Whats the point of stress tests when all you do is game? (overclocking)

[skiersteve123]

I thought about seeing if my cpu can do 4.8 lately (currently 4.6 6700k for last 3 years) I never did prime 95, occt or aida, Since I only game I just played the hardest to run game that I had (witcher 3) and played it for 5 hours straight, didn't crash, next day lowered voltage played w/e, every day I just lowered voltage until I saw a crash, finally got too low and left voltage at 1.29 and its been fine for the last 3 years, why is prime and other stress tests used so often when gaming is the only thing you do? I ask this as us gamer nerds seem to oc the most for extra fps but don't photoshop or anything so why not just use games as the stress test? just a random question that's been on my mind last few days

 

[montymint]

if a game crashes it could be the GPU as well.

Stress testing JUST the cpu using prime or something like that limits the work to the CPU and gives you confidence that any given failure lies there.

Like using Furmark to stress a GPU, it takes the CPU mostly out of the equation.

 

[Ayahuasca]

Stressing it with those programmes is usually quicker. With games the CPU utilisation can vary wildly and it may be stable in one game but not another.

That being said, if you don't mind your system locking up whilst gaming then testing an overclock on a variety of games is fine.

I've round ASUS Realbench to be the best real world stress test that finds instability fairly quickly when overclocking.

I think most of us are past the point of doing 12-24 hour prime runs before we lock an overclock in. I wonder how many processors have been fried over the years doing that

 

[pc-guy]

The purpose of stress tests is to find any instability in your system.

when you are not overclocking and everything runs at stock speeds, there can be compatibility issues or some of the component of the system may be defective.

When over clicking those aforementioned issues become more likely.

thus you use stress test to find if there are any weaknesses. Each stress test is meant to test each component separately so you can determine which component is the weakest link.

if you only stress test by playing a game you will run into a number of issues. When you overclock, you will be overlooking CPU, RAM, GPU. Running a game will be using all of these components, so if on first run your game crash, how would you know which one is at fault and therefore to tweak the overclock.

another issue with only using game is that you won’t be necessarily addressing all of the RAM, the ram instability can be random or fixed location. If the latter that location maybe at the end of the bank, which usually a game won’t address due to the nature of windows memory management as well as the game may not need the full range of ram to work. So this error may not manifest itself while you solely gaming. But if say one day you are streaming or have teamplay or or whatever chat programme and that memory gets addressed then half way into your stream or raid, you system crashes.

Above are just some limited scenarios of not thoroughly testing each component. There are plenty other pitfalls for not properly stress test such as thermal performance, thermal control and throttling etc etc.

 

[LeMson]

Stability or epeen

 

[Darujhistan]

Community wide survey just in:

51% e-peen, 48% can't resist a fiddle, 1% stress testing for stability issues.

 

[deusexlumina]

It's just cool and makes it seem important.

"48 hr stable prime 95 blend overclock"

In reality if it's stable for what you use it for then who cares?

Some people genuinely need the stability, mind, and in the olden days overclocking was a genuine way to half simulation times etc. but you'd want to be sure you wouldn't have errors mid way through, especially when some runs take multiple days/weeks. This was a small subset of people though.

 

[Tetras]

I thought about seeing if my cpu can do 4.8 lately (currently 4.6 6700k for last 3 years) I never did prime 95, occt or aida, Since I only game I just played the hardest to run game that I had (witcher 3) and played it for 5 hours straight, didn't crash, next day lowered voltage played w/e, every day I just lowered voltage until I saw a crash, finally got too low and left voltage at 1.29 and its been fine for the last 3 years, why is prime and other stress tests used so often when gaming is the only thing you do? I ask this as us gamer nerds seem to oc the most for extra fps but don't photoshop or anything so why not just use games as the stress test? just a random question that's been on my mind last few days

Free performance is great and all, but lock ups and data corruption aren't much fun either. If all you do is game on the PC then you're more willing to tolerate the trade-off of a "just about stable" overclock.

