What is a safe temp for a 5930K haswell-e?

S

sblantipodi

2[H]4U
Joined
Aug 29, 2010
Messages
3,765
As title.
Asus suggest a 10° lower than tjmax.
In this case 95° could be tolerated as tjmax is 105°.

I think that 95° is way more over the safe value when over clocking.
Am I wrong?

What is a safe temp for haswell-e?
70° on the hottest core? 80°?

How much?
 
For all Intel CPUs, the "maximum safe operating temperature" is 1 degree less than TJ Max.

V70siDK.png


Intel CPUs are designed to run at full speed right up until they reach the TJ Max temperature. Then they throttle just enough to keep the temperature a hair below TJ Max so it is safe. If this was not a safe temperature, it would be easy for Intel to lower TJ Max but they don't.

I agree with Asus. As long as your CPU has not reached the thermal throttling temperature, it is operating within the Intel spec and will be fine. Leaving 5°C or 10°C of wiggle room just to make sure you never reach the throttling temperature is good advice. Intel usually sets the thermal shutdown temperature approximately 25°C beyond the thermal throttling temperature.
 
The chips are tested at TjMax which is 105C. So you can run it that hot.

70C or 80C or whatever is just some nonsense someone invented on the internet that got repeated over and over enough until it became "fact". Same thing with "don't go over XYZ volts or it will melt your house!".
 
but high temps and high volts equal quicker degradation, just buy the Intel tuning plan and push er.
 
But surely a chip that ran at 95°C 24/7 would have a reduced lifespan compared to one that ran at 70°C 24/7?

I think safe just means it won't fail catastrophically or degrade instantly, but says nothing with respect to rate of degradation or chip longevity
 
I can't speak to the MTBF model and whether it takes temperature into account - I suspect it does, but a) you're not going to be running your chip at high temperature 24/7 b) even if it degrades from 12 years to 6 years, do you really care?
 
I abuse the shit out of the Intel CPUs I get ahold of. I run them up to their thermal throttling threshold all the time. So long as they don't throttle and are stable I call it good.
 
dr/owned said:
I can't speak to the MTBF model and whether it takes temperature into account - I suspect it does, but a) you're not going to be running your chip at high temperature 24/7 b) even if it degrades from 12 years to 6 years, do you really care?
Click to expand...

12 years to 6 no, but 4 to 2 yeah I'd care.

I understand your point, and I guess if you upgrade every year then yeah sure you could probably afford to not give a shit about thermals. That being said I don't think it hurts to keep your temps as low as possible.
 
Araxie said:
intel never make public tjmax.. you have to read the whole documentation document, that is because in part the tjmax its the max temperature of the chip without take into consideration throttling point..
Click to expand...

what is the tjmax of the 5930K?
where can I find this docs?
 
Intel does not publicly document TJ Max for any of their desktop CPUs. On the Core i CPUs, the value of TJ Max can be read from a register in each core of each CPU.

Intel uses temperature sensors that count down as they heat up. All that is known is that when one of these sensors counts down to zero, the PROCHOT signal will go active and the CPU will start thermal throttling. RealTemp GT reads the data from this sensor and displays the information as Distance to TJ Max. As long as these sensors are reporting a number greater than zero, your CPU is not throttling so it is running within spec.

If one of these sensors reaches zero, the CPU will rapidly cycle the multiplier down to 8 and it will also drop the core voltage to control the temperature of the CPU to try and prevent it from going beyond the PROCHOT temperature.

Most Intel CPUs have to get approximately 25°C hotter than the PROCHOT temperature before the THERMTRIP temperature is reached. When THERMTRIP goes active, the CPU is designed to shut down almost immediately.
 
x3sphere said:
RealTemp reports 95 C as Tj Max for my 5930k
Click to expand...

10c less than the usual 105c.
Asus suggest to stay 10c less than tjmax, so 85c is the limit temperature suggested by asus.
Keed in consideration that in September we have many degrees less than in august.
So with room temperature of 25c, considering that in summer room temperature could reach 35c, I suppose that my CPU should not go higher than 75c.
Right?
 
As a general rule, temperature doesn't kill a chip, voltage does.
Give a Haswell-E 3V on the Vcore, and it'll die quickly, or immediately. Run it at 90C, and it'll likely be fine for years.

I have an i7 930 that ran in a Lian Li PC-V351 with less-than-optimal cooling (I worry more about noise than heat), and doing 3D renders for hours on end, it would sit at 76C or so. 5 years later, it's now in a render-node, and doing the same thing, only a few C hotter. At least until my Xeon 5650 gets here.

Excessive heat shortens the lifespan, but from what I've read, less than voltage. If you really worry about your chip, you'd worry less about the temperature, and more about the voltage increase when overclocking. :)

And the normal lifespan probably isn't 12 years anyway - it's probably more like 20-30. So overclocking will reduce that lifespan, but more from voltage than heat.
 
Intel chips are really good about this stuff, I had an E6500 dual core in a machine I built for my dad, the heatsink fell off in shipping and it ran fine that way throttling to 200mhz for over a year before he complained it felt slow and I checked it out when I visited. No lasting damage, still overclocks to 3.6ghz and runs like a champ 6-7 years later.
 
LigTasm said:
Intel chips are really good about this stuff, I had an E6500 dual core in a machine I built for my dad, the heatsink fell off in shipping and it ran fine that way throttling to 200mhz for over a year before he complained it felt slow and I checked it out when I visited. No lasting damage, still overclocks to 3.6ghz and runs like a champ 6-7 years later.
Click to expand...
Yep - it's quite hard to kill a modern chip with heat. They'll clock down, keep running and even in a catastrophic failure, will almost always trigger THERMTRIP before permanent damage is done.

Voltage, though... That'll fry those suckers. :)
 
LigTasm said:
Intel chips are really good about this stuff, I had an E6500 dual core in a machine I built for my dad, the heatsink fell off in shipping and it ran fine that way throttling to 200mhz for over a year before he complained it felt slow and I checked it out when I visited. No lasting damage, still overclocks to 3.6ghz and runs like a champ 6-7 years later.
Click to expand...

A coworker brought in a similar EXXX dual core with the heatsink attached but without the springs to push it against the CPU and no thermal compound at all. Anyways he had used it like this for years. I was amazed at how well it ran with this poor heatsink contact..
 
I turned my Tjmax up from 89 to 105C. It's hit 100C under peak load and hasn't exploded, for whatever that's worth.
 
It would degrade quicker , like temps causes issues .
But its really hard that you will step into those zone with new gen cpu and a good cooler unless you oc very high. If your oc'ing very high for 24/7 use ,get yourself a phase changer .
 
Seems to still be doing fine. Yeah I'll hit 105C if I run P95.

Only problem is one of the cores doesn't like running that hot and P95 will error on those threads after a few minutes.
 
dr/owned said:
Seems to still be doing fine. Yeah I'll hit 105C if I run P95.

Only problem is one of the cores doesn't like running that hot and P95 will error on those threads after a few minutes.
Click to expand...

why don't you set your CPU to throttle before?
105C is too hot and unsecure.

what motherboard do you have that doesn't throttle the CPU?
 
Dan_D said:
I abuse the shit out of the Intel CPUs I get ahold of. I run them up to their thermal throttling threshold all the time. So long as they don't throttle and are stable I call it good.
Click to expand...

I'm rereading this old thread because my H80i GT sucks a lot and my 5930k hit 90c on the hottest core with only 1.3v.

I will run that hot, I'll keep you posted if I fry it, in any case I move this post. Very informative and at the end I agree with you.
If Intel says you can run that hot, why care. :D
 
You must log in or register to reply here.
Back
Top