Soldato
Has anyone used to mobo ai overclocking yet on there asus mobo?
-
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 Alder Lake owners thread ***
- Thread starter mrk
- Start date
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 29,791
Yes but keep an eye on the voltage it uses, I had to give it a small minus offset to keep it reasonable (1.32v)
Soldato
Just curious I xant overclock at all (used to 8 pack bundles not fiddling .
Soldato
Was just curious to see if anyone had used it to see the results .wish overclockers had of done bundles I'm not a bios person at all .. plug in play that's my moto.
ah the Samsung chip versions. I see the 40/40 ones are hynix. I just missed out last drop
Mine are Samsung IC's
Soldato
Yeah 40’s are Hynix and the 36’s Samsungs. Holding out for the Samsung ones but they go quick
Yeah 40’s are Hynix and the 36’s Samsungs. Holding out for the Samsung ones but they go quick
Mine are the CL40’s and they are Samsung.
Soldato
Strange , what speed are they? My gigabyte memory support page shows the 40’s being Hynix
The GB support page is wrong, there are a few others like that as well.
F5-6000U4040E16GX2-TZ5RK-G.SKILL International Enterprise Co., Ltd. (gskill.com)
Trident Z5 RGB - DDR5-6000MHz CL40-40-40-76 1.30V - 32GB (2x16GB)
Soldato
Both sets of G.Skill 36's I have are Samsung, just finishing doing a 24h+ burn-in test on the second set now at the rated speeds and timings. No errors thus far, and they aren't getting too hot either.
Soldato
Where does it say Samsung ? Interested as it will make my memory hunting easier if both the 36 and 40’s are the same chips.
You asked what speed they are, not does it say Samsung.
See my post here and look at CPU-Z - https://forums.overclockers.co.uk/posts/35293938/
What board are they running on? Also does Aida cache and memory benchmark work fine?
Soldato
lol any particular reason you felt the need to run them for 24 hours. What are you burning in? What would you do if the system flipped a bit at the 23-hour mark?
Some screenshots showing coverage elapsed and system info may be useful for others.
Soldato
It's not my choice, the customer who is taking a test build wanted their system burned in the same manner as the other systems they buy, which means 24 hours on all components, and with new hardware, it is sometimes done 48 hours or as much as a full week.
As I'm sure you are fully aware a system that is not not 100% stable even when realistically run like that maybe 5% if the time is seen as not good enough. When you've worked with commercial, industrial and defence as long as I have you realise the way most people use systems isn't important compared to these settings. NATS wanted a system running for a full 30 days before it could be signed off!
Soldato
That is interesting..!
I have only ever built for myself or friend or family, somewhat different and without having that kind of context it is difficult to appreciate that "burning in" reference.
When I have done, no where near as long, similar type testing on high impact loads etc I can find, equal parts bemusement and irritation, it can then crash when doing next to nothing on the PC. But I'm thinking more of overclocking and if using adaptive offsets on a CPU running through its speed range.
Soldato
Soldato
FYI, no system is 100% stable. In fact, it's possible for a stock system to flip a bit at some point. If ran for 48 hours and an error is found, what then?
If data integrity is essential, overclocking the memory probably isn't the best idea! More so when the system is outside of your control. If the customers enviorment differs enough from your own (temperature, humidety etc) voltage swing during POST can result in impedance mismatches that either fail POST entirely, or the memory will become unstable whilst in the OS. I suppose it shows some commendable support to go the extra length, but for the performance gains in the commercial space it seems like shooting yourself in the foot a little.
Last edited:
So it does, think I will still play it safe and grab the 36’s though as my luck blows
You'll probably be waiting till at least March then
Soldato
Depends on the use-case. As this unit is more a workstation-y type affair then I don't think they expect it to never ever crash, however proof of stability is important. I suppose you should look at it a bit like brakes on a vehicle, you expect them to work 100% of the time, conditions dependent, but a failure at someone point can happen due to component degradation or an integrity failure.
I doubt you'd get far if you just said to big clients, pfft, what is the point in testing systems they'll all fail eventually. Fact of the matter is we see more system fail within 2-72 hours that any other failure, but when you are designing a new build, that is going to potentially land a much larger order then some extra testing, and full transparency is required. All the data is presented to the customer/end user even if that means telling them the system crashed once in 96 hours for example, as long as we can show how we investigated and provide a cause where possible, depending on the customer obviously, and how critical the system is.
Most of the day to day stuff is boring AF, nothing you'll have experience with either, COM express, Embedded, and SBC solutions that don't ever even have a hint of fast RAM or the term OC is mentioned in the UEFI/BIOS. However I work in the development of new systems, whether that be COTS or bespoke it is all tested, I don't build systems for OCUK, as you put it "lol".