I have an i5 4690k and I am currently stress testing it with Prime95 28.7. Been stress testing it for almost 4 hours. Highest temp. was 96 C. Only reached 96 once. Never exceeded 90 afterwards. Hovered at around 85. Clock speed is 4.5 ghz and 1.3 volts. Is this is safe temperature?
Safe Temperature for CPU
- Thread starter Avac
- Start date
Solution
Chattahoochee
Honorable
- Feb 3, 2014
- 44
- 0
- 10,560
Absolutely not, keep it under 70c
Temperature guide link- http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html
Temperature guide link- http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html
- Oct 30, 2004
- 32,252
- 14
- 99,960
No, not really..... tho it won't immediately cause damage, the CPU will throttle to prevent that. High temperatures like 95C and over for such long periods is generally accepted to be a concern as it weakens / degrades the CPU
P95 is not recommended Haswell and newer chips, while the older versions don't apply the newer instruction sets that are responsible for the extreme temps, it also does nothing to tell you whether your OC is stable since these instruction sets are not present.
Most people will recommend that your CPU not exceed 80C for any significant period if time. I will run mine up to 85C for a few seconds and immediately shut down the stress test utility .... I do this a few times to cycle it to 85C and back down to room temps to set the TIM.
To set up my OC limits, I generally look to maintain 75%C or below, tho I won't really worry unless **short** term peaks get up near 80C under a stress test like RoG Real Bench. This will insure that everyday gaming / workstation usage will stay in the 65 - 70C area.
P95 is not recommended Haswell and newer chips, while the older versions don't apply the newer instruction sets that are responsible for the extreme temps, it also does nothing to tell you whether your OC is stable since these instruction sets are not present.
Most people will recommend that your CPU not exceed 80C for any significant period if time. I will run mine up to 85C for a few seconds and immediately shut down the stress test utility .... I do this a few times to cycle it to 85C and back down to room temps to set the TIM.
To set up my OC limits, I generally look to maintain 75%C or below, tho I won't really worry unless **short** term peaks get up near 80C under a stress test like RoG Real Bench. This will insure that everyday gaming / workstation usage will stay in the 65 - 70C area.
- Oct 30, 2004
- 32,252
- 14
- 99,960
If you download a stress testing program like OCCT, that ahs some auto and recording functions, you will note that the program shuts down when it hits 85C to protect the CPU. The individual circuits on a CPU are separated by very thin silicon which serves to insulate the individual circuit traces. The insulating value of the material deteriorates over time and this rate of deterioration is greatly accelerated as temperature rises.
So while bringing the temp up to 85C for a second and stopping the utility is a risk, it is a small one and it serves a purpose in getting the TIM to cure. Running a load that hits 85C ... 90 C and, as you said, 96C is like driving your car for sustained periods with the tachometer well above the redline.
So while bringing the temp up to 85C for a second and stopping the utility is a risk, it is a small one and it serves a purpose in getting the TIM to cure. Running a load that hits 85C ... 90 C and, as you said, 96C is like driving your car for sustained periods with the tachometer well above the redline.
- Oct 30, 2004
- 32,252
- 14
- 99,960
Again, I don't see the purpose of doing so. If your goal is to go on forums and tell peeps that your system does this w/ P95, then I guess it serves a purpose. But if your goal is to determine stability of an OC with a modern CPU .....
1. What is the purpose of using an old P95 which does not utilize modern instruction sets ? You have proved that your CPU is stable under P95 as long as you don't use all of the added modern instruction sets which are designed to aid CPU intensive programs in modern CPUs. Kinda like testing the off road capability of your new jeep by going off road in 2WD mode.
2. What is the purpose of testing with a "1 type of load at a time" synthetic test like p95 if that same OC can fail under CPU intensive applications especially in a multi tasking environment which P95 does not test..... not the old versions, not the new versions.
