~4073Mhz actual, ~1.285V SVI3 (core volts after vdroop)

Ooh much faster than your previous run despite the same reported frequency. I'm guessing it's because the sustained clock is more stable with the new tool ?

yep, exactly that! With clocks set using that tool effi is so much better, more stable clocks, more consistent power consumption suggesting higher effective util, it's definitely doing interesting things with using PPT to poke at the SMU. RDNA4 boost is such a PBO-like mess and with some PPT access we can finally bypass or cheese a lot of that, big ups to the creator of that tool