Path: ...!3.eu.feeder.erje.net!feeder.erje.net!news.in-chemnitz.de!news.swapon.de!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: jgd@cix.co.uk (John Dallman)
Newsgroups: comp.arch
Subject: Re: except what, is Vax addressing sane today
Date: Sun, 22 Sep 2024 12:30 +0100 (BST)
Organization: A noiseless patient Spider
Lines: 21
Message-ID: <memo.20240922123047.19028M@jgd.cix.co.uk>
References: <fcc1d80b06e991bf9a27b64ff973b720@www.novabbs.org>
Reply-To: jgd@cix.co.uk
Injection-Date: Sun, 22 Sep 2024 13:30:49 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="63e86f897e11eef93d7dccf1cca35363";
	logging-data="2316505"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX18jcRLMDu9QAiLrer7BrXjrF78NQrEa7vg="
Cancel-Lock: sha1:6jfHJOExbAlZNmYGl5/22AqZxgw=
X-Clacks-Overhead-header: GNU Terry Pratchett
Bytes: 1977

In article <fcc1d80b06e991bf9a27b64ff973b720@www.novabbs.org>,
mitchalsup@aol.com (MitchAlsup1) wrote:
> On Sun, 22 Sep 2024 0:14:49 +0000, Chris M. Thomasson wrote:
> > On 9/21/2024 4:29 PM, Lawrence D'Oliveiro wrote:
> >> Aren't branches that are not taken supposed to be fast?
> > Well, they are not taken, so they should be faster... ;^)
> It is NOT the speed, it is the code bloat.

Yup. Bigger code is always a potential problem, not so much because it
takes up RAM nowadays, but because it takes up memory bandwidth and cache
space. Using up cache space is always bad, because bigger caches are
slower, and instructions seem naturally smaller than cache blocks. 

Wanting smaller code isn't an argument against RISC, but an argument
against poorly optimised ISA design. Variable-length CISC makes it easier
to get smaller average instruction sizes but has other drawbacks. 

For the stuff I work, on ARM64 code is consistently smaller than x86-64,
although the factor varies by platform.

John