 

[opethdisciple]

Well I'll do a little bit of stability testing just to get some confidence that the CPU/memory or what ever I am overclocking is somewhat stable and then yes it's just time to test out some games.

That's what I did recently. Overclocked my ram. Got it to pass 400% for 80% of ram in hci then started playing games. It needed further voltage tweaks to be game stable.

 

[Things change I've changed]

Personally for me, (when I used to overclock that is), it gave me reassurance that I had a solid and stable system, and it only takes a good day or two of stress testing, if you don't do this and you start having stablity issues or blue screens etc it's hard to know why it is occurring and accurately fault find.

 

[~>Dg<~]

unless its for a job or professional purpose just overclock and do what you do normally. most progs bench situations you will never come close to replicating.

 

[BubbySoup]

My i7-2600k was overclocked for a few years and ran all the games I played just fine. It wasn’t stable benching though, would usually lock up after about 10-15 mins.

I was like, “Oh, ok then” and went back to my games.

 

[Cooper]

..same way people cry over temps & power usage.

Stress tests:
240w!!! 95c!!! Intel sux

Gaming..

80w-100w, 50-60c...no different to AMD, then

 

[MrRockliffe]

A stress test pushes the internals to their limit, more so than any game or application. Hence, if it can survive a stress test, you know it’ll be stable in a game

Ideally you should be testing through a range of scenarios.

 

[Grim5]

..same way people cry over temps & power usage.

Stress tests:
240w!!! 95c!!! Intel sux

Gaming..

80w-100w, 50-60c...no different to AMD, then

5 years later when that cpu is running at 100% load in games, " OMG guiz can someone. Explain why my cpu is that running at 90c in gamezzzz"

 

[pc-guy]

5 years later when that cpu is running at 100% load in games, " OMG guiz can someone. Explain why my cpu is that running at 90c in gamezzzz"

Lol. I think I have seen the YouTube videos of people using those exact words.

anyway, stress test or not. If you are maverick, then live by your sword. No point in saying being cautious and thorough is wrong.

let me put it another way, if you come to sell your parts on MM people often will ask what clocks or frequency you are running and volts etc. You will back these up with system info and screenshots of overclocks. If you haven’t done your tests and sold parts to people and then these parts can’t be tested 100% to be stable with the overclocks, you are effectively mis-selling. On MM market it will earn you a bad rep and a few strikes with the mods and that you won’t be able to trade. On auction, they will ban you also and you end up losing money in postage and returns etc etc.

 

[Rroff]

Personally I consider just "game" stable is actually a bad idea - even if a game doesn't crash while playing you can have issues like files becoming corrupt potentially causing issues with anti-cheat, corruption creeping into save games that might not initially be obvious, maybe even small amounts of corruption in OS components, etc. that eventually causes issues and so on.

 

[Grim5]

Personally I consider just "game" stable is actually a bad idea - even if a game doesn't crash while playing you can have issues like files becoming corrupt potentially causing issues with anti-cheat, corruption creeping into save games that might not initially be obvious, maybe even small amounts of corruption in OS components, etc. that eventually causes issues and so on.

And doesn't certain certain anti cheat/anti piracy software (like Denuvo) use AVX, so if the system isnt stable with AVX load then yes you can still get game crashes.

 

[C64]

Depends on the games some will be okay some will crash mid game last thing you want is be in the middle of a great warzone round and have a random lock up.

I think some games can be as good as a stability app for finding instability.

GTA V seems to be a good one to test for stability

It's true though if it's stable for everything you use it for thats good enough if you don't care about data loss etc

Just when I think ive found the best app to test it crashes on the other one, example ive had real bench gpu test passes fall down in seconds on heaven 4.0 and ive had heaven 4.0 passes fail firestrike

 

[Terminal_Boy]

Depends on the games some will be okay some will crash mid game last thing you want is be in the middle of a great warzone round and have a random lock up.

I think some games can be as good as a stability app for finding instability.

GTA V seems to be a good one to test for stability

It's true though if it's stable for everything you use it for thats good enough if you don't care about data loss etc

If the game is putting a high, constant load on the system it acting exactly as a stree test.