Have to agree with Brent's observations - that seems like a "ton" of camshaft for the power level and RPM. With darn near zero "real" information, I suspect that dropping cam down will make for a happier combination. Something in that package is not playing well yet (hard to believe we are saying that about a mid-800 horsepower FE). Might be something harmonically uncomfortable in the valvetrain relationship between cam and spring package - I have had a couple where having a bunch of spring did not really help and actually cost power. Not really bounce so much as just unseating.
Might be a cross section deal? Seems like a fairly odd result for sure - I would expect that high an RPM torque peak to give power at 7000 RPM or so.
In any case its really cool to see solid progress!
I agree with Barry. It appears that the engine may be experiencing a sonic-choke condition or the inability to feed the engine due to an overspeed of air/fuel somewhere in a critical area(or areas) of the port. A Pitot-probe measurement of airspeeds through the entire port ( port floor, roof ,corners, walls and center) would offer clues. If a Pitot tube is not available , a precise CSA mapping of the entire port will reveal localized airspeeds with some math.
Excessive intake reversion due to improper camshaft valve events may cause this as well. Additionally, imprecise exhaust event timing will have an affect on the blow-down efficiency of the port which will negatively impact the beneficial overlap cycle by increasing exhaust pumping losses. This leads to intake pumping losses ...................etc.
The question is this ; Does the induction system and valve lift curve satisfy the demands of the piston ? Is the airflow and air speed managed throughout the lift curve with respect to piston speed and position?
Just some thoughts.