Path: csiph.com!x330-a1.tempe.blueboxinc.net!newsfeed.hal-mli.net!feeder3.hal-mli.net!newsfeed.hal-mli.net!feeder1.hal-mli.net!border3.nntp.dca.giganews.com!Xl.tags.giganews.com!border1.nntp.dca.giganews.com!nntp.giganews.com!local2.nntp.dca.giganews.com!news.giganews.com.POSTED!not-for-mail
NNTP-Posting-Date: Sat, 04 Feb 2012 00:49:31 -0600
Message-ID: <4F2CD4FA.4050004@SPAM.comp-arch.net>
Date: Fri, 03 Feb 2012 22:49:30 -0800
From: "Andy (Super) Glew" <andy@SPAM.comp-arch.net>
Reply-To: andy@SPAM.comp-arch.net
Organization: comp-arch.net
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:9.0) Gecko/20111222 Thunderbird/9.0.1
MIME-Version: 1.0
Newsgroups: comp.arch
Subject: Re: M68k add to memory is not a mistake any more
References: <ggtgp-8D1AEA.03180231012012@netnews.mchsi.com> <jgeu27$9rg$1@dont-email.me> <a7csv8-emn1.ln1@ntp6.tmsw.no> <jgiibd$hkq$1@dont-email.me>
In-Reply-To: <jgiibd$hkq$1@dont-email.me>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Lines: 27
X-Usenet-Provider: http://www.giganews.com
X-Trace: sv3-1jfsBnAEAAFoVXg/4uY3B658P+6eQcc19nGgQLioKURyup7QaqVyhBsKjk77FHC0Mp2y2R3W+aqS7Fe!45KDH8Q3M5fTCrM23OV5hNO43h81cOJpDQNLpWFPeOipIhqd1Q+gVXBRuzHxFRU=
X-Complaints-To: abuse@giganews.com
X-DMCA-Notifications: http://www.giganews.com/info/dmca.html
X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers
X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly
X-Postfilter: 1.3.40
X-Original-Bytes: 2564
Xref: x330-a1.tempe.blueboxinc.net comp.arch:5696

On 2/3/2012 10:14 PM, Stephen Sprunk wrote:
> On 03-Feb-12 00:42, Terje Mathisen wrote:
>> Stephen Sprunk wrote:
>>> I'm a fan of load/store architectures and dislike mem-op-mem (and
>>> mem-op, for that matter) instructions in general, but the latter seem to
>>> be back in favor these days due to higher code density and therefore
>>> higher utilization of decode bandwidth.
>>
>> I do like mem-op, even if I've been conditioned (by the 486 and Pentium)
>> to avoid them and instead manually schedule the load part to happen as
>> early as possible.
>
> Of course, with a larger register set and no mem-op instructions, it
> would have been much easier for compilers to do load hoisting as well,
> not to mention much lower stack pressure due to x86's limited register set.
>
> I wonder if x86 could have gone another generation or two without OoO if
> it hadn't become necessary so soon for the CPU to crack mem-op
> instructions and do its own load hoisting.  That transistor budget could
> have, instead, gone into wider execution and/or larger caches.

Possibly.

However, I think it is worth noting that OoO P6 x86 is really the Intel 
chip that killed the RISC upstarts.

All of the competitors could have played the same games.