[SOLVED] Isn't "OC Stability" a subjective matter?

Feb 6, 2021
4
0
10
0
Hi there, new member here and 1st time poster.

A couple of years ago I decided to build a new PC for myself when my son bought a high-end gaming PC. I didn't want to spend too much money, so I built the system as stated below. When Cyberpunk 2077 came out, I wanted to play it. Needless to say, I would need a huge upgrade, but didn't want to shell out the $$$. Decided to overclock CPU and GPU to see how far I can go.

In terms of "stability", I figure if I run Furmark CPU and GPU stress test for 2 minutes and plateau at a certain temperature without crashing, the system is considered stable? I understand these stress tests may be a bit unrealistic in real life applications, so I don't know if I want to run them for hours at a time. I suspect my system would do fine running Furmark for an hour, but I don't like the idea of the CPU temps staying at 75C - 85C for a long time. I play Cyberpunk 2077 for 1-2 hours at 100% GPU load and it didn't crash at all. Isn't the system considered "stable" if you can run your most demanding software/games for hours without crash? Is it reasonable to push your system beyond its intended purpose?
 
To answer your question I think it could be considered subjective but a better term would be conditional. If it's stable enough to use for what you want to do with it then it's OK, but if you're applying some sort of standard for when it's crashy-ness get's unbearable isn't that forming an objective criteria of sorts?

So on condition that it not crash so often that you want to beat your screen to death with the keyboard, it's OK enough and a fairly subjective standard.

But if you say 'on condition it will run for 4 hours solid with no crashing in my games' (whereupon you take a break and reboot anyway) you just applied a standard so objectively it's a good enough overclock. If you get a new game and add it to the list of games it must stay stable for 4 hours duration yet it crashes in 10 min's you changed the standard so objectively it's now a bad overclock.

Both cases are conditional: one uses subjective criteria, one uses objective criteria.

FWIW, your use of Furmark with a timed duration is fairly objective. But IMO Furmark is a terrible tool: it's akin to saying I want a house strong enough to withstand a thermonuclear blast. It's do-able (Cheyenne Mountain) but do you really need that? I'd use something like Valley for 2 hours at the highest resolution you game at, very hard on the GPU and yet real-world. 3DMark also has a stress test option, but not in the free version I don't think.

I'd want to use something that's based on proven stable routines so you know it's the overclock alone that brings on the crash. Valley and 3DMark fit that, I'm not sure Cyberpunk has got to that level of assurance yet.
 
Last edited:

Colif

Win 11 Master
Moderator
I think it depends on who you trying to prove its stable to.

Games in many ways are better tests than stand alone tests, and if the PC can handle playing Cyberpunk for a few hours without crashing, it might be good enough.

But if you had BSOD and obvious problems, it becomes less subjective.
 
Reactions: CountMike
To answer your question I think it could be considered subjective but a better term would be conditional. If it's stable enough to use for what you want to do with it then it's OK, but if you're applying some sort of standard for when it's crashy-ness get's unbearable isn't that forming an objective criteria of sorts?

So on condition that it not crash so often that you want to beat your screen to death with the keyboard, it's OK enough and a fairly subjective standard.

But if you say 'on condition it will run for 4 hours solid with no crashing in my games' (whereupon you take a break and reboot anyway) you just applied a standard so objectively it's a good enough overclock. If you get a new game and add it to the list of games it must stay stable for 4 hours duration yet it crashes in 10 min's you changed the standard so objectively it's now a bad overclock.

Both cases are conditional: one uses subjective criteria, one uses objective criteria.

FWIW, your use of Furmark with a timed duration is fairly objective. But IMO Furmark is a terrible tool: it's akin to saying I want a house strong enough to withstand a thermonuclear blast. It's do-able (Cheyenne Mountain) but do you really need that? I'd use something like Valley for 2 hours at the highest resolution you game at, very hard on the GPU and yet real-world. 3DMark also has a stress test option, but not in the free version I don't think.

I'd want to use something that's based on proven stable routines so you know it's the overclock alone that brings on the crash. Valley and 3DMark fit that, I'm not sure Cyberpunk has got to that level of assurance yet.
 
Last edited:

CompuTronix

Intel Master
Moderator
kaiustheking,

On behalf of Tom's Moderator Team, welcome aboard!

Concerning the ongoing stability debate, to put it into perspective, keep in mind that with every software crash that occurs through either stress testing, overclocking or normal operation, be it a lockup or the dreaded Blue Screen Of Death (BSOD), each can potentially result in corruption of the Operating System (OS) and / or your valuable data files. Therefore, if you're going to live dangerously, tempt fate or otherwise flirt with disaster (digital Russian Roulette), then it's absolutely critical to maintain a current total system backup, as well as a current backup of your data files.

When stress testing or overclocking, you can use a spare SSD or HDD with a basic software load, so that in the inevitable event of a crash, if the software becomes corrupt, your normal software load remains uncompromised.

Once again, welcome aboard!

CT:sol:
 

ASK THE COMMUNITY