG service tech try to tell me that that was too cold, it "hid problems" because the components would not fail as quickly at those temperatures (as opposed to 72F he recommended). I still do not understand the advantage in having things fail sooner!

I recommend keeping things cool, but not forcing rapid temperature changes in either direction. Rapid expansion and contraction are bad for almost any material, and circuits and boards are mostly just "dirty glass"!

I have an interesting mix of systems:

Intel i7 7700 that idles around 27c and ramps up to the 40's when gaming. This is in a home built gaming PC with a giant Noctua fan with a passive radiator (air cooled) that is the size of a small block Chevy motor...

2019 MacBook Pro with a quad i5 Intel that idles around mid 30's. Never hits the 50's that I can remember.

2020 Mac mini with a 6 core i7 that starts at boot around 100c and then settles down to idle around mid 30's, low 40's. I have used third party software to ramp the fan on this one up to max to keep it cool, plus run it upside down so the air intake is more "open". During normal use (VM host, surfing, graphics editing), it never gets this hot but just odd that it does this at boot. The boot temp makes me nervous on this one but it never stays at that temp more than 5 seconds or so.

My old HP z800 with 2 6-core Xeons (2.67ghz) would idle around mid 40's and when compiling code, would stay around 75c.

All of these systems are in my home office which is normally around 76-78F in summer, much less in winter. Well, what passes for winter in South Texas anyway...

The semiconductors have a temperature range where they "produce" the expected/designed behavior. [extreme] low or high temperature conditions may/will result in anomalous current/voltage values which will disturb/confuse the digital circuits and at the end will lead to incorrect bits, the unit will misbehave, the system will be unstable.

1 members found this post helpful.

'Too cold' is a bit arbitrary - extreme overclockers have been drowning CPUs in liquid nitrogen (and more recently helium too) for years, although that isn't usually sustainable for every day operation (you can probably find some videos online of how they do this, but basically its just too labor intensive for regular use - there are refrigerant based systems that can sustain sub-zero temperatures for everyday use, but they are quite expensive and exotic). The big issue with 'too cold' in practice is condensation, but that requires fairly low (but not always sub-zero) temperatures to achieve (as Jan K points out) - standard air/water cooling will not do that. The one caveat here is some air/water blocks have a built-in TEC (if you don't know what this is, this likely doesn't apply to you - these are specialized coolers) that can potentially achieve low enough temperatures to condense, but I've also seen some newer iterations of that scheme that intelligently moderate their temperatures to stay just this side of condensing (or at least, they claim to - I'm not spending the kind of money they want for those solutions to find out!).

Anyways, none of the temps you have reported are dangerous and I agree with the general advice to avoid massive swings - 'thermal stress' or 'thermal shock' can be a problem for the underlying PCB even if the actual chip is not destroyed by it. This is a problem 'over the years' not 'all at once' generally speaking. Also keep in mind where/how your radiators are exhausting heat - if they're dumping all that heat onto a bank of capacitors that could also lead to problems eventually (capacitors generally do not like running 'hot' for long periods of time (again, years, not hours) and mfgr usually downrate them relative to increases in temp). Basically the consideration here is twofold - 'too high temps' can kill hardware instantly, but 'not too high but still high' temps can shorten hardware lifespan. Ideally you want your hardware running at or around 'ambient' (that is, low delta T) for as much of its life as possible. Manufacturer #s these days have gotten a little silly because the thermal density is so high (especially for Intel CPUs and AMD GPUs) that they're basically just saying 'yeah sure 95* C is fine 24x7 whatever please buy our product' but the question you should following up with is 'for how many months/years like that?' The Macbook that sevendogsbsd mentioned is an example of that - sure the chip 'can survive' that 100* C burst for a few seconds, but how long would that machine last if it was kept at 100* C 24x7? A completely anecdotal example - I had a dual GPU graphics some years ago that idled around 70* C and would regularly hit load temperatures closer to 100* C - according to the manufacturer this was 100% OK, and the card was indeed OK for around 4 years, and then it died suddenly. It's one of the only videocards I've ever had die, and easily the hottest running. When it was the removed the PCB had visible warping which I also assume was from thermal stress. Yes it had a big giant fan on it, that ran almost constantly. Sure this could all be coincidental (in my sample size of 1), but I also know this is one of a small handful of GPUs I've ever had die, and it was easily the hottest running of any GPU I've ever had.

Finally something else to keep in mind: not all CPU thermal sensors are created equal. I honestly don't know with the newer Ryzen chips, but for a long time AMD's on-die sensors reported non-real units (which were treated as degrees centigrade nonetheless) - so a typical 'ambient idle' may read something absurd like 3* C and the slightest load will see it 'shoot up' to 30-40* C instantly, when in reality its just that the sensor is very inaccurate under some point, and they only worried about accuracy at/near the thermal cut-off point (so it gets more accurate as it gets hotter). So take 'idle temps' with a grain of salt. I think, but don't know for certain, Intel has more 'accurate' sensors in this regard, at least since Pentium 4 era (I remember there was some 'gotcha' about Pentium 3 and temperatures, but I don't remember exactly what it was).