Variant vs. Tagged Type

Variant vs. Tagged Type

[Frank Piron]

Hi,
is it possible to make some general statements
concerning the alternative variant vs. tagged
type with respect to runtime performance?
For example:
if we have an array A 1) of type access to a variant
2) of type access <TYPE>'Class
what can be said about the runtime performance
of the expression
A(I).<component>?
Thanks in Advance,
Frank

Re: Variant vs. Tagged Type

[Marin David Condic]

Well, in most respects, a tagged type is just a special case of a 
variant record. (Conceptually) The "variant" is the tag. In that sense, 
most access should be just as fast either way - depending on compilers 
and optimizations, etc. (The only way to know for sure is to measure 
with *your* compiler in *your* environment with code that is typical of 
what you want to do. Maybe Ada should develop a performance benchmark 
suite to test some of these things?)

However, you do have to be careful with tagged types in that some things 
cannot be done statically at compile-time. Things like dispatching to 
operations on the tagged type. In general, the static things should be 
no worse than any other sort of operation on a variant record, but if 
you use features that are dynamic, you'll incur some additional 
overhead. (How much? You can only know by checking into your compiler.)

OTOH, you also need to ask just how critical the time concerns are. In 
many apps, the slight additional overhead connected to tagged types is 
not enough to cause any sort of problem. In other apps, you might be so 
tight on the time budget that even the variant record approach might be 
too much to bear. (The code has to look up the variant at run time to 
know how to interpret the record, right? That's not free.) Lots of 
designers get their panties in a bunch over performance concerns when a 
step back and a deep breath reveals its all "the small stuff." Like life 
in general, we often worry about things that never happen.

MDC


Frank Piron wrote:
> Hi,
> is it possible to make some general statements
> concerning the alternative variant vs. tagged
> type with respect to runtime performance?
> For example:
> if we have an array A 1) of type access to a variant
> 2) of type access <TYPE>'Class
> what can be said about the runtime performance
> of the expression
> A(I).<component>?
> Thanks in Advance,
> Frank
> 


-- 
======================================================================
Marin David Condic
I work for: http://www.belcan.com/
My project is: http://www.jsf.mil/NSFrames.htm

Send Replies To: m   o   d   c @ a   m   o   g
                    c   n   i       c   .   r

     "So if I understand 'The Matrix Reloaded' correctly, the Matrix is
     basically a Microsoft operating system - it runs for a while and
     then crashes and reboots. By design, no less. Neo is just a
     memory leak that's too hard to fix, so they left him in... The
     users don't complain because they're packed in slush and kept
     sedated"

         --  Marin D. Condic
======================================================================

Re: Variant vs. Tagged Type

[Steve]

"Frank Piron" <nospam@konad.net> wrote in message
news:oprx5zpwdbz12uz9@news.onlinehome.de...
> Hi,
> is it possible to make some general statements
> concerning the alternative variant vs. tagged
> type with respect to runtime performance?

Only for a specific Ada compiler.
Actual implementation of any feature within different compilers may differ
radically.  The only way to compare performance is through testing with a
specific compiler.  In certain cases you may find order-of-magnitude
differences in performance between implementations.

Steve
(The Duck)


> For example:
> if we have an array A 1) of type access to a variant
> 2) of type access <TYPE>'Class
> what can be said about the runtime performance
> of the expression
> A(I).<component>?
> Thanks in Advance,
> Frank
>

Re: Variant vs. Tagged Type

[Jeffrey Carter]

Frank Piron wrote:

> is it possible to make some general statements
> concerning the alternative variant vs. tagged
> type with respect to runtime performance?
> For example:
> if we have an array A 1) of type access to a variant
> 2) of type access <TYPE>'Class
> what can be said about the runtime performance
> of the expression
> A(I).<component>?

It sounds as if you're engaging in premature optimization (the root of 
all evil). Do it whichever way is best suited to the problem, and see if 
the result is fast enough. 99.99% of the time, the microsecond 
differences you're worrying about don't make a difference.

-- 
Jeff Carter
"Many times we're given rhymes that are quite unsingable."
Monty Python and the Holy Grail
57

Re: Variant vs. Tagged Type

[Frank Piron]

Am Wed, 05 Nov 2003 17:40:45 GMT hat Jeffrey Carter <spam@spam.com> 
geschrieben:

> Frank Piron wrote:
>
>> is it possible to make some general statements
>> concerning the alternative variant vs. tagged
>> type with respect to runtime performance?
>> For example:
>> if we have an array A 1) of type access to a variant
>> 2) of type access <TYPE>'Class
>> what can be said about the runtime performance
>> of the expression
>> A(I).<component>?
>
> It sounds as if you're engaging in premature optimization (the root of 
> all evil). Do it whichever way is best suited to the problem, and see if 
> the result is fast enough. 99.99% of the time, the microsecond 
> differences you're worrying about don't make a difference.
>
Currently i'm working on an ORACLE access library on top of
oci. Yesterday i've done some benchmark tests. I experienced
differences in runtime performance up to a factor of 4 while
fetching and extracting 10_000 rows in using a variant instead
of an array of class wide pointers for the db_row_type. The
compiler is gnat 3.15p on windows.
Fortunately ada gives alternatives ;-)
Frank