The DSP budget

You try to add a second reverb. The unit refuses. The CPU meter is showing about a third. So you delete a delay, try again, still refused, delete something else, and eventually conclude that the meter is broken or the unit is faulty.

It is not broken. You have hit a different limit, and the meter was never going to show it.

Neural state this plainly: bypassing or disabling blocks does not reduce CPU consumption. A bypassed delay costs exactly what an active one costs.

So the reflex of switching things off to make room does nothing at all. To reclaim processing you have to delete. Bypass is a performance control, not a budget control — and once you know that, a lot of previously baffling behaviour stops being baffling.

The constraint that usually bites first is enforced against row pairs rather than against the preset. The reported rule is one reverb on the row 1/2 pair and one on the row 3/4 pair, with spring reverbs excepted from the count.

That single fact turns an impossible preset into a five-second fix: move the second reverb to the other path rather than deleting it. It is also the only explanation that fits a refusal at a third of the meter.

When a preset exceeds the available processing resources, the Global EQ and Input Gate blocks are automatically disabled. There is no error and no fuss.

So if your Global EQ works on most presets and mysteriously does nothing on one, that preset is over budget. It is not a bug and it is not your EQ — delete a block from that preset and the EQ comes back.