RAM more stable with time?

techNOguy123 · Jul 11, 2018

I recently built a PC with the following specs:

Ryzen 7 1700
Asrock X370 Pro4 (BIOS v4.70)
G.Skill Ripjaws V 16GB DDR4 3200 16-18-18-38 (non Samsung B-die)
MSI GTX 1080 Duke
EVGA G3 650w
Thermaltake G21 case

Originally, I had a CM Hyper 212 LED Turbo installed. I got a BSOD when using XMP 3200MHz/1.35v @stock CPU clock/voltage (3.0GHz/1.18v). When the CPU was overclocked, it was 2hr stable using Prime95 @3.9GHz/1.344v/stock RAM(2133MHz/1.21v)/28C-55C. However, I could not get it stable beyond 10 minutes using XMP @3200MHz up to DRAM 1.375v.

I then switched to a CM Masterliquid 240 Lite AIO (26C-48C). I also decided to try out DRAM 1.4v for XMP 3200 which was stable for 30 minutes. So I backed it all the way down to 1.35v and now it is stable up to 2hrs. Is this normal? What could have caused this? Latest BIOS has been installed since the beginning.

zyh1987 · Jul 11, 2018

one thing that i could think of is the cooling on cpu also cools your memory controller.

which helped memory overclock.

TJ Hooker · Jul 11, 2018

Stability is a probability/luck thing. Maybe you just got unluckly the first time and lucky the 2nd time.

techNOguy123 · Jul 11, 2018

zyh1987 :

I thought that but it's only a 6C difference and I think some people OC to much hotter temps as well.

TJ Hooker :

Initially with the Hyper 212, I tried intermediate speeds as well, 2666 MHz at 1.35v and that was also unstable. After installing the AIO, I tried 3200 MHz at 1.4v, 1.385v, 1.375v, and 1.35v. All were stable.

TJ Hooker · Jul 12, 2018

When you say "stable for X minutes", you mean you ran the stress test for X minutes before it crashed right?

Also, what P95 test are you running? Overall P95 is more of a CPU stability test than a memory one from what I remember, I think memtest86 is one of the best for testing RAM stability.

techNOguy123 · Jul 12, 2018

I meant I only tested it up to x mins/hrs during which no crash occurred unless otherwise stated. I only tested up to 2 hrs max because I noticed crashes usually occur within the first 5-10 mins and I rather not subject unrealistic stress beyond a duration I would normally perform a workload (rendering/encoding).

In prime95, I tested both small FFTs and blended. I've never tried memtest86 but have used memtest64 once (I think it was 3200/1.375v) during which it passed a 2 hr run.

Search

RAM more stable with time?

techNOguy123

Distinguished

zyh1987

Estimable

TJ Hooker

Titan

techNOguy123

Distinguished

TJ Hooker

Titan

techNOguy123

Distinguished

TRENDING THREADS

Latest posts

Moderators online

Share this page