Re: cannot find sparc64.gz
> > I'm on a 333Mhz UltraSPARC IIi and I would find it difficult to describe
> > it as poor. Clock speed doesn't really mean much in terms of real CPU
> > system / performance IMHO. Most of the folks I know turn their noses up
> > at it because it's not up to the same clock speed as their Athalon and
> > then they wonder when my machine out performs theirs.
> I'm on a 333 UIIi too. I use my U10 as my development and multimedia
> workstation and, stopwatch at hand, I sadly have to see that my workstation
> is slower than my old PC (Athlon based, 2000+ processor), and I'm talking
> about real-life tasks (e.g. using the same programs with the same operating
> systems), not mere bogoMIPS calculations.
What sort of tasks? I can believe heavily FP intensive tasks (like
multimedia) are easier on an Athlon (even better on a PPC 9450 Iwould
have thought) but I find that the significant cache on the processor I'm
using makes a big difference on lots of other stuff. Plus the lower (in
terms of clock cycles) latency in memory access seems to make a
difference but that is purely an impression I have nothing to back that
up with.
> Obviously, system's speed is not related to CPU's clock only (for example, my
> CMD646 integrated IDE controller is only capable of multiword DMA versus the
> UltraDMA2 capabilities of my old VIA southbridge on the Athlon m/b (using the
> same hard disk), and the main memory isn't designed to skyrocket at DDR-like
> timings (it was designed years before DDRs, it's understandable).
> Just to make it clear: I like my workstation, I bought it with hard earned
> money and I'm quite happy about it, just I am less than an enthusiast in the
> workstation's performance on the multimedia field (mplayer, as an example,
> even with SDL optim. enabled and the like, can't play a simple
> dvd/divx/mpeg/you name it at a human-recognizable frame rate. I understand
> perfectly that my video card is a 4-Mb model and it was bleeding edge when my
> workstation was built, but the processor's aid in getting things solved in
> this field is not sufficient, and so I visually feel that the system's
> performance is poor.
Hmmm... I'm working on some autovectorisation stuff at the moment - this
might help this sort of app. Any idea where the bottlenecks on this are
/ have you odne any code profiling?
> Never said the mighty UIIi is poor at all, and excuse me
> for the post's length.
If you excuse me for being rabidly anti-x86 and completely going off on
one.
Cheers,
- Martin
--
Martin
inkubus@interalpha.co.uk
"Seasons change, things come to pass"
Reply to: