Re: What are the differences between x86 and Sparc

From: Tuesday (no1home_at_comcast.net)
Date: 07/30/03


Date: Wed, 30 Jul 2003 01:03:25 GMT


"Will Hartung" <willh@msoft.com> wrote in message
news:bg71vj$lcpeu$1@ID-197644.news.uni-berlin.de...
> "Daniel Seichter" <daniel@dseichter.de> wrote in message
> news:bg5tel$e8f$1@news.dtag.de...
> > Hello,
> > can anyone tell me the main differences between running solaris on x86
or
> on
> > sparc?
> > Which speed is the Ultra 150 with 650MHz equal to a x86 architecture?
> >
> > I hope to find some answers here, because I don't know, if I should buy
a
> > XEON or a SPARC
>
> I can't answer your question specifically, and of course it really depends
> on your application and needs.
>
> But when I went shopping for a use Blade 100, I ended up with an Ultra 10.
> The compelling difference for me was that while the MHz was faster on the
> Blade (The Ultra was 440 vs the Blade 500), the Ultra had 4X the chip
cache
> (2MB vs 512K). Now, they may be different processors as well (I don't
really
> know), but with the lower price of Ultras and the higher cache, I felt it
> was actually a better value. I think for many apps, I got a faster machine
> for less money.
>
> How it compares with Intel stuff, I haven't any idea.
>
> Regards,
>
> Will Hartung
> (willh@msoft.com)
>
>

2MB of cache on the UltraSPARC IIi is direct mapped, 512K on the UltraSPARC
IIe
(which for bizarre marketing reasons Sun renamed to UltraSPARC IIi) is 4-way
set
associative. For that reason the two caches are comparable in overall terms
of per-
formance, except for large datasets...in which case the larger cache is
better, even if
it is direct mapped.

MasterToyFixer



Relevant Pages

  • Re: Should we build i386 or i686 rpms?
    ... Yes, more registers and more cache, and things get much faster. ... Too bad Alpha never really cought on as mainstream processor. ... My experience with Alpha processors and competing 32/64-bit counterparts ... If 256 megs was enough to make 32-bit application happy on UltraSparc, ...
    (Fedora)
  • SUMMARY: 240 vs 280
    ... In addition to the number of PCI slots - ... They basically pointed out that the 240 has the Sparc IIIi w/ only 1MB cache ... This should not be a big thing for a Samba/Xinet/NFS file server, ...
    (SunManagers)
  • Re: x86, ARM, PARISC, PPC, MIPS and Sparc folks please run this
    ... > So explain this from the Sparc architecture code: ... sparc and sparc64 both have it. ... >> kernel has access to the more efficient cache flushing instructions. ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)
  • SUMMARY: V440 or V480 for large email server?
    ... was a bit dated and it is time to look for replacement. ... > We are looking for replacing our old 280R server running Sun Java Messaging ... > L2 cache play a role in email application. ... should I go for UltraSPARC IIIi or UltraSPARC III? ...
    (SunManagers)
  • Re: reducing number consing / cache experimental results
    ... it seems to be specific to CMU-CL and Sparc v9. ... In my experiments you have to exceed the L2 cache size to see things ... exceeding L2 cache, ... The last time I ran it with his benchmark ...
    (comp.lang.lisp)