Failing RAM? In this economy?
Now to go beg to the RMA gods
You must log in or # to comment.
f
F
They’re very few addresses, you can tell the OS not to use them. Take a look at this: https://www.memtest86.com/blacklist-ram-badram-badmemorylist.html
Dawg you’re the GOAT with that link
It didn’t even occur to me it could be done, makes sense.
Seems like its always the same core that is creating the error. Could be a bad overclock. Try reseting the bios to default. Also try reseating the RAM (and clean out the slots with compressed air) and rerunning the test. I’ve had similar issues in the past that were fixed by this.
Once stable overclocks can fail after many years as it slowly wears out some transistors.
Not good. The pattern is not consistent.
Take the RAM out and carefully clean the contacts with e.g. alcohol and a piece of lint-free cloth.
F
had a 2x8 kit fail as well, thankfully I clocked it down from 3600 to 3200 and it passed the ~2 hour test.
perhaps relax the timings and/or increase voltage?
Timings are already base unfortunately as you can see 2134mhz
Gonna RMA them
I’ve got two 16GB sticks that I bought in 2020 pullin me through—I’m praying they last long enough til prices go back down haha
Try blowing on it and inserting again
This usually makes me start working.
F. Godspeed.
Dang, i’m a bit worried for my own RAM as well. I keep getting random reboots on my pc every now and then. I initially suspected my distro to be the problem since it started happening after i hopped, but after feeling like i checked everything on the software side, i’m checking hardware now just to be safe anyway. Started with removing the OC from my RAM. I also have my 5800X3D undervolted though since i saw it recommended at the time. Now that it’s a few years old i might just not be stable anymore at that voltage.
Time to go back to weaving memory
The software for the AGC was written by programmers at the Massachusetts Institute of Technology (MIT) Instrumentation Lab, and was woven into core rope memory by female workers in factories.[2] Some programmers nicknamed the finished product LOL memory, for Little Old Lady memory.[3]
LOL Memory indeed
They want RAM, not ROM ;)
i had to pronounce some ddr5 dead this morning. at least it was ‘only’ one 8gb stick, but it’s in an omen prebuilt, out of warranty.
user’s just a high school kid. gonna have to limp along with the remaining 8gb (not much for gaming then) or random crashes with 16gb until he or the parents can afford to replace it.
Linux? I’ve heard if you can identify the bad blocks (assuming it’s just one group of bad blocks) you can tell the kernel to not use those. If you’re technically inclined.
he has windows, but i’ll mention the option to him next chance i get.
There is a option for Windows: https://www.memtest86.com/blacklist-ram-badram-badmemorylist.html#badmemorylist
Although it seems like it may be broken.
I ran this too on mine, I had the mode set to memmap. Badram is old, grub these days expects memmap.
There were a few areas but they were close enough, so this is what I ended with.
Blocks of 64K were enough to fix this.
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash memmap=64K\\\$0x3c8f0000,64K\\\$0x49068000,96K\\\$0xa24660000"
What motherboard is it? I had problems with a 4-stick ddr4 configuration recently and found out the mobo didn’t have the same specs on all 4 slots 😅
MSI B450 Tomahawk
Dude you are running DDR 4 Ram at 2134MHz and with 15 timings. The Timings are good but the frequenzy is shit. You want atleast 3000 MHz with modern Ryzen CPUs for their Virtual Memory Fabric. You can probably get equally good Ram for a few bucks because these sticks suck.
Unless you already lowerd the Frequenzy in that case RIP.
Yeah I lowered it to make sure that wasn’t the problem
I’m not sure what you’re talking about. DDR4 without XMP typically runs at 2133 or 2400mhz. It’s probably best to run a mem test at the stock frequency.
You can edit the grub config (assuming you’re using Linux) and add reserved ranges for the affected physical address.
find the linux boot command and add a memmap=<size>$<start> argument This tells the kernel to avoid the bad parts.
In your case it looks like a relatively small chunk of bad memory so… memmap=64K$0x130FE0000
I’ve used this trick to stabilise systems with faulty RAM and it works.
