Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Performance: RM5200 vs. R12K

48 views
Skip to first unread message

Georges A. Tomazi

unread,
Apr 19, 2003, 4:47:28 PM4/19/03
to

Hi -

Is there some benchmarks available on the net comparing QED's RM5200
to MIPS' R12000 CPUs ?

Or perhaps some experiences here?

Thanks,

Georges

--
Georges A. Tomazi - g...@sunwizard.net

Neil Rothwell

unread,
Apr 19, 2003, 5:11:00 PM4/19/03
to
On Sat, 19 Apr 2003 22:47:28 +0200, Georges A. Tomazi <g...@diapason.com>
wrote:

>
>Hi -
>
>Is there some benchmarks available on the net comparing QED's RM5200
>to MIPS' R12000 CPUs ?
>
>Or perhaps some experiences here?

Gleaned from various sources over the years:

All O2s

Proc SPECint95 SPECfp95
12000 400 19.3 13.6
12000 300 14.5 10.4
12000 270 13.1 9.8
10000 250 12.4 9.7
10000 195 10.1 8.8
7000 300 ? 7.5
5200 300 8.0 6.9
10000 175 9.1 6.6
10000 150 7.4 6.2
5000 200 5.3 5.7
5000SC 180 4.8 5.4
5000PC 180 3.7 4.6


--
-------------------------------------------------
Neil Rothwell
Freelance Mechanical Design Engineer
E-Mail: ne...@rothers.demon.co.uk
Home Page: http://www.rothers.demon.co.uk
-------------------------------------------------

yi ge mei you wei ba

unread,
Apr 19, 2003, 5:49:43 PM4/19/03
to
Neil Rothwell wrote:
>

> All O2s
>
> Proc SPECint95 SPECfp95
> 12000 400 19.3 13.6
> 12000 300 14.5 10.4
> 12000 270 13.1 9.8
> 10000 250 12.4 9.7
> 10000 195 10.1 8.8
> 7000 300 ? 7.5
> 5200 300 8.0 6.9
> 10000 175 9.1 6.6
> 10000 150 7.4 6.2
> 5000 200 5.3 5.7
> 5000SC 180 4.8 5.4
> 5000PC 180 3.7 4.6
>


Hmmm. Considering these figures and the fact that they fixed the
video timing problems for the r10-12 series, why did they bother
with the rm-5200 ? Is an r10k-250 *that* much more expensive to
build ? It's not as if the rm5200's are cheap - at least to buy.

<<>>

unread,
Apr 19, 2003, 7:34:46 PM4/19/03
to
In news:3EA1C61D...@pacbell.net, yi ge mei you wei ba wrote:

> Is an r10k-250 *that* much more expensive to build ?

Well I beleive the R1xK has a much larger die area than the R5xK, plus
it is a more complex chip, all of which probably means that yeild is
significantly lower. Cache may be the other factor, I forget the
figures for R1xK and R5xK in O2. Plus it is SGI after all.

> It's not as if the rm5200's are cheap - at least to buy.

If you mean from SGI then *compared* to the 250MHz R10K they were, and
probably still are, much cheaper to buy. I think at the time the split
product line made sense. A R5K/R7K line from $5000 - $10,000 (low end
;) and a R1xK line from $10,000 on up into Octane territory.

If you mean second hand, well that market is always screwy. One can get
a nice 250MHz R10K O2 for just shy of $500 (or even better if you're
handy) but you can't seem to touch a 300MHz R5200 for much less than
$900. And the R5200 modules still sell for more than a complete 250MHz
R10K O2.

Thanks for the numbers Niel. I'd really like to replace my desktop PC
with a high spec 12K/R14K O2 some day, probably when prices aren't
insane.

Chris

--
#include <disclaimer.h>
cm0...@hotmail.com


Stefan Eilemann

unread,
Apr 20, 2003, 7:24:22 AM4/20/03
to
On Sat, Apr 19, 2003 at 09:49:43PM +0000, yi ge mei you wei ba wrote:
> Neil Rothwell wrote:
[spec numbers]

>
> Hmmm. Considering these figures and the fact that they fixed the
> video timing problems for the r10-12 series, why did they bother
> with the rm-5200 ? Is an r10k-250 *that* much more expensive to
> build ? It's not as if the rm5200's are cheap - at least to buy.

In my opinion:

- noise
- second internal hard disk

Stefan, who just downgraded his O2 to R5k.

Georges A. Tomazi

unread,
Apr 20, 2003, 8:44:05 AM4/20/03
to

Hi Neil -

On Sat, 19 Apr 2003 22:11:00 +0100, Neil Rothwell
<ne...@rothers.demon.co.uk> wrote:

[...]

>Gleaned from various sources over the years:

Thanks ! That is precisely what I was looking for !

I was considering upgrading (downgrading ?) my O2/R10K@270MHz to a
RM5200@300MHz so you saved me some bucks...

I'll have to find a R12K/400MHz and it seems to be a pretty hard task.

Georges A. Tomazi

unread,
Apr 20, 2003, 8:45:56 AM4/20/03
to

Hi Stephan -

On Sun, 20 Apr 2003 13:24:22 +0200, Stefan Eilemann <ei...@sgi.com>
wrote:

[...]

>In my opinion:
>
>- noise

Is the O2/R5K that quiet compared to the O2/R10-2K ?

[...]

Andreas Backhaus

unread,
Apr 20, 2003, 3:03:45 PM4/20/03
to
"Georges A. Tomazi" schrieb:

>
> Hi Stephan -
>
> On Sun, 20 Apr 2003 13:24:22 +0200, Stefan Eilemann <ei...@sgi.com>
> wrote:
>
> [...]
>
> >In my opinion:
> >
> >- noise
>
> Is the O2/R5K that quiet compared to the O2/R10-2K ?

I can't find the difference to be that big. It depends more on the disks
you have in it. Since I put that 36GB IBM drive into my O2, it's loud as hell.


Cheers
Andreas

Colin Anderson

unread,
Apr 20, 2003, 5:09:01 PM4/20/03
to
Georges A. Tomazi <g...@diapason.com> wrote in message news:<ma55avghsmknpd09u...@4ax.com>...

I've used both an R12K/300 and an R12K/400 in O2... while both were
much more zippy than the R5K/180 and R10K/195 models I had been using,
I really couldn't feel any difference between the 300 and the 400.

If you already have a 270 and desire more performance, consider
getting an R12K/300 (single or dual) Octane. Processors being the
same, an Octane is much faster for most tasks than an O2. (At least
until more developers take better advantage of the O2's hardware --
examples being Brandon's IRIX DiVX player and Cesar's Pegamento).

- Colin

0 new messages