Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


Hetzner tech says CPU temperature of 100°C is "okay"
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Hetzner tech says CPU temperature of 100°C is "okay"

DanioDanio Member

Recently I've been dabbling with video transcoding on my Hetzner EX41 server, when I noticed that the CPU would run at 100°C and severely throttle. I didn't keep it running for long, but I saw it go from 4.2 GHz down to 3.3 GHz, which is even lower than the base frequency of the CPU.

To their credit, after putting in a ticket Hetzner quickly replaced the fan which improved the situation. Now the CPU can maintain 3.8 to 3.9 GHz under load, but still it would be running at 100°C. According to the technician, a temperature of 100°C is ok, as long as it does not throttle to below base frequency. So there is nothing more they can do for me.

Now, I understand the CPU has thermal protection - so it will throttle to stay under 100°C. But surely it's not a good idea to be running it that hot long term? I don't think it will be very good for the lifespan of the machine. The capacitors on the motherboard are very sensitive to higher temperatures. And I'm losing out on performance...

The CPU is an Intel Core i7 7700.

What do you guys think?

Anyone else with an i7 6700/7700 (Hetzner) server that can share their experiences, what are your temperatures like under load?

«1

Comments

  • LeviLevi Member

    Keep your backups and in case failure just restore when they replace server. It is not your job to monitor temperature or care about it. It is not your hardware and they know better than you what to do. Relax.

  • You are too worried. This is normal when the CPU is fully loaded. Even if it is running for 24 hours, I think the hardware can still run normally because it is server-grade hardware.

  • deankdeank Member, Troll

    100c is okay as long as that's absolute max it is going to reach. T-junction is like 105c.

    If the CPU is in 1U, sometimes you can't do much to lower temp and have to play dangerously.

    Thanked by 3Ouji Danio Pwner
  • hardgamershardgamers Member
    edited July 2020

    From https://www.intel.com/content/www/us/en/support/articles/000005597/processors.html

    Could my processor get damaged from overheating?
    It's unlikely that a processor would get damaged from overheating, due to the operational safeguards in place. Processors have two modes of thermal protection, throttling and automatic shutdown. When a core exceeds the set throttle temperature, it will reduce power to maintain a safe temperature level. The throttle temperature can vary by processor and BIOS settings. If the processor is unable to maintain a safe operating temperature through throttling actions, it will automatically shut down to prevent permanent damage.

    Is it bad if my processor frequently approaches or reaches its maximum temperature?
    Not necessarily. Many Intel® processors make use of Intel® Turbo Boost Technology, which allows them to operate at very high frequency for a short amount of time. When the processor is operating at or near its maximum frequency it's possible for the temperature to climb very rapidly and quickly reach its maximum temperature. ** In sustained workloads, it's possible the processor will operate at or near its maximum temperature limit. Being at maximum temperature while running a workload isn't necessarily cause for concern. Intel processors constantly monitor their temperature and can very rapidly adjust their frequency and power consumption to prevent overheating and damage.**

    Thanked by 1Danio
  • Hetzner techs probably have in-house betting pools for these kind of stats.
    They'd know their gear best.

    Thanked by 1Hetzner_OL
  • deankdeank Member, Troll

    I know that people on PC enthusiast forums swear by a statement that lower temp is better.

    Perhaps true, but not necessarily in products' lifetime (10 years). The chips and its hardware are generally designed to withstand over 100c as long as they are server(industrial)-grade.

    Only capacitors are really susceptible to temp, and server-grade ones handle well over 100c.

    Thanked by 1vimalware
  • NeoonNeoon Community Contributor, Veteran

    Well, see, you pay monthly rent, if it breaks, Hetzner replaces it.
    As long it does not throttle, I would not gave a single fuck.

    But it likely does at 100 celsia.

  • LeviLevi Member

    @GodSpeed said: it is server-grade hardware

    i7 is not server grade. It is desktop component based.

    Thanked by 1kkrajk
  • NeoonNeoon Community Contributor, Veteran

    @LTniger said:

    @GodSpeed said: it is server-grade hardware

    i7 is not server grade. It is desktop component based.

    Well, if you put that i7 in a 1U case its, server authorized grade.
    Same you put a indoor wifi access point into a plastic bag and put it outside.

    Thanked by 1Lm85H4gFkh3wk3
  • NyrNyr Community Contributor, Veteran

    While I wouldn't want that temperature in my office during summer, what's the problem if it's fine for Hetzner in their data center?

