• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Stability Testing

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

pqwoerituytruei

Member
Joined
Nov 24, 2011
Should i be using memetest86+ to verify that ram is stable or should i run multiple instances of memtest in windows
After manually applying my XMP profile's timings i was able to get the system to crash in about 10-15 minutes of running, not spew errors just hard crash and start booting, no bsod, it just resets the system
i was having some issues

EDIT: attached Typhoon burner reports for the ram i have tested
 

Attachments

  • tet.png
    tet.png
    48.9 KB · Views: 567
  • Typhoon Bruner HTML Reports.zip
    5.1 KB · Views: 22
Last edited:
Memtest86 doesn't really stability test an overclock. It will test for corrupted/bad sticks but it is really only beneficial to use for stock settings, since it does stress the ram, only evaluates accuracy.

Most people here will use prime95 set to blend, and then select custom and specify to use between 80%-100% of system memory installed. I don't have any experience with the "mem bench" you show, I don't believe it was included in the version of the calculator I used.

Can you update your signature?

Why are you manually applying XMP profile timings? A 3600 should have no problem running 3200, but including the type of memory you're using as well as the ICs (since you need that to run the calculator) would be helpful. There have been some sporadic issues with older chipsets, but this could be chalked up to BIOS issues.
 
I have not moved my new hardware into my daily system at this time, hence the outdated signature
The ryzen calculator runs memory testing is multiple instances of this software https://hcidesign.com/memtest/ v6.0

I was having issues posting even with everything auto without even using xmp, took it all apart and used contact cleaner to see if that fixed it and was checking to see if XMP was stable, make sure something is not defective
error was saying "Base 64K memory failure"

My test setup is a X470 Gaming Plus, Ryzen 3600, G.Skill F4-3200C16-8GVKB, GTX 650 Ti Boost, Seasonic 520W S12II, Hyper 212 LED, Fractal R2 140mm fan, and a 120GB SSD CT120BX500SSD1

My board, looks to be ignoring 1/2 the settings in the ram EEPROM for XMP
 

Attachments

  • msi bug ?.png
    msi bug ?.png
    90.2 KB · Views: 578
Last edited:
Sorry to say it seems that there are a lot of growing pains in the x470 bios support for Zen2 (3000 series chips). If you're having trouble posting at stock/auto JDEC timings then you're not going to be stable at XMP.

Are you using the newest BIOS for the board?
 
latest bios, been checking for updates daily
after using contact cleaner on CPU/GPU/RAM (slots included) JDEC settings seem to work consistently
i think one of my sticks is defective, also think there could be a nvidia driver bug, possible since nobody is running a GTX 600 series on ryzen 3ed gen
also suspect there is a BIOS bug involved as randomly the system will not make any attempt to post
 
If you think you have a bad memory stick, take one stick out at a time and test each one. Personally, I think your problems likely revolve around running a Zen 2 CPU in a x470 board. At this point in time, lots of others are having similar problems. And MSI 470 boards seem to particularly problematic in this regard from what I can glean from reviews. Running a gen 3 Ryzen in a 470 board is pretty hit and miss right now. For stress testing, I like using the Realbench stress test.
 
I have isolated the stick that is giving me problems
I believe i have 3 problems
1 bad stick of ram, i have manged to get the bios to beep about it being bad in all 4 slots, but i have not managed to get the other stick to do this
1 BIOS Bug
1 NvIdia driver Bus
i guess newegg's RMA testing is not perfect, but in my testing is some configs it will work for hours, while in others it acts completely dead
I got the ram open box to save a buck, yea iim a cheapskate
 
The good thing is that all or most all RAM manufacturers offer a lifetime warranty on their products. RMA time.
 
memtest86+ would not give a false positive would it?
 

Attachments

  • 20190825_000259.jpg
    20190825_000259.jpg
    2 MB · Views: 497
Its always possible but unlikely. If one tested bad in all four dimms youve found the offending module. Contact gskill and they will take care of you.
 
