I just 'upgraded' from a 3950X to a 5950X and am having major troubles with memory stability. No matter what motherboard out of the two I run this CPU on, no matter which memory kit out of the 3 I have... I get errors instantly in Karhu/TM5. Even at JEDEC auto settings. If I put my 3950X back in boom, instantly stable at 3733 or any speed I want below.
Is this indicative of an issue with the memory controller in my CPU? IE did I get a bad unit? I have heard there are issues with motherboard updates currently and both of the boards I have are up-to-date with Agesa 18.104.22.168 C and on my Asus TUF Gaming Plus I cannot revert bios to check. I am not however getting WHEA errors, just purely not stable full stop. When I run Karhu once it hits 7% I get errors equal to the number of threads I am running...
Any help would be appreciated, at the moment it's a £750 paper weight... really hurts.
Anyone with any ideas? Sat here twiddling my thumbs, this is the only official channel I know to go down for support before RMA or returning to the retailer.
Sorry to hear you weem to be having the same issues I've had before my 5950x ended up dying on me ... hope for you it's not the exact same one tho (spoiler alert : my RMA was started 3 weeks ago and they warned me last friday that they won't have a replacement to send to me before another 2 to 3 weeks)
What you can try for now is setting your memory to 3200Mhz and disabling PBO ... also, some motherboard builder seemingly started rolling out a new AGESA version that seems to help lots of people with same issues
"Dying on me" what do you mean by that? I did feel like mine was getting worse while it was in...
I mean mine's dead, computer not starting anymore, and "CPU Error" led light up on the motherboard ... currently under RMA waiting for AMD to be able to source a replacement.
Ah yeah I presumed you meant dead dead, I guess I wondered if you could tell it was dying or if it just went kaput randomly? Part of me was hoping it's a bios issue but it may be worth me starting the RMA process now as it is unlikely.
So on mine, while analysing the WHEA errors over time, they did seem to get more and more precise in implying the issue causing those WHEA was indeed the CPU and not the RAM (getting errors related to L1 cache etc, so CPU internal really) ...
Dead of the CPU itself however was quite abrupt : was removing a stick of ram to check for memory issues and pc never turned back on ... tried reseating the cpu and everything but nothing would get it to boot again, so no real precursive signs either :/
Nightmare! Strange thing with mine is I am not getting any WHEA errors as with everyone else, purely just general memory errors as if my timings aren't stable, however they're perfectly fine on 3000 series with the same controller. I can set the loosest timings possible or all auto and I still receive errors equal to threads IE Karhu 16,000MB allocated to 16 threads, at 7% it reports 16 errors. Regardless of configuration it is not possible to achieve stability.
I hope your RMA goes smoothly and you get a good unit next time around! It just took 4 days to get a reply from my motherboard manufacturer and they asked me for RAM model numbers to check the QVL even though I listed them in my request and specifically said they ARE on the QVL... The back and forth with these companies drives me crazy.