Second repair request, damaged mouse?
Second repair request, damaged mouse?
I've got an Asus x570 Tuf Plus motherboard and have been using a G.Skill Trident Z Neo (2x16gb) kit since 2020 with no issues. Once I upgraded to 64gb (4x16gb, same Trident Z Neo kit, same timings), after a few months, I had to RMA the new RAM kit. This was sometime between late 2023 to early 2024. The replacement kit was tested and worked fine until now. Will attempt to RMA this replacement kit, but now I'm wondering, is there possibly an issue with my motherboard? Is it a probable coincidence that two kits just went bad used in, iirc, the same slots, whereas the older kit is still going strong (original kit is in paired slots away from motherboard, newer kits in paired slots closer to mother board)? I'm not really familiar enough with RAM failure rates, especially with a reputable brand like G.Skill, to know if I should begin to suspect my motherboard or not. And if the motherboard is the culprit, are there still good AM4 motherboards out there? My CPU is a 5900X. If it's literally just some bad RAM, it'd be a pain in my wallet to throw the baby out with the bath water upgrading to AM5, DDR5, etc., but if I'd have to chance it with a used motherboard to get something decent with AM4, I guess I'd consider it. Not really sure what the status is on the motherboard tier lists these days. I can list out other specs if necessary, but the only other thing that I can think of that could be causing an issue, the PSU, is a nice Seasonic PX-850, and I have a hard time thinking it's causing an issue. Really just need to hear some people's experience re: the motherboard vs ram.
You can avoid using the RAM by removing the old unit before installing the new one, placing the new RAM where the old was, and after some time you can revert to four slots.
Are you asking if the latest setup is near the CPU and could overheat? It's unlikely to be a big concern. Having two bad kits in a row is uncommon, especially with high-end models that have many volts. You probably won't need to worry. By then, you'll likely switch to a newer socket like AM5 or AM6 if needed.
You might need to slightly increase VSOC and your VDDG as well. The four 16s could be challenging to manage.
Reviewed the documentation for the board. The manual suggests additional cooling when using 4 DIMMs or overclocking, possibly due to heat concerns. It also advises specific RAM placement guidelines—kit 1 should fit in A2-B2, kit 2 in A1-B1, particularly if combining kits from different sets. I'm not sure if these differences matter, but it seems worth checking. Many users have noted that AMD boards tend to struggle with four DIMMs.
I considered swapping the original unit into reverse positions to test if the older version failed first. That would help confirm whether the issue lies with the motherboard rather than the RAM. It sits near CPU (A1-B1), but my setup offers ample airflow and keeps the RAM cool. Honestly, even with a fully air-cooled build, the careful selection of Noctua parts has kept things stable. The old system used a massive 4x4GB DDR2 Corsair module with a comical clip on the air cooler, but I think it would just introduce too much turbulence now. Plus, the fan closest to the CPU touches a fan and plastic isn’t very good at conducting heat. I’m not sure if this was your plan, but it seems reassuring. I’ll probably go for an RMA and hope G.Skill can send a replacement. Moving forward, I’ll just run monthly checks to stay on top of things, just in case. That’s why I asked—great thought!
Looking at AM4 boards, swapping to something similar would cost around $300, especially new ones, which would be a huge expense right now. Thanks for the tip. I’m planning to set DOCP and keep it simple, then research before installing if I can get a replacement from G.Skill.
Would lowering the OC (based on the idea that XMP/DOCP is an OC) also help? I’ve mentioned it often but never got it right. I’ll investigate further. I’m currently using 3x140mm (two front, one top-front) and 2x140mm (one top-rear, one rear), with NF-P14s Redux-1500 PWM, plus two 120mm fans on my NH-U12S (2xNF-F12). Even with my 3090 FTW3 Ultra and 5900X, everything stays well ventilated.
My original unit is in A2-B2; I’m sorry I didn’t remember the DIMM labeling. I was on my phone and didn’t have the manual. As far as I know, the old kits stayed in A1-B1, while failing ones always went to A2-B2. The serial numbers are sequential across the three G.Skill units I’ve tried, so I’m confident they’re distinct. Manufacturing dates are clearly marked too.
I didn’t go all out with my hardware—bought the F4-3600C16D-32GTZNC twice, which is CL16 (16-19-19-39) at 1.35V, ideal for Ryzen but not overly aggressive for DDR4. Even if I got slower speeds from four DIMMs due to AMD’s memory controller, it wouldn’t cause memory failure. Could be overheating or improper factory installation of the DIMM slots, but running the standard DOCP profile on a capable board seems safe.
I haven’t explored AM4 content much since AM5 arrived. Thanks for all the advice!
Unless you manufacture the chips yourself, or perhaps you're familiar with g.skill, it's hard to say for sure. The standard voltage for DDR4 is 1.2 volts, and for speeds around 3200 (no JEDEC spec for 3600), the timing is significantly slower. The challenge would be figuring out how to achieve those speeds with a 12.5% higher voltage. In the future, memory performance might improve thanks to better manufacturing, but quality differences matter—different dies react uniquely to voltage and heat. Some discussions suggest many low-quality components were used during or after the pandemic due to shortages, but I’m not sure if that affects memory specifically.