[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

RE: different alpha cpu compatibility issues?



Some of the incompatibilities can be over come with PAL code, however
relying on PAL code to do such things as  Byte/Word manipulations will slow
the older machines to a crawl if you are doing much.  If you compile for the
21064 ( EV4 ) is "safest", however speed improvements could be noticed on
newer processors if the application was optimized for that core.
 
---------------------------------------------------------------------------
Tony Hoover, Network Administrator 
KSU - Salina, College of Technology
mailto:hoover@sal.ksu.edu

"Don't Blend in..."
----------------------------------------------------------------------------


> ----------
> From: 	warp@xs4all.nl[SMTP:warp@xs4all.nl]
> Reply To: 	axp-list@redhat.com
> Sent: 	Tuesday, March 28, 2000 1:12 PM
> To: 	axp-list@redhat.com
> Subject: 	Re: different alpha cpu compatibility issues?
> 
> Eric Ding wrote:
> 
> Hi,
> 
> > I notice that there are various types of machines, all of which fall
> under
> > the category of "alpha" -- for example, in the config.guess of gcc
> 2.95.2,
> > it checks if it's running on alpha vs. alphaev5 vs. alphaev56, etc.
> > 
> > Are there compatibility issues between different alpha boxes?  For
> example,
> > for Applixware, the build machine which we currently use for releases is
> > an EV56 machine, according to /proc/cpuinfo, 
> 
> Yes, for example, code that is specificly optimized for an EV56 CPU
> will not run properly (if at all) on older EV4(5) machine's and can
> produce errors like Signal illegal instruction.
>  
> This is because the EV56 CPU's supports instructions and
> optimizations which are lacking in the older EV4 CPU's.
> 
> (Same applies for optimized EV6 code on a lower CPU model,
>  etc.).
> 
> Hope this helps,
> 
> Regards,
> Antoine
> 
> -- 
> To unsubscribe: send e-mail to axp-list-request@redhat.com with
> 'unsubscribe' as the subject.  Do not send it to axp-list@redhat.com
> 



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index] []