Gamers Red Hot with Fury over Intel Core i7-7700 Temperature Spikes

Megalith

24-bit/48kHz
Staff member
Joined
Aug 20, 2006
Messages
13,000
It seems that Intel’s i7-7700 processors have a flaw where temperatures rise as high as 90°C (194°F) at times, even at stock. This problem is reportedly happening even on liquid cooling setups with the chips running at fairly low voltages. Intel has chimed in to say that these fluctuations are actually totally normal. Have any of you encountered this issue?

Owners of Intel's new i7-7700 processors say the chips have been randomly revving up to extremely high temperatures, and Chipzilla won't give the issue so much as a second look. Reg reader Bastard-Wizard says that he and many other i7-7700 owners are finding that the chips will occasionally kick themselves into overdrive, running at temperatures as high as 90°C (194°F) at times. Intel says the i7-7700 will run at temperatures up to 100°C (212°F). "My own chip suffers from it, (without any overclocking) which is quite an annoyance," our tipster writes. "This despite a delid modification and a proper water loop, resulting in the fans ramping up and down very frequently, and the temperature appearing to frequently spike near the danger zone."
 
How exactly do you "suffer" from 2-3 second spikes of temp below the danger zone for the CPU? That's not enough time for the heat to even propagate through the silcon, much less the headspreader.
Could cause a crash or stuttering in middle of a game cause of throttling. Unless you one of those people that constantly watches temps you would never notice it happening.
 
  • Like
Reactions: DF-1
like this
Well, I suppose the precious "max temp" reading on HWINFO is defiled by such a spike when you're testing your shiny new cooling setup. How can you brag about how cold your new watercooling OC'd rig is when playing Witcher 3 if your chip decides to randomly spike to 90c?? Outrageous! /sarcasm
 
Could cause a crash or stuttering in middle of a game cause of throttling. Unless you one of those people that constantly watches temps you would never notice it happening.

Wait, It could cause a crash...but you would never notice it happening unless you watch temps? Do you realize how that sentence came out?
 
Wait, It could cause a crash...but you would never notice it happening unless you watch temps? Do you realize how that sentence came out?
Yes it could cause a crash but still very unlikely it would.
 
Could cause a crash or stuttering in middle of a game cause of throttling. Unless you one of those people that constantly watches temps you would never notice it happening.

Except the thing doesn't throttle until 100C, and no one has seen spikes bast 90C. And "throttling" means stepping down that hot core's speed a couple hundred Mhz, so it would be unlikely to cause a crash or stutter.
 
How exactly do you "suffer" from 2-3 second spikes of temp below the danger zone for the CPU? That's not enough time for the heat to even propagate through the silcon, much less the headspreader.

Yeah, I'm confused as well...the CPU is jumping 20+ degrees C (assuming a fairly high 60-70C typical temp) and then back down in a matter of seconds? I don't think so.
 
Hopefully this doesn't happen too often. I could see this being a real annoyance with fan curves in many motherboard bios. Some may let you adjust them, others just have fixed settings. Either way, 90c usually results in 100% output to the fan.

Has anyone nailed down any specifics for this one? Does it only happen in Windows?
 
Hopefully this doesn't happen too often. I could see this being a real annoyance with fan curves in many motherboard bios. Some may let you adjust them, others just have fixed settings. Either way, 90c usually results in 100% output to the fan.

Has anyone nailed down any specifics for this one? Does it only happen in Windows?

It happens all. the. time. I open Newegg and it spikes from 30c to 65-70c just loading the images on that site. I swear that it's a problem with the sensors, as it jumps up and back down instantly with load. However, I can't exclude the possibility of bad TIM application. I'm running at only 1.23v, and adaptive volts don't go above 1.28. and temps hit up to 85c under heavy, but normal apps like encoding.

MB has no user input to smooth the spike into a curve, so I *had* to use SpeedFan and test out how CPU load affects the motherboard CPU socket thermometer. Socket gets up to 45-46c on my AIO in a slow, normal fashion. I set max fans when it hits 43 or so.

I'm weighting a delid, but don't want to drop $60 on a use-once tool.

doesn't anyone have a voltage time graph next to a temp time graph?

I can produce one if needed, but the spikes seem related to load more than voltage.

Yeah, I'm confused as well...the CPU is jumping 20+ degrees C (assuming a fairly high 60-70C typical temp) and then back down in a matter of seconds? I don't think so.

30c jumps over the course of a second, yes. Doing simple shit like browsing the web.
 
Last edited:
Could cause a crash or stuttering in middle of a game cause of throttling. Unless you one of those people that constantly watches temps you would never notice it happening.

The CPU doesn't throttle until 100C so that's total bullshit.
 
I have Prime95 and use it as my primary heat testing application for the CPU. Using HWinfo64 I've never seen any spikes on my 7700K. The hottest it ever gets now is 75C with the H80i cooler set to it's quietest mode under full Prime95 load for however long I want to leave it. If I set the cooler to performance mode, load temps never break 72C. It's hot, but not silly hot at all I think. Runs about the same as my GPU in fact, interestingly...
 
I should add that I have NOT tested using IBT or OCCT. IBT is silly in my opinion as it always spits out the most unrealistically high temperatures I've seen of any program ever. 20C over the peak of anything else I run. Obviously flawed since no real life scenario I can generate gets close to that. And I don't like OCCT because it can run spikey due to it's own behavior as observed by other users.
 
