The Traveller computer issue perhaps depends on the definition of computer, meaning what belongs to this thing.
To the largest degree, size efficiency is related to processor, memory and other transistor based components. Power and infrastructure components also experience some size shrinking, but due to the vastly increased capabilties of the core components there often is more infrastructure around, too.
As a result the volume occupied by electronics e.g. in automotive products rapidly increases, where most volume is indeed occupied by infrastructure (wiring), protective components (isolation, shock proving) and interface technology.
Extrapolating this into Traveller schemes, I have less problems with the vast volumes occupied by a computing system controlling perhaps several million components on a starship.
Really, a model 3 starship computer with 12 kl volume really does not differ much to a double twin E10K cluster in a climate control and protection cubical along with all the wiring to the outside world....
Regarding the jump drive and power issues we should note, that even superconductors have practical limits (a few 1000 A/mm2). Dealing with Traveller energy amounts, connectors might still take up quite a volume
.
Some realworld info: HTSCs (high temperature superconductor cable) with a diameter of 15 cm are good for power transmission of 115 MW.
Thinking of the vast amount of jump fuel used there might be a striking energy flow (well, depends on JDs power efficiency, which isnt noted anywhere AFAIK....).