this post was submitted on 04 Oct 2024
14 points (93.8% liked)

Buildapc

3730 readers
15 users here now

founded 1 year ago
MODERATORS
 

I'm saving up to get a 5700X3D around Christmas, upgrading from a 5600G, but I want to make sure I prepare properly before I do the swap.

The RAM I bought couldn't match the C18 @ 4000 M/Ts advertised and still remain stable, but I managed to manually overclock to C16 @ 3666. Should I drop to JDEC specs before I upgrade, or is it a non-issue?

you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 2 points 3 days ago* (last edited 3 days ago) (1 children)

No problem bubs.

I think a reasonable assumption to make is that silicon quality improves as a process matures; a later batch Vermeer X CPU (like the 58/5700 X3D) may perform in some ways above and beyond it's original spec.

With that said, I feel it's unlikely to reach your kit's profile (anything north of 1900 running stable on the fclk for an am4 chiplet CPU is practically unheard of) but I'd be interested in how the testing goes, since I've not tried to tune on a new AM4 system in some time now.

If you're able to identify the exact memory ICs in your kit, you may have some better expectations; f what primary and secondary timings would work best. The only way I've been able to do this in the past is by looking up the memory kit part model or using janky, windows specific software called thaiphoon burner (which isn't always accurate in any case).

[–] [email protected] 2 points 3 days ago (1 children)

I'm pretty sure they're Hynix C-bin. About middle of the road, as far as OC'ing goes. Unfortunately, not Samsung B's, so I can't really push these too far (and there's a manufacturing flaw in my mobo that causes the CPU temp sensor to freeze if mem voltage is too high).

[–] [email protected] 2 points 3 days ago* (last edited 1 day ago)

That's fine bubs, we can try for something like 3600 CL18, maybe hit 3800 at the same CAS latency?

The neat thing about the X3D parts is that higher frequency, low latency memory is less critical to application performance. I think you'll be fine perf wise if you scale back a bit.