And that was my impression of the p-System too. The key thing that made Java fast, and wasn't clearly appreciated (except by a few people, not me) at the time was that the JVM was small enough to execute entirely in the CPU's cache. When the p-System came out, caches were tiny, a handful of bytes at most. By the time Java was defined, caches were growing larger, and the language designers knew it.

When the p-System came out cpu caches were not yet a thing. Main memory was fast enough to match cpu speeds. Your cache was your register set.

CPU caches were not a thing on microcomputers. The 360 Model 85 had a CPU cache in 1968.

UCSD Pascal allowed you to write compiled binaries on the Apple II -- that's very impressive technically when everyone else was either coding in BASIC or having to code in assembler. My point being that this was, in its way, far more impressive than MS BASIC, but did not create a software behemoth.

The binaries were p-code. I agree, using Pascal at that point (very small RAM, the need to extend the Pascal to be more than a tool for learning) was surely a worthy technological achievement, and, if I remember correctly, the system allowed the user programmers to produce the programs which were bigger than the available RAM. That had otherwise to be custom implemented in the non-trivial programs at the time. To compare, original MS Word (for DOS) was also at least partially p-code and was bigger than the RAM. Before it, Wordstar also had to have the mechanisms of having only one part of the program in the RAM at the time. However I still find the later Turbo Pascal significantly more amazing product.