Re: Intervention needed?

["Jeffrey R. Carter" <spam.jrcarter.not@spam.not.acm.org>]

On 3/22/19 12:10 PM, Lucretia wrote:
> See in: https://users.rust-lang.org/t/if-ada-is-already-very-safe-why-rust/21911/47
> 
>> He also told me that Ada compilers aren’t allowed to do certain kinds of optimizations that for example c, c++ (and Rust and other PL via LLVM) are doing.
> 
> How true is this?

In the late 1980s-early 1990s there was a company named Tartan that sold C 
compilers for TI chips. TI chips came with a free C compiler, but people were 
willing to pay for the Tartan compilers because of their excellent optimization.

Tartan also sold Ada compilers with excellent optimization (see the "Ada 
Outperforms Assembly" paper). There were some papers published (probably in /Ada 
Letters/) in which the compiler writers demonstrated that their Ada compilers 
were faster than their C compilers, and discussed reasons why. One reason I 
recall was that in Ada, the compiler knew what was aliased and what wasn't, 
while in C everything is implicitly aliased, and this allowed optimizations of 
Ada that could not be done for C.

So for Ada 83 circa 1990, this statement is clearly false. With another 30 years 
of optimization technology behind us, I don't know for sure, but I suspect that 
hasn't changed.

-- 
Jeff Carter
"Crucifixion's a doddle."
Monty Python's Life of Brian
82