[mercury-users] Mercury IL and Managed C++

[Jonathan Morgan]

Ralph Becket wrote:
> Jonathan Morgan, Thursday, 17 November 2005:
> > ...though I think C is likely to be best for most
> > applications (and probably faster too).
>
> You might be surprised :-)
>
> You'll probably take five to ten times as long getting the first
> version of your C application running, time which you could have spent
> optimising your Mercury program.  And Mercury isn't that much slower
> than C: I reimplemented a SPEC95 compression benchmark and with a modest
> amount of effort had the Mercury version taking 1.4x the time of the C
> version.  The difference is likely to be smaller in real applications:
> this particular SPEC benchmark involved no memory management and so was
> very much on C's turf.

Just to clarify that, I was talking about the C backend (either one, in
fact) of the Mercury language, and comparing it with the IL backend of the
said Mercury language.  I wouldn't be foolish enough to enter an argument on
speed when faced with one of the fastest logic programming languages in the
world.  I was just commenting that the IL backend has probably had less work
on optimising it than the C backend (though it would get all the generic
Mercury optimisations, I suppose), and that IL is likely to be slower than C
anyway.

Anyway, I thought the Mercury theory was that it took longer to make a
working version of a program, but shorter to make a correctly working
program.

Jon

--------------------------------------------------------------------------
mercury-users mailing list
post:  mercury-users at cs.mu.oz.au
administrative address: owner-mercury-users at cs.mu.oz.au
unsubscribe: Address: mercury-users-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-users-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------