Intermittent instability with new RAM

Hello everyone, specs are below.

 

So I'm currently experiencing a strange issue with my computer. A few days ago I got a 32gb memory kit by Corsair (CMW32GX4M2C3200C16) because the price was good. I previously had 16gb of corsair ram which I took out of my PC and installed the new 32gb. I experienced a BSOD about 4 hours later. I rebooted the PC and went into Prime95 and ran some Blended tests and it started spewing out errors. I rebooted which fixed the issue and everything went back to normal for the rest of the day.

 

Then, I started experiencing crashes with my software/games. I decided to run Prime95 Blended tests again and it was still experiencing more "fatal errors". I turned the PC off and on the first 2 tries, the PC would stall when trying to get into my diagnostics USB drive. I managed to get into memtest and ran it for a few hours without issue.

 

The intermittent nature of this problem makes it very confusing to me: as it is, every time I boot up the PC, the first thing I do is I run Prime95 blended tests. If it doesn't make errors in the first few seconds, I'm good until the end of the day. If it does make errors, I reboot until it doesn't. It can take 2-3 tries.

 

 

Any tips and ideas would be very appreciated.

 

 

CPU Intel i9-9900k

Motherboard Gigabyte Z390 AORUS PRO Wifi (F11 BIOS, most recent)

RAM 32gb Corsair Vengeance 3200mhz

GPU EVGA GTX 1080ti FTW3

Storage Samsung 960 EVO 500gb

PSU Corsair RM750x

Operating System Win 10 x64

Can you clarify what settings your are running in relation to the memory (frequency, timings, voltage)?

 

If you crash from a memory related error, your bios may alter setting to ensure your next boot. That does not mean they are optimal to run. The entire point is to make to post so you ca change things. Be wary any time you crash from a memory related error.

I'm running at XMP specs, so in the case of this SKU it's:

DDR4 3200 (PC4 25600)

Timing 16-18-18-36

 

As for the BSOD, that only happened once and I'm not sure if anything was changed.

I'd recommend downloading a copy of MemTest86 and see if you have faulty memory, BSODs are generally not a good sign.

 

You could always try inputting the memory timings manually and/or downclocking the memory and rerunning MemTest86 and see if the errors clear.

Well as I have said I have ran Memtest and found no issue, but the instability is intermittent which confuses me greatly. I'll try to run it overnight again and see if something happens. Then I'll downclock the memory to see if it solves it.

 

The problem is that whenever I try anything, I have to monitor the situation for hours or days to see if the situation changes at all, because most of the time there's no problem.

Your board and CPU should have no difficulty with 3200 and 16-18-36-T2 is not overly aggressive. That part should be easy, so we need to be careful about classifying something as a memory issue versus CPU setting issue. What are you running on the CPU clocks? Did you run Prime 95 as a select test? How were you able to determine it was a memory related crash? I am assuming it because everything was fine, but then immediately went sour after the kit was introduced.

 

I think I misread the original post and believed you were crashing on the 1-5 second intro memory training each boot, then the BIOS/System would power cycle and try again, and again, before booting. That is memory problem. Now you still have have one, but it is harder to pinpoint. Timings and DRAM voltage should be set by XMP. I don't know if GA is like Asus where "XMP tuning" is a whole separate mode in the BIOS. When you engage it, your prior settings are wiped. Make sure things are still the same.

 

Find out what the BIOS is using for VCCIO and VCCSA voltage.

Your board and CPU should have no difficulty with 3200 and 16-18-36-T2 is not overly aggressive. That part should be easy, so we need to be careful about classifying something as a memory issue versus CPU setting issue. What are you running on the CPU clocks? Did you run Prime 95 as a select test? How were you able to determine it was a memory related crash? I am assuming it because everything was fine, but then immediately went sour after the kit was introduced.

 

I think I misread the original post and believed you were crashing on the 1-5 second intro memory training each boot, then the BIOS/System would power cycle and try again, and again, before booting. That is memory problem. Now you still have have one, but it is harder to pinpoint. Timings and DRAM voltage should be set by XMP. I don't know if GA is like Asus where "XMP tuning" is a whole separate mode in the BIOS. When you engage it, your prior settings are wiped. Make sure things are still the same.

 

Find out what the BIOS is using for VCCIO and VCCSA voltage.

Hello c-attack, thank you for your response.

 

To answer your questions one by one: although I used to have the CPU OC'd to 5ghz on all cores, I'm currently running it at stock speeds.

-I did run Prime95 extensively prior to getting the RAM without any issue

-I can't definitely say that the ram is the issue, however, running Prime95 with the new ram, Blended tests (which test a lot of memory) would instantly return errors. Meanwhile, doing Small FFTs which mostly stressed the CPU wouldn't cause any problem. As I've said, I was also able to reboot the PC 1-3 times and then Blended tests would run without issue.

-Correct, the issues started after I got the new ram. What confuses me is the fact that it's so intermittent. In my experience if ram is unstable it'll be unstable every time you run the same synthetic test until you lower the clocks or increase the voltage

-I did validate that the XMP settings are the ones my DIMMs are rated for.

 

In any case I don't have much to go on so I'm just monitoring the situation.

I think in your position I would likely return the kit, either for the same thing again (if that's what you really wanted) or check out a different spec model. It may be possible to narrow this down to specific timings (with a few weeks of work), but my general opinion is if it throws up errors at 3200, it's going to be a pain in the neck for the duration. Better to cut ties and return now if you can, rather than find yourself in the same position 6 months down the line with fewer options and a massive headache.

I think in your position I would likely return the kit, either for the same thing again (if that's what you really wanted) or check out a different spec model. It may be possible to narrow this down to specific timings (with a few weeks of work), but my general opinion is if it throws up errors at 3200, it's going to be a pain in the neck for the duration. Better to cut ties and return now if you can, rather than find yourself in the same position 6 months down the line with fewer options and a massive headache.

After 2 days without issue, I had another crash, so now I'm pretty well convinced that the RAM is the problem. I RMA'd the kit. I could have tinkered with the timings and frequency and waited for days to see if it was still unstable but as it is my 16gb kit will have to do.

 

Thanks everybody for your help, cheers!

Well then...

 

I just had a BSOD with the stop code MEMORY_MANAGEMENT with the old ram... this is getting weird now. The PC wouldn't boot up until I re-seated one of the ram sticks, before then it was turning off and on repeatedly very quickly.