    Thanked by 1webcraft
  • @LTniger said:

    @GodSpeed said: it is server-grade hardware

    i7 is not server grade. It is desktop component based.

    "server grade" is just a joke tbh. alot of the time its the same silicon just with some traces cutoff to remove features and OCCASIONALLY better chip binning.

    Thanked by 1kkrajk
  • @deank said:
    100c is okay as long as that's absolute max it is going to reach. T-junction is like 105c.

    If the CPU is in 1U, sometimes you can't do much to lower temp and have to play dangerously.

    7700 Tjunction is 100c.

    Running at constant max temps will shorten the CPU life but it shouldn't give out just like that.

  • deankdeank Member, Troll
    edited July 2020

    I've had a P4 chip that ran for 14 years at near 100c. Caps on its motherboards gave out much earlier than the chip did. Actually, the chip never died. It was retired.

    No need to worry about shortening a chip's lifespan. Mindless OCing will shorten a chip's lifespan for sure though.

  • serv_eeserv_ee Member
    edited July 2020

    To be fair youre comparing apples to oranges here.

    Same could be said that my FX9590 maxed its lifespan at 2 years because it wasnt properly cooled (AIO suggested, 140mm fans on an Noctua actual).

    The hotter the chip runs the sooner the die or IHS gives out. Im not saying it will be tomorrow or even a year but sooner or later it will.

    Cant remember which generation Intel CPUs those were but they already had problems with their IHS when they decided to use bird poop instead of proper glue. Google results came in with almost every generation from Sandy and up but I dont think thats the case.

  • deankdeank Member, Troll

    I've personally never seen a chip die due to high temp. There are, after all, automatic safety switches on motherboards. I've seen chips burn out due to OCing though because dudes disabled those safety measures.

    Motherboards will likely die before its CPU does unless you are pumping voltage into a chip.

  • serv_eeserv_ee Member
    edited July 2020

    I have had one die on me but that was mainly me being stupid to buy that chip in the first place.

    Then again I had an Dell server at home with two Opterons which were passive cooled and nothing ever happened to them.

    So... ¯_(ツ)_/¯

  • DanioDanio Member

    Indeed it is pretty hard to kill a CPU. One way is from electromigration, due to giving a higher voltage to overclock. Over time, the chip will need more voltage to maintain the same clock speed, or outright not being stable anymore at those clock speeds that worked fine before.

    @serv_ee said:
    Cant remember which generation Intel CPUs those were but they already had problems with their IHS when they decided to use bird poop instead of proper glue. Google results came in with almost every generation from Sandy and up but I dont think thats the case.

    The Intel bird poop era lasted from 3rd gen up until 8th gen. As of 9th gen they are soldering the IHS again. :)

  • @Danio said:
    Indeed it is pretty hard to kill a CPU. One way is from electromigration, due to giving a higher voltage to overclock. Over time, the chip will need more voltage to maintain the same clock speed, or outright not being stable anymore at those clock speeds that worked fine before.

    @serv_ee said:
    Cant remember which generation Intel CPUs those were but they already had problems with their IHS when they decided to use bird poop instead of proper glue. Google results came in with almost every generation from Sandy and up but I dont think thats the case.

    The Intel bird poop era lasted from 3rd gen up until 8th gen. As of 9th gen they are soldering the IHS again. :)

    Yea, soldered not glued, sorry.

  • jlayjlay Member

    The system will likely survive, CPUs are really good these days about controlling current in respect to temperature and so on. With that said, I'd probably press the issue. Either get them to fix it, or get a replacement system (perhaps in one of their other DCs for assurances).

    Reason being, the moment dust collects in your server or the units around yours are also under extreme load, chances are yours will start thermal throttling again.

    Thanked by 1Danio
  • It may be a bit surprising how hot systems can run and still function fine. I've had to use gloves to de-rack systems before. It wouldn't surprise me if the cpus in them had hit 100C.

    It's not ideal. But when you're running a low end datacenter you don't aim for a chilly datacenter. If your balls aren't sweating, it's a good day.

    Let the DC worry about temps and hardware failure if they own the hardware. If the temp is low enough outside, we shut off ALL CRACs in the winter and run on external air only to get back in the black.

