Lisp and VLIW (was Re: Genera)
P. T. Withington
Mon, 6 Apr 1998 12:34:51 -0400
On 4/6/98 11:39, eric dahlman wrote:
>That was my impression. What I am really interested is a sort of
>empirical assessment of the situation. For instance, I would like to
>compare the performance of a new (fictional) VLIW lisp compiler
>"Monty" and GCC version 7.3 (equally fictional) which is really good
The LispM claimed to use 10% additional hardware in support of Lisp.
Compiler technology stunk back then, so the solution was to do more in
hardware. But that was considered too expensive back then.
Now we have the situation were 100% (or 200%, or 300%, etc.) additional
hardware is available, because hardware is so cheap. But is it the case
that compiler technology is so good that it doesn't need (or can't use)
that extra hardware? Or is C so low level that there is no parallelism
to discover? Let's bring back lousy compilers for high level languages
so we can utilize all this cheap hardware!-).