E3-1230 to EPYC 7313: worth the jump for homelab?

Finally got my hands on an EPYC 7313 for the homelab. Before I migrate everything, I benchmarked it against my old E3-1230 v3 that's been humming since 2014. E3-1230 v3: 4c/8t, 84W TDP, passively cooled in winter
EPYC 7313: 16c/32t, 155W TDP, Noctua NH-U14S barely keeping up Single-thread: EPYC ~35% faster
Multi-thread: EPYC ~4.2x faster
Idle power at wall: E3-1230 47W, EPYC 89W The jump is real but the efficiency curve is ugly. At 10% load the EPYC is burning almost double the watts for marginally better latency. For my use case — Plex, three VMs, ZFS scrub monthly — I'm not sure the TCO works. Status page says my E3 has 99.97% uptime over 4 years. Anyone actually measured 7313 at 65W PPT? Curious if the efficiency sweet spot exists.

Hey folks, super stoked you're doing real numbers!

I ran a 7313 in my rack for like six months and honestly? Gonna agree the idle hurt. I was pulling 95W at the PDU doing basically nothing. Moved to a 7232P for the lab and never looked back. Less cores but the perf-per-watt is

Your E3 is a legend though. Mine died in 2019 and I still miss it.

IMO the EPYC only makes sense if you're actually saturating those cores regularly. YMMV, but for typical homelab stuff the E3 generation is weirdly hard to beat on efficiency.

@uma I underclocked a 7313P last year — 65W PPT, 2.6GHz all-core — and got surprisingly close to stock 7313 performance in my workloads. Take it with a grain of salt, but the efficiency gains were real. Might be worth testing before you commit to the full power draw.