[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4EBABCB9.7020608@hushmail.com>
Date: Wed, 09 Nov 2011 18:47:37 +0100
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: linux-x86-64i target, using sse-intrinsics.S
2011-10-29 01:46, magnum wrote:
> I just posted (to the wiki, patch 0035) a first take on implementing the
> pre-built sse-intrinsics.S just like the one j5c4 had, but optional.
>
> New make target: linux-x86-64i
>
> I also added a make target (make intrinsics) for rebuilding the .S file.
> This must be performed manually (just removing the .S file will also
> initiate a rebuild).
>
> I'm not sure how we should proceed to enable this for all x86-64 targets
> in a clever way. I don't like the idea of 10 more targets. Maybe we
> could implement them without including them in the target list, but
> print something like "append 'i' to any x86-64 target for icc asm". But
> this 'i' might be confused with the current sse2i targets. Anyone with
> ideas go ahead and make a patch (I'm fine with just a linux target).
Like I just replied to the slighly incorrect jumbo-8 announcement today,
we need to decide how to handle this.
1. We need to decide how to "design" the make targets (or rather the
target *names*). Like I originally wrote above, adding 10 new targets
will make a very long list. One alternative is to *always* use the icc
file whenever possible, with no special targets. I'm not sure what is best.
2. Should we support this for 32-bit at all? I suppose I can cross
compile a 32-bit .S file with icc (haven't tried it) but I have no idea
if it will perform better or worse than gcc on a 32-bit machine. I
suppose this should be verified on a machine that is really 32-bit.
magnum
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
