Question PC crashing when approaching/exceeding 50% ram usage

[kicz]

Ugh, this has been going on for way too long and I can't figure it out.
Setup:

• AMD RYZEN 7 2700 (Scythe Mugen cooler)
• HyperX 32GB KIT DDR4 2666MHz CL16 FURY Black (2x16gb)
• MSI B450 TOMAHAWK MAX
• Seasonic M12II-520 80Plus Bronze 520W Retail

I've ran memtest overnight a few times, no errors. Using Windows 11, every time RAM usage approaches 50% Chrome (ugh) starts giving me the aw snap message, programs freeze and slow down, eventually leading to a full black-screen crash.
Is it possible the RAM is faulty even if it passes memtest?
Any ideas?

 

[In2Photos]

Have you tried using different RAM slots? Maybe the RAM is fine, but the motherboard has issues? Is your BIOS up to date? Any overclock on the RAM?

 

[Tech Junky]

Google Chrome’s Memory Saver and Energy Saver Modes Are Now Rolling Out

Google Chrome's new Memory Saver and Energy Saver modes should now be enabled by default on Chrome 110. Google claims that the new features can make Chrome use up to 40% and 10GB less memory.

www.thurrott.com

.

I enabled these when they were still hidden in the flags / experimental section and they make a big difference.

When you mention 50% as the key factor when things start crashing though it sounds like one of your sticks might be bad or having issues. A while back I upgraded to 32GB because Chrome is a hog and I got sick of things lagging and crashing or needing to close tabs to keep it from happening. One thing that keeps me from going AMD is all of the posts about various issues with RAM compatibility and the price compared to Intel is substantial. I do like the idea of AMD for a couple of niche features that Intel won't do unless you go up a tier from consumer to Xeon.

I would take 1 stick out boot, if it's fine then swap the sticks into the same slot and boot again and if it's fine then try pairing them up in different slots to see if things clear up.

If it's still giving you a hassle it's time to think about swapping it for different RAM or go through the system with a fine tooth comb and see if there's something loose or not seated completely. Sometimes when we troubleshoot things we miss something in the process of putting it back together. I had this happen with a cooler in the past where I thought I had clamped it back down and didn't and resulted in high CPU temps until I found the screws weren't tightened down again.

 

[igor_kavinski]

You can turn off GPU acceleration to see if that helps.

How to Turn Hardware Acceleration On and Off in Chrome

Google Chrome uses hardware acceleration to pass tasks off to hardware better equipped to handle graphics-intensive processes.

www.howtogeek.com

Try using Firefox for a while to see if the Firefox also throws similar errors.

If your BIOS has the XMP option enabled, disable it and try to run your RAM manually at 2666 MT/s.

 

[Shmee]

Ugh, this has been going on for way too long and I can't figure it out.
Setup:

• AMD RYZEN 7 2700 (Scythe Mugen cooler)
• HyperX 32GB KIT DDR4 2666MHz CL16 FURY Black (2x16gb)
• MSI B450 TOMAHAWK MAX
• Seasonic M12II-520 80Plus Bronze 520W Retail

I've ran memtest overnight a few times, no errors. Using Windows 11, every time RAM usage approaches 50% Chrome (ugh) starts giving me the aw snap message, programs freeze and slow down, eventually leading to a full black-screen crash.
Is it possible the RAM is faulty even if it passes memtest?
Any ideas?

Which memtest did you run? Was it UEFI version memtest86? Even if so, it could be possible that RAM is faulty, without errors being found.

Though the culprit could also be the memory controller on the Ryzen 2700. Zen and Zen+ didn't have the most sturdy memory controllers, especially when it came to higher memory capacity, DDR4 > 3200, or more than 2 DIMMs.

Though DDR4 2666 shouldn't be an issue, you could try running memory at default settings without XMP, updating the board's UEFI if you have not already, and also tweaking settings in UEFI setup manually.

If it does turn out to be an issue with the CPU itself, now could be a great time to upgrade to a Zen 3 based 5700X or similar CPU upgrade. At least in the US, a 5700X can be had for around $200 or less.

 

[kicz]

Have you tried using different RAM slots? Maybe the RAM is fine, but the motherboard has issues? Is your BIOS up to date? Any overclock on the RAM?

Sadly I've tried this already, same result. No overclock on the RAM, BIOS is fully updated.

You can turn off GPU acceleration to see if that helps.

How to Turn Hardware Acceleration On and Off in Chrome

Google Chrome uses hardware acceleration to pass tasks off to hardware better equipped to handle graphics-intensive processes.

www.howtogeek.com

Try using Firefox for a while to see if the Firefox also throws similar errors.

If your BIOS has the XMP option enabled, disable it and try to run your RAM manually at 2666 MT/s.

Thanks I will give both of these a try.

Which memtest did you run? Was it UEFI version memtest86? Even if so, it could be possible that RAM is faulty, without errors being found.

Though the culprit could also be the memory controller on the Ryzen 2700. Zen and Zen+ didn't have the most sturdy memory controllers, especially when it came to higher memory capacity, DDR4 > 3200, or more than 2 DIMMs.

Though DDR4 2666 shouldn't be an issue, you could try running memory at default settings without XMP, updating the board's UEFI if you have not already, and also tweaking settings in UEFI setup manually.

If it does turn out to be an issue with the CPU itself, now could be a great time to upgrade to a Zen 3 based 5700X or similar CPU upgrade. At least in the US, a 5700X can be had for around $200 or less.

Thanks, I'll keep trying for now it's just kind of working until I approach the limit and then quickly closing everything when it starts to lag.

 

[mikeymikec]

Test each memory module on its own overnight with memtest86?

I'd also check control panel > security and reliability > reliability > view maintenance history I believe the option is labelled. I'd look for hardware errors, system failures, that kind of thing.

 

[solidsnake1298]

When I first got into manual RAM overclocking I found memtest86 to be unreliable at detecting errors. I, and a lot of manual RAM OCers, use "Ram Test" by Karhu Software (not free). But OCCT (free) now has a dedicated memory stress test. Try that instead of memtest86.

 

[solidsnake1298]

Oh, another thing came to mind. The memory controller on the Ryzen 1000/2000 series wasn't particularly strong. Especially with more than 16GB of memory. Increasing the voltage to just the SoC may help.