3. What is the purpose of testing a synthetic load which no modern application or combination thereof can apply ? If you intend to search for the next mersenne prime, the sure.... but if your intent is to run applications and games, since P95 will make your CPU 5C or more hotter the any application based stress test will result in, why unnecessarily limit your OC because of temp concerns when nothing you will ever run on your computer will approach these temps.
P95 make s sense for testing CPU coolers for example where it applies a consistent non varying heavy load. This is useful for eliminating variables which may occur under testing and is useful for determining cooler performance with reference to other coolers. But using P95 for determining the highest safe and stable overclock just doesn't satisfy the goals for this task. As indicated before, I have had stable P95 OCs fail when using a multi tasking benchmark like Real Bench so stable under P95 does not equal stable under actual applications in a multitasking environment
http://rog.asus.com/forum/showthread.php?43233-Realbench-v2-Discussion-Thread-Download-Links
http://www.hwinfo.com/download.php
http://www.tomshardware.com/answers/id-2387010/intel-4690k-temps-voltage.html
"That's the problem.
Do NOT run any versions of Prime95 later than 26.6. Here's why:
Core i 2nd, 3rd and 4th Generation CPU's have AVX (Advanced Vector Extension) instruction sets. Recent versions of Prime95, such as 28.5, run AVX code on the Floating Point Unit (FPU) math coprocessor, which produces unrealistically high temperatures. The FPU test in the software utility AIDA64 shows the same results.
Prime95 v26.6 produces temperatures on 3rd and 4th Generation processors more consistent with 2nd Generation, which also have AVX instructions, but do not suffer from thermal extremes due to having a soldered Integrated Heat Spreader and a 35% larger Die.
Please download Prime95 version 26.6 - http://windows-downloads-center.blogspot.com/2011/04/pr...
Run only Small FFT’s for 10 minutes.
Use only Real Temp to measure your Core temperatures, as it was designed specifically for Intel processors: Real Temp - http://www.techpowerup.com/downloads/2089/real-temp-3-7...
Your Core temperatures will test 10 to 20C lower with v26.6 than with v28.5.
Please read this Tom’s Sticky: Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temp...
Thanks,
CT
" - CompuTronix (http://www.tomshardware.com/community/profile-111321.htm)Temperatures I get never pass 75.
- Oct 30, 2004
- 32,252
- 14
- 99,960
You quoted my post but what follows rally doesn't address what the post says. I must assume it wasn't presented clearly so lemme take another shot.
I don't dispute that running P95 28.5 has no purpose and can be dangerous. But what I was addressing is what's the point of testing with any version of P95 ? What do you walk away with after the test completes ? Are we trying to find out if the system can run P95 or actually be able to run applications ? Are we happy if we pass P95 and then crash while running say Handbrake ? Generally peeps use Prime95 because they assume that they can ascertain stability of their OC's, no ? It served pretty well in that respect up thru socket 1366. It does not accomplish this goal very well anymore. This is like trying to see if you can run the 100 yard dash below a certain time while running in beach sand.
Using 28.5, as you have identified, you will have drastically limited your OC because of the unrealistic load placed on the CPU, which will never, ever be duplicated outside of synthetic benchmarks. So, yes, what is the point of limiting your OC to 4.3 Ghz, because it gets too hot under p95v28.5, when under real loads, even real multitasking loads, it doesn't get anywhere near those temps ? So the test, is both dangerous and, if you do undertake it, even when completed you have not accomplished you goals of getting the highest stable OC in the environment which the PC will be used. But it will prove the system is stable when those newer instruction sets are present.
Using 26.6, a user will have proved their OC is stable as long as you don't run AVX, SSE2 or any other of the modern instruction sets.... but the user has not proved that the CPU is stable when when those instruction sets are present. So what is the point ? If you pass on 26.6 are you necessarily stable when those instruction sets are present ? No. The test excludes all the features of modern CPUs; would you test /tune a new color TV by watching B&W movies or even Japanese Anime in color ? What will you know about flesh tones afterwards ?
