what are the types to tests to run to test for ram stability?

[blackestvoid]

building new ryzen for myself (myself = friend)
ram will be oc
can run the tests on sata hdd win 7 os while waiting for other components to arrive? still waiting for ssd and graphics card to arrive, borrowed from existing system
if everything appears stable, is there a need to run the tests again after changing to ssd on win 10 and new graphics cards?

which tests to run?
memtest64 vs memtest86 vs memtest86+
SuperPi
Prime95 - is this for cpu or ram?
anymore?
understand some of them are from x86 era, still suitable?
is there a sequence which to run 1st or is it even necessary to run all or just one of them is enough?

also heard there is a cold boot test, seems like some rams are unable to cold boot when showing stable for the tests, true?

 

[singupingu]

Mind listing down your existing components, your new components and your otw components?

Sent from Finally Unbanned using GAGT

 

[blackestvoid]

x370 taichi running trident z rgb 3200 trying for 3466

borrowed for testing
gtx1050
wd black hdd with win 7

incoming
gtx1080, maybe with ti - undecided
samsung evo 960 nvme with win 10

 

[MrAwesomeX]

You do know all your answers can be Googled .

 

[singupingu]

What cpu model?

You will need to fresh install windows nonetheless. You can't just swap major components and motherboard without reinstalling. Even if you can, performance won't be as good as fresh install.

Fresh install a win 10 using windows creation tool saved on a thumb drive, activate it with code purchased from carousell if you want coz it will be bound to motherboard anyway. Flash the latest bios for your motherboard.

Use memtest64 to test the ram out of the box. After which then you oc the ram and try with memtest64 again.

Prime95 and superpi use to test cpu more than ram although ram utilization plays a part as well. Do test them too. To see how much difference your oc made, you can use superpi or cinebench. Use Prime95 to see if it is stable or not for 30 min to 2 hours. In sequence, memtest64, prime95, superpi and cinebench.

After all said and done and stable liao, then after you get the ssd, you mirror a copy over. Set boot priority to ssd from bios, then reformat the hdd and return it back.

Sent from Finally Unbanned using GAGT

 

[Encrypted11]

Not a ryzen user, but some good resources
1. http://www.overclock.net/t/1628751/official-amd-ryzen-ddr4-24-7-memory-stability-thread

2. "KIV" (as per oc memory stability expert who teach me how hlots123 secondary, third and forth timings too so satki ) Raja@asus recommendations
https://rog.asus.com/forum/showthread.php?73665-Our-preferred-memory-stress-test
http://www.overclock.net/t/1569364/official-intel-ddr4-24-7-memory-stability-thread/60#post_24371115

Stressapp is more stressful than HCI for testing memory alone. HCI's load is a bit more spread to the cache, while Stressapp is more memory bus isolating. They both have their use, but if testing memory alone, then Stressapp is a quicker way of finding errors.

3. Look for "The stilt"'s Ryzen timings, start from those.
He has improved generic profiles for Samsung & Hynix memory both single/double side. Some of them are even placed in the C6H official ram presets.

4. Make sure you don't stress test your ram in your os disk. Do it in some isolated DOS or portable linux environment unless its remotely stable. There's a good risk of corrupting data especially the methods like running GSAT within Windows via ubuntu bash on potentially unstable memory.

Extra, test your CPU side OC separately if not its harder to isolate OC problems. Introduce p states the last if necessary, stay away from overclocking refclk (bclk).
If you overclock the bus, PCIe is overclocked as well and can introduce data corruption for NVMe ssds as per stilt in Ryzen strictly technical.

 

[blackestvoid]

You do know all your answers can be Googled .

google is now a history search engine, many old results
very difficult to search for something recent
can give you results like ram test
seems not many using latest bios or google not reflecting these results

 

[blackestvoid]

What cpu model?

You will need to fresh install windows nonetheless. You can't just swap major components and motherboard without reinstalling. Even if you can, performance won't be as good as fresh install.

Fresh install a win 10 using windows creation tool saved on a thumb drive, activate it with code purchased from carousell if you want coz it will be bound to motherboard anyway. Flash the latest bios for your motherboard.

Use memtest64 to test the ram out of the box. After which then you oc the ram and try with memtest64 again.

