Date: Wed, 13 May 1998 11:58:57 +0300 (IDT)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Nate Eldredge <nate AT cartsys DOT com>
cc: Lennart Marien <lennartm AT metronet DOT de>, djgpp AT delorie DOT com
Subject: Re: Fast Pallette changing?
In-Reply-To: <19980513011428.AAD27360@ppp118.cartsys.com>
Message-ID: <Pine.SUN.3.91.980513115809.26082J-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On Tue, 12 May 1998, Nate Eldredge wrote:

> the I/O instructions are not particularly slow either.

Actually, the I/O instructions *are* slow: they can take as much as 28
cycles (at least on a 486) in protected mode, depending on the CPL vs
IOPL numbers.

Of course, issuing a real-mode interrupt is much more expensive.
 
> That said, the only true answer is to benchmark both methods and see. :)

Hear, hear.  Far too many people are optimizing in the wrong place
because they *think* they know what's faster.