If this had come out in December of last year it would have been a foot note because what we're your alternatives? Now that's not the case, Intel should comment on it, they can't just say "it's normal live with it" because if it is it should not be.
 
Starting to sound like faulty sensor input/reporting or interference, isn't it?

I agree, running any stress test I would expect the temp to rise until it hits a maximum but I'm seeing it fluctuate by as much as 15 degrees every few seconds.
The major annoyance here is of course the constant changes in fan speeds this causes and I haven't quite found a way to change my fan profiles to stop this fully.

edit:
added a quick example to show package temp fluctuations over a 1 minute period at near constant CPU load
 

Attachments

  • example.jpg
    example.jpg
    36.8 KB · Views: 46
Last edited:
Could this be the reason the 7xxx series has been having those inconsistent FPS? (For those unaware, some have been complaining of choppiness with the new processors - high highs in FPS, but also low lows)...
 
The major annoyance here is of course the constant changes in fan speeds this causes and I haven't quite found a way to change my fan profiles to stop this fully.

Use the CPUTIN (socket) temp on your motherboard. It's not 1:1 with your cores (my socket maxes at 46c, the cores 85-90c), but it's way more stable than any sensor on the chip itself. See how hot it gets and set fans accordingly.

Set a 2nd curve for the cores, but have that kick in only when cores over 70C.
 
Could this be the reason the 7xxx series has been having those inconsistent FPS? (For those unaware, some have been complaining of choppiness with the new processors - high highs in FPS, but also low lows)...


Probably the lack of cores is that reason.
 
Assuming it isn't an occasional faulty reading, what is the chip doing for a few seconds that it draws enough power to cause the temp spike?
 
Use the CPUTIN (socket) temp on your motherboard. It's not 1:1 with your cores (my socket maxes at 46c, the cores 85-90c), but it's way more stable than any sensor on the chip itself. See how hot it gets and set fans accordingly.

Set a 2nd curve for the cores, but have that kick in only when cores over 70C.

Thanks I'll try that.
 
I wouldn't notice it. I usually push my cpus to the bitter end so they can end up at 90C even without "spikes".
 
Assuming it isn't an occasional faulty reading, what is the chip doing for a few seconds that it draws enough power to cause the temp spike?

Loading web pages. Scrolling through web pages. Occasional background task. You know, things a phone can manage without getting warm.
 
Intel's statement says they don't want you to overclock your CPU that is designed for overclocking. If Intel can provide an explanation of the benefit for having an unlocked multiplier for non-OCing purposes I'd love to hear it. They market it, it must be for a reason.
 
One day, there will be some hardware worth purchasing to replace my 2600k. Obviously today is not that day. Maybe I will have better luck next year. Waiting to upgrade isn't too bad, its not like I haven't been waiting for 6 years or anything...
 
One day, there will be some hardware worth purchasing to replace my 2600k. Obviously today is not that day. Maybe I will have better luck next year. Waiting to upgrade isn't too bad, its not like I haven't been waiting for 6 years or anything...

..right there with ya, buddy...and I know we got lots of company.
 
Loading web pages. Scrolling through web pages. Occasional background task. You know, things a phone can manage without getting warm.

The CHIP isn't getting warm, one sensor is getting warm, probably right next to a core. 2-3 seconds isn't enough TIME for the entire chip to get warm. It is enough time for one small spot of the chip to get that hot under a sudden load.

Your cell phone does the same, you just don't have per-core accurate temp reporting for everything on your phone.
 
The CHIP isn't getting warm, one sensor is getting warm, probably right next to a core. 2-3 seconds isn't enough TIME for the entire chip to get warm. It is enough time for one small spot of the chip to get that hot under a sudden load.

TBH, I'm half convinced it is *solely* a sensor issue. I was running some single threaded app the other day and one core was running at 70c while the other cores idled along at a bit less than 40c, which is about what the MB reported at the socket. You'd think, relatively quickly, that the other cores would heat up as well, but nope. No heat transfer. 30c difference in cores as long as the load persisted. The hot core dropped immediately with load.

It's not a heatsink mounting issue either, as I've done this a couple of times and different cores would spike depending on which one got the thread. They all idle at the exact same temp, not a degree different.
 
When my 7700K was running hot, it turned out to be a BIOS issue with too much voltage. Gigabyte fixed the BIOSes with an update (after I spend $140 on new cooling).
 
TBH, I'm half convinced it is *solely* a sensor issue. I was running some single threaded app the other day and one core was running at 70c while the other cores idled along at a bit less than 40c, which is about what the MB reported at the socket. You'd think, relatively quickly, that the other cores would heat up as well, but nope. No heat transfer. 30c difference in cores as long as the load persisted. The hot core dropped immediately with load.

It's not a heatsink mounting issue either, as I've done this a couple of times and different cores would spike depending on which one got the thread. They all idle at the exact same temp, not a degree different.

Well, if you look at the design, it's very possible they were operating at those temps. Heat will be diffused very rapidly UP into the IHS, simply due to the much larger surface area. It will move from one core to another very slowly since thermally each core only sees each other core on the "edges" of a very, very thin wafer. The better your cooler is, the more likely this is, since the heat will be removed from the IHS faster than it can spread and diffuse back into the "cooler" cores.
 
Back
Top