• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

***The Official 8HR Stability 5820k/5930k/5960X Overclocking Thread***

People seem reluctant to post any stability test for 8 hrs this thread proves it... :( I'll take your pints onboard... As for hci memtest thread the test you mention start a thread... Stipulate what you want and I'll do it :)
 
That's debatable though as to if this thread proves anything at all. The whole concept of running AIDA for 8 hours proves a complete lack of understanding of CPU architecture. I've been running my machine since last September and have run AIDA multiple times over. There is little benefit of using it as a 'benchmark' thread especially when excluding cache test and FPU which is where more current is called upon and would potentially knock over a lot of the overclocks in here, so you're back at square one if you're after brownie points ;)

HCI Memtest running one test per thread is the most strenuous memory test, outside of the DOS version(HCI). Especially when really pushing cache and memory speeds simultaneously. Even these aren't completely unconditional, as no overclock is completely stable.

So when push comes to shove if you want to talk about unconditional stability, try really pushing where X99 dominates most with memory ;)

9vJsfB9.png
 
Silent_Scone said:
HCI Memtest running one test per thread is the most strenuous memory test, outside of the DOS version(HCI). Especially when really pushing cache and memory speeds simultaneously. Even these aren't completely unconditional, as no overclock is completely stable.

So when push comes to shove if you want to talk about unconditional stability, try really pushing where X99 dominates most with memory ;)

9vJsfB9.png
Click to expand...

So what happens if you're not overclocking cache or the memory, how does HCI prove further stability over Aida64?
 
ketma said:
So what happens if you're not overclocking cache or the memory, how does HCI prove further stability over Aida64?
Click to expand...

madindehead said:
Yeah, I'd like to know this too. My memory and cache haven't been touched at all.
Click to expand...

Why would you buy a platform inherent for memory bandwidth and not overclock memory lol?

Note memory overclocking is obviously anything over stock SPD 2133
 
Silent_Scone said:
Why would you buy a platform inherent for memory bandwidth and not overclock memory lol?

Note memory overclocking is obviously anything over stock SPD 2133
Click to expand...

Because I read the Anandtech DDR4 scaling article that basically said there are very small minimal in performance above 2400MHz on the RAM. That's why. (Article for those interested http://www.anandtech.com/show/8959/...to-3200-with-gskill-corsair-adata-and-crucial)

I have 2400MHz RAM, could potentially push it higher. But it's not really an issue.

Still, that's not answering the question: What should I stress test my CPU with?
 
Anandtech is using very slow GPU's to test games - other articles on other websites do find that some games show a noticeable performance improvement with higher speed memory.

BF4 being one example where memory speed makes a huge difference.
 
madindehead said:
Yeah, I'd like to know this too. My memory and cache haven't been touched at all.
Click to expand...

madindehead said:
Because I read the Anandtech DDR4 scaling article that basically said there are very small minimal in performance above 2400MHz on the RAM. That's why. (Article for those interested http://www.anandtech.com/show/8959/...to-3200-with-gskill-corsair-adata-and-crucial)

I have 2400MHz RAM, could potentially push it higher. But it's not really an issue.

Still, that's not answering the question: What should I stress test my CPU with?
Click to expand...


HCI Memtest is one of, if not the best tool for eeking out memory errors.
Even if you are running your memory at manufacturer settings or even over clocking them, its still a good idea to thoroughly test your sticks just so you know they are 100%.

As for CPU testing, try Asus realbench. This will stress it far more than AIDA but not as stupidly stress full as prime95, more like a good real life hammering.
Put it through its paces by using the stress option select your max memory and 8hrs. See how you get on with that
 
Dave2150 said:
Anandtech is using very slow GPU's to test games - other articles on other websites do find that some games show a noticeable performance improvement with higher speed memory.

BF4 being one example where memory speed makes a huge difference.
Click to expand...

This was certainly the case for DDR3, is this the same for DDR4? Most games show very little improvement when tested with overclocked RAM.
 
You should try 3200 - 3333 :) It's definitely a lot more "snappier". General use, things will be faster - but you won't lose sleep over it. These speeds are easily achieved and all CPU samples should manage 3000mhz.
 
I'm sure it will be 'snappier' in use and I can appreciate using a more intense stress test for overclocking RAM like HCI but I'm confused by the statements saying Asus Realbench is more taxing than Aida64, - is this confirmed? What about Intel XTU?

I know Prime is a no-go but I'm struggling to see what Realbench brings to the table over an 8hr test run on Aida.
 
See here from Raja at ASUS:

The stress test in Realbench is more difficult to pass than other third party tests such as AIDA. As a result, you may notice higher processor frequencies are possible in other tests. We use Realbench because it is a worst case scenario load and features real-world applications such as Handbrake concurrently with Luxmark while stressing memory to evaluate processor stability. Doing so allows us to gauge frequency expectations more realistically than “lighter load” stress tests.
Click to expand...


AIDA much like other stress tests uses synthetic algorithms of various kinds to determine stability. Basically, it (RB) is a much more realistic work load. You will find people who are stable in IBT and AIDA will sometimes in certain cases fail Realbench almost instantly. One needs to use at least 2, maybe more suites to determine stability along with using the machine.

Cache instability is one of the most erratic of all, and not running cache with this 8 hour stress test is rather futile. It depends whether cache is accessed at point of failure, so not running it for 8 hours obviously greatly shortens the possibility of it falling over lol.

Hence why such a thread proves at best your overclock is not completely unstable.
 
Last edited:
I have been considering overclocking my RAM, but I need to take a look at the settings. Apparently the RAM I have does clock fairly well.

I will sort out Intel XTU and Asus RealBench tonight, and might run one of them overnight.
 
You must log in or register to reply here.
Back
Top Bottom