It seems like you're questioning whether the issue lies with the motherboard, processor, or RAM.
It seems like you're questioning whether the issue lies with the motherboard, processor, or RAM.
Parts: 13900k, bought right at release. This means it could be damaged from excessive voltage or oxidation. RAM : Crucial 64GB Ballistix DDR4 3600 MHz Mobo: asus tuf gaming x570 wifi Cooler is 280mm aio GPU : MSI Suprim 4090 There are 2 SATA and 4 NVME drives I replaced the PSU with a platinum Seasonic Focus 850W 2 days ago. Machine wouldn't post. Ok, RAM. Tried CMOS reset (which would have released XMP profile). Nope. Eventually I pulled one of the sticks. Posts, works. I tried putting a stick back. No post. I switched both sticks to the other socket. No Post. I put the sticks in separate sockets on different banks - posts, boots, all memory shows up. (this is a 64 gig kit) Putting the sticks into the XMP profile now causes a blue screen immediately on boot. It was working fine. Something has failed but I've got 4 culprits : 1. The memory itself. But it fails immediately in dual channel, and is working fine otherwise. My hunch is the memory is fine, but I can't put it into a speed needed to test in memtest86 (testing at this slow speed in single channel won't tell me anything) 2. The socket. Apparently Intel goofed and 13900k sockets can warp over time. I used a light AIO though so dunno. 3. The motherboard. 4. The CPU itself. Given it may be burned out 2 different ways, this is the biggest suspect. I don't know what to do now other than try to RMA the CPU with Intel, and see if a new CPU works. If it doesn't, then I would need to get another RAM kit (kinda expensive) or another motherboard to see what the problem is. I'm tempted to just give up and replace all 3 with a new AMD motherboard + CPU + DDR5 RAM.
Consider removing the CPU and thoroughly cleaning its contacts using IPA alcohol. Perform a Boot MemTest86 to check for any CPU issues. You might also clean the entire motherboard with an oil-free contact cleaner and IPA. Use a rubber eraser to clean the RAM contacts, then run MemTest again to see if it resolves the problem.
The CPU bending issue could be resolved simply. Purchase an inexpensive contact frame from Amazon or Aliexpress. It seems reasonable. Just test one of the functional RAM settings until you're more confident. Perform a Memtest86 in that setup. You likely won't feel any extra cost. Have your BIOS updated? Edited July 28, 2024 by leclod
They managed to achieve this, though I haven’t seen any CPUs fail due to warping. It affects thermal performance more than causing permanent damage. This isn’t something I’d worry about right now. The problem lies with the memory system, so the existing Intel instability concerns aren’t connected. With all 64GB available at once, it’s unlikely the issue is in the RAM and probably points to the CPU or motherboard instead. Also, just to be sure, which memory slots are you using for each attempt, and where are they currently located during boot?
The RMA board should likely resolve your issue, which seems confusing and strange. You’ve already ruled out the CPU as a problem, so removing the board should work. It’s possible the issue was due to socket or CPU problems, making random failures rare. If you have a degraded chip, replacing it now is probably best while you can.
Concerning bending: it matches what I read, but it stopped working after I changed the power supply, which required some adjustments to the cables and aio. For memtest86: I noticed I can swap kits with my girlfriend's system and run RAM tests at full speed. I also received the latest BIOS update.
Sockets: any combination that doesn't support dual channel is acceptable; a single slot will do. The issue lies with the dual channel functionality, which stopped working abruptly. It seems like RAM training triggered by a new CPU setup caused the problem. I'm uncertain if it's the board or the CPU—since the CPU has its own memory controller and appears faulty—I suspect the CPU is the culprit. I plan to test the memory, escalate RMA for the CPU, and if issues persist, isolate the board.
If you can't repair the board, it still makes sense to replace it together with the CPU just to be certain