P95 puts a uniform load on the CPU, the analogy being facing a tennis ball machine throwing balls at you from across the court at 98 mph every 4 seconds. What about a multi-tasking environment ? .... 4 tennis ball machines, one at each corner of the court with you in the middle shooting them out at 80 - 98 mph with no consistent interval or sequence ?
What does one walk away with after passing this P95 26.6 test ? Only that your cooling system is / is not adequate for your OC. Other than that, whereas Prime 95 may be useful for getting you build posted on an overclocking site leader board, it does nothing to show the stability of an OC using "real applications".
I have gone 4 hours on 26.6 on multiple builds, only to have my OC attempts humbled by 2-3 minutes of RoG Real Bench. Temps are also substantially higher under 26.6 than they are under RB. Older generations of CPUs were performance limited because of heat. This does not **seem** to be the case anymore, tho I have insufficient data to have 100% confidence since I don't invest 100s of hours in doing OC's. Back w/ SB, I would get to 4.7 or 4.8 within a few hours, now I find myself spending a weekend, but that's as much time as I'm prepared to invest
From my observations, what seems to trip up CPU OCs more than anything else now is not consistent heavy loads but widely varying loads where the system struggles to maintain proper voltage for the associated load from millisecond to millisecond. AVX for example adds 0.10 - 0.13 volts to vcore when present. So w/ 26.6 one has no way of knowing how the system will respond when it gets that AVX voltage bump and then drops it down a millisecond later.
This BTW is why I stopped using RealTemp back in Ivy Bridge era, Real Bench hasn't been updated since SB, almost 4 years ago. HWiNFO is updated weekly (betas) and incremental version upgrades about once a month. Accordingly, I have found it a more reliable and better supported tool for my usage, especially since it provides all monitoring information necessary (temps w/o voltages to my eyes is but half the picture) w/o running multiple monitoring applications. I find it useful to be able to see when AVX kicks in and how high it goes because I am not only limiting my OC by temperature, but also by max allowable voltage. RealTemp does not provide that information. I want to be able to see for example that:
Max temp was 79C but average temp throughout the test of 70C tell me it was very short lived.
Max voltage was 1.51 but average was 1.36 so again, must have been very short lived.
Retesting just the Open CL portion of the test (about 2 minutes) in the above cases lets you view temps and voltages in real time allowing you to see how temps follow voltages and just how long those voltage and temperature spikes last. If it pops up for millisecond 5 or 6 times, I am not concerned.... if it sits there for a solid 2 minutes, I'll lower Vcore.
I have no issue using both if you have the time.... but as passing 26.6 and failing under RB has been an oft repeated experience, I just don't bother anymore as I find it time intensive and doesn't bring anything to the table. I do find P95 useful for setting my TIM tho.... I can quickly gets temps up to 85C and shut it down letting it cool close to room temp in about 2 minutes....after 4 or 5 cycles, can be sure that TIM has cured up to a temperature point it will never see ever gain.
I don't dispute that running P95 28.5 has no purpose and can be dangerous. But what I was addressing is what's the point of testing with any version of P95 ? What do you walk away with after the test completes ? Are we trying to find out if the system can run P95 or actually be able to run applications ? Are we happy if we pass P95 and then crash while running say Handbrake ? Generally peeps use Prime95 because they assume that they can ascertain stability of their OC's, no ? It served pretty well in that respect up thru socket 1366. It does not accomplish this goal very well anymore. This is like trying to see if you can run the 100 yard dash below a certain time while running in beach sand.
Using 28.5, as you have identified, you will have drastically limited your OC because of the unrealistic load placed on the CPU, which will never, ever be duplicated outside of synthetic benchmarks. So, yes, what is the point of limiting your OC to 4.3 Ghz, because it gets too hot under p95v28.5, when under real loads, even real multitasking loads, it doesn't get anywhere near those temps ? So the test, is both dangerous and, if you do undertake it, even when completed you have not accomplished you goals of getting the highest stable OC in the environment which the PC will be used. But it will prove the system is stable when those newer instruction sets are present.