Prime95 and superpi use to test cpu more than ram although ram utilization plays a part as well. Do test them too. To see how much difference your oc made, you can use superpi or cinebench. Use Prime95 to see if it is stable or not for 30 min to 2 hours. In sequence, memtest64, prime95, superpi and cinebench.

After all said and done and stable liao, then after you get the ssd, you mirror a copy over. Set boot priority to ssd from bios, then reformat the hdd and return it back.

Sent from Finally Unbanned using GAGT

its not about os
i want to see if ram would still be stable if passing all the tests in one setup vs after changing hardware components
do different components affect ram? eg change graphics card, ssd, but motherboard and cpu no change?

 

[blackestvoid]

Not a ryzen user, but some good resources
1. http://www.overclock.net/t/1628751/official-amd-ryzen-ddr4-24-7-memory-stability-thread

2. "KIV" (as per oc memory stability expert who teach me how hlots123 secondary, third and forth timings too so satki ) Raja@asus recommendations
https://rog.asus.com/forum/showthread.php?73665-Our-preferred-memory-stress-test
http://www.overclock.net/t/1569364/official-intel-ddr4-24-7-memory-stability-thread/60#post_24371115


3. Look for "The stilt"'s Ryzen timings, start from those.
He has improved generic profiles for Samsung & Hynix memory both single/double side. Some of them are even placed in the C6H official ram presets.

4. Make sure you don't stress test your ram in your os disk. Do it in some isolated DOS or portable linux environment unless its remotely stable. There's a good risk of corrupting data especially the methods like running GSAT within Windows via ubuntu bash on potentially unstable memory.

Extra, test your CPU side OC separately if not its harder to isolate OC problems. Introduce p states the last if necessary, stay away from overclocking refclk (bclk).
If you overclock the bus, PCIe is overclocked as well and can introduce data corruption for NVMe ssds as per stilt in Ryzen strictly technical.

even more tests to add to the mix
this is a new so if there us to be any problems, best is to corrupt the OS now and get detected rather than later
the question is if oc ram currently pass all tests in one setup, just changing graphics card alone, would ram become unstable?

 

[singupingu]

Changing gpu, ssd, hdd won't affect your ram oc. As long the cpu, mobo, ram untouched, change anything you want

Sent from Finally Unbanned using GAGT

 

[ragnarok95]

You do know all your answers can be Googled .

Yes can be googled. Then what’s the use of this forum???

 

[Encrypted11]

When you go to the extreme edge of your ram's capability, swapping rams to different slots can have an effect on stability. (Basically matching RAM to slots) That might be on your part swap concern but gpu shouldn't change it.

If you want a good ram test stressapptest is enough (explained by Raja). Recommendation by him is 1 hour but I run it overnight.

And like you said, test your rams out of cold booting state because you will turn it off and cold boot at some point and cold boot memory training is different.
http://www.overclock.net/t/1624603/rog-crosshair-vi-overclocking-thread/2520#post_25932285

 

[Encrypted11]

Also when you use and try to replicate the "known good" settings on (not just OCN), there will always be drive bys on the thread with very cropped screenshots of hard to believe benchmarks scores with no proof of real world stability. When you encounter such bragging posts (common on reddit for upvotes as well), self filter out and ignore accordingly.

Also pay attention to the max safe limits.
There's an amd robert tutorial that's good for starters.

https://youtu.be/vZgpHTaQ10k

Btw I'm letting go of a 1tb 960 pro if keen can pm

 

[blackestvoid]

guys thanks for the advice
ran memtest64 overnight, stable
upon cold boot it failed to load
definitely cold boot test is the 1st

 

[Encrypted11]

guys thanks for the advice
ran memtest64 overnight, stable
upon cold boot it failed to load
definitely cold boot test is the 1st

Additional reference to kiv:
https://rog.asus.com/forum/showthread.php?86946-Is-this-a-good-alternative-for-memory-testing

Less of the scold people part though.

 

[hlots123]

kiv, gud luk!
taichi thread
ryzen thread

p/s Encrypted11 is the expert, i is jus kpo & fumble around...

 

[blackestvoid]

adding usb test to the lot
had one usb thumbdrive showing up but contents unreadable
verified thumbdrive is working fine on another computer
but this only affects this one particular thumbdrive, plug in usb hdd works