Path: news.eternal-september.org!eternal-september.org!feeder3.eternal-september.org!i2pn.org!i2pn2.org!.POSTED!not-for-mail
From: quadibloc <quadibloc@gmail.com>
Newsgroups: comp.arch
Subject: Why I've Dropped In
Date: Mon, 19 May 2025 19:15:36 +0000
Organization: novaBBS
Message-ID: <0c857b8347f07f3a0ca61c403d0a8711@www.novabbs.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Injection-Info: i2pn2.org;
	logging-data="1065765"; mail-complaints-to="usenet@i2pn2.org";
	posting-account="GSAUMsvIs05PgSAevbIzdWiOy1BcuThtiv166p5NnMk";
User-Agent: Rocksolid Light
X-Rslight-Posting-User: 7260c650ae4d5ba82d3b6b1eab0ac1b8653ff052
X-Rslight-Site: $2y$10$ZOmn88yAAa13N3K5njhNyet0EdmVzIlzUZqsA8S3QCFaiZDzcUaR.
X-Spam-Checker-Version: SpamAssassin 4.0.0

I haven't posted in this newsgroup for a while.
In case anyone has wondered why I've visited once again, it's for the
obvious reason. I've turned my attention again to Concertina II.
I thought that I had a great idea to reduce the overhead for certain
block types. Then I saw it had a flaw, But I could fix that flaw, but it
resulted in an inelegant asymmetry. Then I saw how to fix that! And then
I realized the fix still was missing something important.
So there will be some delay before I take another step in going around
in circles with Concertina II.

My brilliant idea? Reduce the displacements from 16 bits to 15, so that
memory-reference instructions could fit in 1/4 of the opcode space.
Then I could have blocks with variable-length instructions, as long as I
didn't allow instructions to cross block boundaries, with only 16 bits
of overhead.
But I knew I would never stay satisfied with only 15-bit displacements.
On further thought, I thought that I could still have 16-bit
displacements, and have a complete instruction set. But I ended up
realizing that I wouldn't have a complete instruction set for the blocks
with variable-length instructions.

John Savard