Sun's 4-way Opteron V40z server

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

I thought it interesting how high the state of the art in x86 server
technology is these days. The range of features I'm seeing here were
only available in enterprise Unix servers of only a couple of years
ago. Just about everything seems to be hot-swappable, and there's
lights-out management capabilities, among other things.

http://www.anandtech.com/printarticle.aspx?i=2354

Yousuf Khan

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On 28 Mar 2005 13:58:40 -0800, "YKhan" <yjkhan@gmail.com> wrote:

>I thought it interesting how high the state of the art in x86 server
>technology is these days. The range of features I'm seeing here were
>only available in enterprise Unix servers of only a couple of years
>ago. Just about everything seems to be hot-swappable, and there's
>lights-out management capabilities, among other things.
>
>http://www.anandtech.com/printarticle.aspx?i=2354
>

$5000/processor gets you a nice Altix with a real processor. ;-).

RM

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Mon, 28 Mar 2005 17:21:39 -0500, Robert Myers
<rmyers1400@comcast.net> wrote:

>On 28 Mar 2005 13:58:40 -0800, "YKhan" <yjkhan@gmail.com> wrote:
>
>>I thought it interesting how high the state of the art in x86 server
>>technology is these days. The range of features I'm seeing here were
>>only available in enterprise Unix servers of only a couple of years
>>ago. Just about everything seems to be hot-swappable, and there's
>>lights-out management capabilities, among other things.
>>
>>http://www.anandtech.com/printarticle.aspx?i=2354

Them Newisys boys did their homework!

>$5000/processor gets you a nice Altix with a real processor. ;-).

And no software to run on that processor :>

Actually it would be rather nice to see a straight-up, head to head
comparison of this Opteron-based Sun Fire v40z vs. the Itanium based
SGI Altrix 350 vs. the Power5 based IBM p630. All three sell in the
same basic price range, all are backed by (presumably) solid support
and all run Linux.

Of course, doing such a task would probably be a bit beyond most of
the hardware review websites, not to mention the fact that it would be
extraordinarily difficult to give a really meaningful picture of
performance for more than a narrow handful of applications.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

"Robert Myers" <rmyers1400@comcast.net> wrote in message
news:hn0h411js4i102fp3e60o9d54ab0vvg46r@4ax.com...
> On 28 Mar 2005 13:58:40 -0800, "YKhan" <yjkhan@gmail.com> wrote:
>
> >I thought it interesting how high the state of the art in x86 server
> >technology is these days. The range of features I'm seeing here were
> >only available in enterprise Unix servers of only a couple of years
> >ago. Just about everything seems to be hot-swappable, and there's
> >lights-out management capabilities, among other things.
> >
> >http://www.anandtech.com/printarticle.aspx?i=2354
> >
>
> $5000/processor gets you a nice Altix with a real processor. ;-).
>
> RM

Or a power5 box for that matter. I5?

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

Is that all an Altix costs? $20,000 for a 4P system?

Yousuf Khan

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On 29 Mar 2005 09:43:01 -0800, "YKhan" <yjkhan@gmail.com> wrote:

>Is that all an Altix costs? $20,000 for a 4P system?
>

This is a year-old link, but it still works:

http://news.zdnet.com/2100-9584_22-5138719.html

RM

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Tue, 29 Mar 2005 02:08:46 GMT, "Delbert Cecchi"
<dcecchi_nospam@worldnet.att.net> wrote:

>
>"Robert Myers" <rmyers1400@comcast.net> wrote in message
>news:hn0h411js4i102fp3e60o9d54ab0vvg46r@4ax.com...
>> On 28 Mar 2005 13:58:40 -0800, "YKhan" <yjkhan@gmail.com> wrote:
>>
>> >I thought it interesting how high the state of the art in x86 server
>> >technology is these days. The range of features I'm seeing here were
>> >only available in enterprise Unix servers of only a couple of years
>> >ago. Just about everything seems to be hot-swappable, and there's
>> >lights-out management capabilities, among other things.
>> >
>> >http://www.anandtech.com/printarticle.aspx?i=2354
>> >
>>
>> $5000/processor gets you a nice Altix with a real processor. ;-).
>>
>> RM
>
>Or a power5 box for that matter. I5?