Using 26.6, a user will have proved their OC is stable as long as you don't run AVX, SSE2 or any other of the modern instruction sets.... but the user has not proved that the CPU is stable when when those instruction sets are present. So what is the point ? If you pass on 26.6 are you necessarily stable when those instruction sets are present ? No. The test excludes all the features of modern CPUs; would you test /tune a new color TV by watching B&W movies or even Japanese Anime in color ? What will you know about flesh tones afterwards ?
P95 puts a uniform load on the CPU, the analogy being facing a tennis ball machine throwing balls at you from across the court at 98 mph every 4 seconds. What about a multi-tasking environment ? .... 4 tennis ball machines, one at each corner of the court with you in the middle shooting them out at 80 - 98 mph with no consistent interval or sequence ?
What does one walk away with after passing this P95 26.6 test ? Only that your cooling system is / is not adequate for your OC. Other than that, whereas Prime 95 may be useful for getting you build posted on an overclocking site leader board, it does nothing to show the stability of an OC using "real applications".
I have gone 4 hours on 26.6 on multiple builds, only to have my OC attempts humbled by 2-3 minutes of RoG Real Bench. Temps are also substantially higher under 26.6 than they are under RB. Older generations of CPUs were performance limited because of heat. This does not **seem** to be the case anymore, tho I have insufficient data to have 100% confidence since I don't invest 100s of hours in doing OC's. Back w/ SB, I would get to 4.7 or 4.8 within a few hours, now I find myself spending a weekend, but that's as much time as I'm prepared to invest
From my observations, what seems to trip up CPU OCs more than anything else now is not consistent heavy loads but widely varying loads where the system struggles to maintain proper voltage for the associated load from millisecond to millisecond. AVX for example adds 0.10 - 0.13 volts to vcore when present. So w/ 26.6 one has no way of knowing how the system will respond when it gets that AVX voltage bump and then drops it down a millisecond later.
This BTW is why I stopped using RealTemp back in Ivy Bridge era, Real Bench hasn't been updated since SB, almost 4 years ago. HWiNFO is updated weekly (betas) and incremental version upgrades about once a month. Accordingly, I have found it a more reliable and better supported tool for my usage, especially since it provides all monitoring information necessary (temps w/o voltages to my eyes is but half the picture) w/o running multiple monitoring applications. I find it useful to be able to see when AVX kicks in and how high it goes because I am not only limiting my OC by temperature, but also by max allowable voltage. RealTemp does not provide that information. I want to be able to see for example that:
Max temp was 79C but average temp throughout the test of 70C tell me it was very short lived.
Max voltage was 1.51 but average was 1.36 so again, must have been very short lived.
Retesting just the Open CL portion of the test (about 2 minutes) in the above cases lets you view temps and voltages in real time allowing you to see how temps follow voltages and just how long those voltage and temperature spikes last. If it pops up for millisecond 5 or 6 times, I am not concerned.... if it sits there for a solid 2 minutes, I'll lower Vcore.
I have no issue using both if you have the time.... but as passing 26.6 and failing under RB has been an oft repeated experience, I just don't bother anymore as I find it time intensive and doesn't bring anything to the table. I do find P95 useful for setting my TIM tho.... I can quickly gets temps up to 85C and shut it down letting it cool close to room temp in about 2 minutes....after 4 or 5 cycles, can be sure that TIM has cured up to a temperature point it will never see ever gain.
TRENDING THREADS
-
Question No POST on new AM5 build - - - and the CPU & DRAM lights are on ?
- Started by Uknownflowet
- Replies: 13
-
-
-
Question Is there a consensus 'best' coolant for a custom loop?
- Started by Imacflier
- Replies: 12
-
-
Facebook
Twitter
Instagram