...
How do you know negative vcore offset is broken on my board? there was no vcore offset from MSI with this board until recent bios updates.
...
Sorry I left out a few words before. I meant to say "has been broken on MSI boards." You can tell me if it's fixed if you want to check load voltages with a repeatable load (i.e. Prime95). I've seen that it's been added to boards since 2018, but not that it's worked. That said I'm not an expert on MSI bios.
All of this said, you didn't answer how your determination of "no difference" was made. I will also note that in the big picture, the amount of difference (double digit changes to CinebenchR20 scores) I'm detecting is mostly academic. It's also important to say that this is one benchmark. For example the way memory speed works with the Infinity Fabric in Zen, Zen+, and for the most part Zen2 (although an infinity fabric divider is now available for the latter) has the likelihood of having a bigger impact on other things. Maybe even the single core CBR20 since it seems to utilize all the cores at different times. A load where each thread gets assigned a render and then completes that render and then gets a new render assigned to it is the ideal picture for infinity fabric already, as cores are not sharing a lot of data (at least this is my inference by watching the test). I just don't want anyone to read this and think memory speed matters less (or more) than it does.
I'm still not sure if we're fully talking about the same thing in terms of an "all core boost." Any time I open a monitoring program it also reports at each of my cores
has boosted to the maximum value of 4250MHz. When we are discussing an "all core boost," we're referring to a simultaneous all core boost, have you seen all the cores hit this clock at the same time under stress test or other load to all cores? It's nothing for them to boost to max if they are essentially at idle.
As for your manual OC, it looks nice, and it appears to be an appropriate reward for your ambient temps. I'd be freezin' my butt off! If my CPU would hit a manual OC that was higher than the "all core boost" from PBO at 1.4V or less, then I would OC manually. I agree it's more fun, its certainly easier lol. Either my memory is impaired or something else has changed, because I seem to remember pegging 4150MHz running prime95 when I first set the system up. Now I'm sitting between 4000MHz and 4025Mhz running small FFTs. In a day or two I'll come up and manually OC and run CBR20. I don't want to claim the 4150MHz because again my memory is missing out.