Here’s the YouTuber:
Intel & AMD have both 'lied' about their real world power consumption. Here's the Truth:Sponsored Message:Whokeys 25% discount code : TN20 Windows 10 Pro O...
youtu.be
He’s a creator. Please let me know your thoughts.
Yikes, looks like that needs a lot of digesting. I haven't watched it all yet, but a few thoughts (I think lsg1r covers most of these):
- BIOS settings are super important to efficiency for both CPUs, especially if they're being run power limited (or not). X and K CPUs are generally not ideal from this perspective, for either AMD & Intel. AMD usually optimises their non-X CPUs much better for power efficiency out of the box (a trend that goes right back to Ryzen's first gen and continued with e.g. 2700 non-X, 3900 non-X, 7900 non-X), but the settings motherboards use vary quite a bit. Gamers Nexus in their testing, will usually use the manufacturer's recommended settings and turn off any automatic overclocking (like exceeding the stock power limits).
- With Zen 3, Intel was often more efficient at light/low loads when a Ryzen system isn't optimised, primarily because of their excessive boost behaviour, but I don't know if that's still the case and it could be changed in the BIOS.
- I think it is fairly well known and accepted that the lower-end AMD CPUs (like 7600/7600X) are less competitive than Intel CPUs (like i5-13500) in productivity work.
- Anandtech and Gamers Nexus (the articles & videos I linked) show different results to him, though Gamers Nexus were using a non-X CPU and Anandtech were deliberately tuning the CPUs for efficiency.
- In GN's eco mode
testing, the stock 7900X beat the 13900K by quite a long way and a power tuned 7950X smashed it out of the park. So, I'm surprised his results are so different.
- Adobe apps are (as far as I know) regarded as being primarily single-core / single-thread dependent and I suspect that like Zen 3, Zen 4's out of the box boost behaviour is what is letting it down in these results.
- GN and Anandtech have used heavier multi-thread apps in their testing, like Blender, which is more likely to turn out in AMD's favour.
Regardless of which one you buy, I'd recommend setting the power limits and boost behaviour to best suit your usage profile (K and X CPUs are optimised to win benchmarks, not constrain your power bill) and I'd choose a more power optimised model because they're partly configured that way out of the box (e.g. 13700 non-K or 7900 non-X).