I already did, i got my RMA approval Friday afternoon, no need to run i can take it to the PO tonight
with my ddr3 experience, when ram is bad it is consistency bad not 10% i'm gonna crash your system this session and the other 90% i'm gonna work till you run out of time to test me
it is as if ECC would actually be useful
 
Test Stick A in slots 2 and 4 for 2 passes, all good
Test Stick B in slots 2 and 4 for 2 passes, all good
Test both sticks in slots 2 and 4 for and get error on 1st pass (sticks were swapped compared to the pic i took)

Memtest86+ XMP results
Code:
 Memtest86+ | Stick Serial #   | Test 1 Slot      | Test 2 Slot | Test 3 Slot | Test 3 Slot | Test 4 Slot | Test 5 Slot      |
   5.01     | Stick A          |      2           |      2      |      4      |             |             |      4           |
 multi-core | Stick B          |      4           |             |             |      2      |      4      |      2           |
  Results   |                  | Err tst 7 pass 2 |   --------------No error, ran 2 passes--------------  | Err tst 7 pass 1 |
It took about all day to run all this testing, i was only able to get errors in memtest while using 2 sticks
 

Attachments

  • 20190825_162053.jpg
    20190825_162053.jpg
    1.7 MB · Views: 477
Last edited:
I have isolated the stick that is giving me problems
I believe i have 3 problems
1 bad stick of ram, i have manged to get the bios to beep about it being bad in all 4 slots, but i have not managed to get the other stick to do this
1 BIOS Bug
1 NvIdia driver Bus
i guess newegg's RMA testing is not perfect, but in my testing is some configs it will work for hours, while in others it acts completely dead
I got the ram open box to save a buck, yea iim a cheapskate

What do you mean by "configs"? Software, even stress testing software, typically will use all the memory or the same areas of memory. The ones that need to access the bad part of the bad memory stick will trigger failure. Memtest86 and other dedicated memory testing tools do a much more thorough job of checking in areas that may not be used by some other apps.
 
configs being number of sticks used and what slots they are as well as xmp be enabled or not
even if i oc it it can pass hours of testing then randomly crash at idle
 
Test Stick A in slots 2 and 4 for 2 passes, all good
Test Stick B in slots 2 and 4 for 2 passes, all good
Test both sticks in slots 2 and 4 for and get error on 1st pass (sticks were swapped compared to the pic i took)

Memtest86+ XMP results
Code:
 Memtest86+ | Stick Serial #   | Test 1 Slot      | Test 2 Slot | Test 3 Slot | Test 3 Slot | Test 4 Slot | Test 5 Slot      |
   5.01     | Stick A          |      2           |      2      |      4      |             |             |      4           |
 multi-core | Stick B          |      4           |             |             |      2      |      4      |      2           |
  Results   |                  | Err tst 7 pass 2 |   --------------No error, ran 2 passes--------------  | Err tst 7 pass 1 |
It took about all day to run all this testing, i was only able to get errors in memtest while using 2 sticks

Are you sure you are putting the sticks in the correct slots? Most newer motherboards would use the two slots furthest away from the socket. Sounds like you are alternating slots. Did you check the motherboard documentation to see what is the correct slot configuration when running two sticks of RAM?
 
Last edited:
In my initial testing with the non-beta bios i was able to get 3 beeps from just one (and just that stick) of my stick when in either slot 2 or 4
according to the only documentation i can find 3 beeps means Base 64K memory failure - bad memory
This was stick A
guess i need to run more testing in the morning... I have not shipped it out yet...

On another note, my mom's system had issues with not waking from sleep mode (only sometimes), like it would get stuck, for some reason when i added a second stick of ram the issue vanished (budget Asus AM3 micro atx board) ever seen a issue like that?
 
I managed to get some ram from a different system, nothing fancy ram from a dell
i was able to get the same exact error
odd it always occurs on cpu 2
 

Attachments

  • 20190828_000959.jpg
    20190828_000959.jpg
    1.5 MB · Views: 424
In the interest of science i disabled SMT, Memtest86+ crashes around 3 minutes into the test at 1/2 way through test 7
 
the core disabling feature is actually quite useful
i set it to 4 cores 2+2 and no more memetest errors 8hrs 15min and counting
additionally setting to to 4 core 4+0, infant does nothing as there are still 12 threads
 
Back