So long and short of it i OC my ryzen5 1600 tp 3.7ghz with offset +0.031 volts(dont have normal setting to put the likes of 1. 4v etc) . Any higher on speed and it resets the computer on prime95 temp testing and if i try to bump volts up to cope the cpu temp gets to 79 before i stop the test so i guess thats my limit until i upgrade the cooler but thats fine. Anyway prime95 ran small ffts without avx for 20 mins temps capped at 76. Ran a 1hr stress test on realbench and it passed too. (feel free to add anything i might need to check up to this point). Anyway i was reading the next step os probably an 8hr stress test on realbench. Can i start this and go out for thd day or do o really need to monitor this periodically?
New to overclocking. - stress test question
- Thread starter Mikey360
- Start date
As you're probably waiting for an answer I will oblige. Stress testing with any single application only confirms stability with that application and nothing else. A. Prime stable overclock is not necessarily game stable or vice versa. An initial amount of stress testing with Prime or Realbench is always beneficial to confirm general stability but there are deminiting returns from stress testing for hours on end. To me it sounds like your system has proven itself stable enough to use as intended. If you get any stability issues then you can address them when they crop up. To answer your question directly if your temperatures are under control for an hours stress testing they are unlikely to go out of control if you leave the house, so it should be fine. Any issues Windows will most likely stop the process anyhow and / or restart.
I agree with PieEater. I had a machine years ago that usually failed Prime95 after mins or a couple of hours, yet it managed everything else I threw at it without crashing. I know some people like to run Prime95 for hours or days, I don't personally do that, if a machine can manage an hour or two that it's pretty damn stable and unlikely to crash in 'normal' use.
Also hitting 79c isn't that bad, remember this is a program that is deliberately hard on the CPU. I did a 4.3ghz overclock on my Ryzen 3700x that hit high 80s in Prime95, in normal gaming it was nowhere near that hot.
And no, I wouldn't leave the house with Prime95 running, but then I wouldn't be doing an 8 hour test anyway...
Oh when i was reading a guide from tomshardware it basically said you didnt want to push the cpu past 80 on a prime95 15 min test. To clarify i only ran prime95 withkut avx for 15 mins to check temperaturd stability of cpu then moved on to longer realbench tests. I passed an 8hr realbench test on this anyway. But now im currently talking to you on my. Other thread about stability undervolting which is doing weird things lol
The thing is about temps is they are a little open to what people perceive as hot. I've seen people saying ' my pc has been hitting 90c while gaming for the last year, do i have a problem?' - Yes they certainly do, but guaranteed someone will comment and say '90c is within spec for that CPU as it's limit is 100c etc blah blah blah'.
There's no way I would put up with any PC hitting 90c in normal use/gaming. Or 80c. Or even 70c. I like mine to be as low as possible at all times, that's why I'm using a 2011 Coolermaster HAFX full tower that only just fits under my desk and has every optional fan fitted.
Toms hardware says it shouldn't go past 80c on a 15 min test - I think that's confusing in a way as you should hit heat-soak (time to steady temp) in well under that and it should stay at that temp, so it shouldn't really matter how long it runs, it should stay at the same peak with a couple of degrees.
Even the largest air cooler hits steady state under full load in under 2 mins. And even the largest AIO loop should do the same in 4 or 5 mins.
I'd probably go with the fastest overclock you managed with the extra voltage (when it hit 79c how long had it been running?). Run a cinebench test to check that it actually performs better than it does one click lower on speed/voltage. If it's a higher Cinebench score I'd use in normal use/gaming and keep an eye on the temps.
You can always push a bit more when you upgrade the cooler. +0.031 volts isn't much, I needed +0.19 volts offset to hit 4.3ghz all cores on 3700x. I set an alarm in Coretemp to tell me if it hit 75c and it didn't, outside of stress testing.
Oh thanks for the info. Il maybe keep toying with it when i get more free time. It hit 79 degrees at about 13 mins actually but jt had been 77/78 up to that so like you say, probably had hit its point and was varying within a degrees or two.
Again stress testing is not relevant to real world applications, you are very unlikely to reach the temperatures you get whilst stress testing in every day use. So what you are seeing is worst case scenario and is well within acceptable boundaries.
The main reason to control temperatures on Ryzen is that the lower the temps the higher your boost speeds and the longer they will be maintained so good cooling is worthwhile investing in.
The main reason to control temperatures on Ryzen is that the lower the temps the higher your boost speeds and the longer they will be maintained so good cooling is worthwhile investing in.
Maybe on the temps but the voltage pulls are higher on games I hit 1.5 vcore on games and 1.4 in all stress tests
I'd say the stress apps are things to speed up finding the choke points on overclocks and games and day to day use are the fine tuning for an overclock.
asus real bench
heaven 4.0
3d mark fire strike
IBT
prime
and all the others i've had pass then the overclock can fall down in a game
overclocking is a long drawn out process you might find stable and then once in a blue moon getting a crash so you +1 a voltage here or there
you don't need to be running things for 12 hours i've never done this the cpus do tend to have the same sort of areas of clocks and voltages that are stable vs not
Last edited:
yea but I'm making that point in the context of don't use stress apps because it puts unrealistic loads on the system but gaming puts the most realistic load
That's interesting, as I more than likely will go ryzen next upgrade... can you not nail the overclock on so it runs 24/7 at xxGHz then? Not that you'd leave it like that but it's handy to be able to do that with intel chips for oc testing, as in you wouldn't have to take into account whether the chip decides to throttle itself if it goes above a certain temp?
You can run a static all core overclock but there is very little point in my experience. I can get 3% better results in Cinebench at the expense of additional volts, heat and noise and the loss of any chance of the chip boosting to higher clock speeds under lighter loads. There are gains to be had tuning the memory and the better your cooling the better the boost behavior, but old style overclocking on these chips is pretty much dead.
RAM stability - Memtest (you have to reboot into DOS environment - test can be very long. Min 2 pass and usually takes overnight)
CPU and general System Stability - Aida 64 run for a couple of hours and keep an eye on temps. If stable and temps in the mid 70s you are good.
GPU testing - Realbench & Firestrike on loop for about 1hr that way you can tell stability as well as how your cooling is coping.
Prime95 is not a very particularly good bench test for stability anymore. But it is very good at loading your CPU so it is often used to test how good the CPU cooling is. You only need to run that for about 30min to get the CPU hot enough to know your cooling is good or bad.
CPU and general System Stability - Aida 64 run for a couple of hours and keep an eye on temps. If stable and temps in the mid 70s you are good.
GPU testing - Realbench & Firestrike on loop for about 1hr that way you can tell stability as well as how your cooling is coping.
Prime95 is not a very particularly good bench test for stability anymore. But it is very good at loading your CPU so it is often used to test how good the CPU cooling is. You only need to run that for about 30min to get the CPU hot enough to know your cooling is good or bad.