  • @Neoon said:

    @LTniger said:

    @GodSpeed said: it is server-grade hardware

    i7 is not server grade. It is desktop component based.

    Well, if you put that i7 in a 1U case its, server authorized grade.
    Same you put a indoor wifi access point into a plastic bag and put it outside.

    If it was an IP67 plastic bag and still magically allowed heat dissipation. I know you're just trying to make a joke, but, no.

  • NeoonNeoon Community Contributor, Veteran

    @TimboJones said:

    @Neoon said:

    @LTniger said:

    @GodSpeed said: it is server-grade hardware

    i7 is not server grade. It is desktop component based.

    Well, if you put that i7 in a 1U case its, server authorized grade.
    Same you put a indoor wifi access point into a plastic bag and put it outside.

    If it was an IP67 plastic bag and still magically allowed heat dissipation. I know you're just trying to make a joke, but, no.

    Its not a joke, people did this and it worked.

  • @Neoon said:

    @TimboJones said:

    @Neoon said:

    @LTniger said:

    @GodSpeed said: it is server-grade hardware

    i7 is not server grade. It is desktop component based.

    Well, if you put that i7 in a 1U case its, server authorized grade.
    Same you put a indoor wifi access point into a plastic bag and put it outside.

    If it was an IP67 plastic bag and still magically allowed heat dissipation. I know you're just trying to make a joke, but, no.

    Its not a joke, people did this and it worked.

    "server authorized grade." Isn't a thing, anyway. But "worked" would be for a very limited time. I was really balking at the idea it would be enterprise grade in a plastic bag.

  • Adam1Adam1 Member

    You could try reducing voltage a little, I do this on my laptop (-100mV) to keep the temps sensible (-10-15c). On windows I use throttlestop.

  • SwiftnodeSwiftnode Member, Host Rep

    @serv_ee said:
    The hotter the chip runs the sooner the die or IHS gives out.

    I highly doubt the IHS is going anywhere. The gasket/sealant used to secure the IHS is rated way above the thermal cut off of any Intel chip.

    @Danio said:
    To their credit, after putting in a ticket Hetzner quickly replaced the fan which improved the situation. Now the CPU can maintain 3.8 to 3.9 GHz under load, but still it would be running at 100°C.

    If it throttles again, ask them to disable BD_PROCHOT in bios, or you can try to do it with some software. (eg. ThrottleStop) That will keep the chip from lowering the clock under high temp, won't save you from a shutdown if the chip hits it's thermal cutoff though.

  • FHRFHR Member, Host Rep

    I would agree with Hetzner - as long as you're getting at least base clock, you're getting what you're paying for. The temperature doesn't need to concern you, it's not your hardware.

    Thanked by 1Stradale
  • CConnerCConner Member, Host Rep
    edited July 2020

    @FHR said:
    I would agree with Hetzner - as long as you're getting at least base clock, you're getting what you're paying for. The temperature doesn't need to concern you, it's not your hardware.

    No where is stated that you're paying for just the base clock. The boost clock is a standard feature of the CPU and can be considered false advertisement if it is not able to boost all of the time whilst customers are not being made aware of that. They're also listing CPU benchmark scores with their machines that did not have their CPU's boost clock disabled.

  • This is on you. Your using a 7700k desktop processor for heavy transcoding. It's stored in a 1u rack around a bunch of other hardware. It's hot there. But When getting a dedicated server assume your colocating but without worrying about replacing/repair hardware. Would you set this up for your use case? Probably not and if you did you would understand it's limitations. Well this is the limitation it won't go above base freq in such an environment. At least you don't have to worry about replacing those 1u fans once they burn out from running at max so much. Not to mention never having you hear the suckers. OMG they must be so stupid loud!!

  • MikeAMikeA Member, Patron Provider

    @sureiam said:
    This is on you. Your using a 7700k desktop processor for heavy transcoding. It's stored in a 1u rack around a bunch of other hardware. It's hot there. But When getting a dedicated server assume your colocating but without worrying about replacing/repair hardware. Would you set this up for your use case? Probably not and if you did you would understand it's limitations. Well this is the limitation it won't go above base freq in such an environment. At least you don't have to worry about replacing those 1u fans once they burn out from running at max so much. Not to mention never having you hear the suckers. OMG they must be so stupid loud!!

    It being a desktop CPU doesn't matter, you can cool those no problem in a 1U, especially a 4 core one.

Sign In or Register to comment.