No intention of slighting IBM's nice offerings, either.

http://www-1.ibm.com/servers/eserver/openpower/hardware/720_browse.html

RM

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

Robert Myers wrote:
> On 29 Mar 2005 09:43:01 -0800, "YKhan" <yjkhan@gmail.com> wrote:
>
>
>>Is that all an Altix costs? $20,000 for a 4P system?
>>
>
>
> This is a year-old link, but it still works:
>
> http://news.zdnet.com/2100-9584_22-5138719.html

It seems to be atypically priced for that market category. According to
this statement:

> Among the Altix 350's competitors are IBM's p650, an eight-processor system with a starting price of about $27,000 for a dual-processor configuration; Hewlett-Packard's four-processor rx5670, with a starting price of about $22,000; and Sun Microsystems' V880z, an eight-processor machine with a starting price of about $60,000.


Also this statement makes no sense:

> Moving to Intel's Itanium chips and to Linux offers the company a chance to reduce its research and development expenses.

If it's trying to reduce R&D $'s, then what's it doing developing things
like NUMAflex specifically for Itanium anyways?

Yousuf Khan

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Tue, 29 Mar 2005 15:12:29 -0500, Yousuf Khan <bbbl67@ezrs.com>
wrote:

>Robert Myers wrote:
>> On 29 Mar 2005 09:43:01 -0800, "YKhan" <yjkhan@gmail.com> wrote:
>>
>>
>>>Is that all an Altix costs? $20,000 for a 4P system?
>>>
>>
>>
>> This is a year-old link, but it still works:
>>
>> http://news.zdnet.com/2100-9584_22-5138719.html
>
>It seems to be atypically priced for that market category. According to
>this statement:
>
>> Among the Altix 350's competitors are IBM's p650, an eight-processor system with a starting price of about $27,000 for a dual-processor configuration; Hewlett-Packard's four-processor rx5670, with a starting price of about $22,000; and Sun Microsystems' V880z, an eight-processor machine with a starting price of about $60,000.
>
But there is "Hewlett-Packard's four-processor rx5670, with a starting
price of about $22,000."

My point wasn't intended to be taken altogether seriously, as
indicated by the ;-) in my first post to this thread. $/processor is
just one very crude way of normalizing the list prices of computers,
which prices themselves may bear little resemblance to the price at
which hardware is actually sold, especially if you count all the
extras that are thrown in to an actual sale.

>
>Also this statement makes no sense:
>
>> Moving to Intel's Itanium chips and to Linux offers the company a chance to reduce its research and development expenses.
>
>If it's trying to reduce R&D $'s, then what's it doing developing things
>like NUMAflex specifically for Itanium anyways?
>
NUMAflex doesn't have anything to do specifically with Itanium. If
Itanium went away, you might see Opteron connected with NUMAflex. The
Altix 350 allows you to run a single system image on a cluster of four
boxes with 4P each, something you can't do on a do-it-yourself
cluster. SGI uses infiniband to create larger clusters.

RM

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

Robert Myers wrote:
>>>Moving to Intel's Itanium chips and to Linux offers the company a chance to reduce its research and development expenses.
>>
>>If it's trying to reduce R&D $'s, then what's it doing developing things
>>like NUMAflex specifically for Itanium anyways?
>>
>
> NUMAflex doesn't have anything to do specifically with Itanium. If
> Itanium went away, you might see Opteron connected with NUMAflex. The
> Altix 350 allows you to run a single system image on a cluster of four
> boxes with 4P each, something you can't do on a do-it-yourself
> cluster. SGI uses infiniband to create larger clusters.

But still, it must require some redesigns to get it to interface with
each different kind of bus architecture to which it needs to connect,
while keeping the latency down.

Yousuf Khan

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Mon, 04 Apr 2005 13:23:04 -0400, Yousuf Khan <bbbl67@ezrs.com>
wrote:

>Robert Myers wrote:
>>>>Moving to Intel's Itanium chips and to Linux offers the company a chance to reduce its research and development expenses.
>>>
>>>If it's trying to reduce R&D $'s, then what's it doing developing things
>>>like NUMAflex specifically for Itanium anyways?
>>>
>>
>> NUMAflex doesn't have anything to do specifically with Itanium. If
>> Itanium went away, you might see Opteron connected with NUMAflex. The
>> Altix 350 allows you to run a single system image on a cluster of four
>> boxes with 4P each, something you can't do on a do-it-yourself
>> cluster. SGI uses infiniband to create larger clusters.
>
>But still, it must require some redesigns to get it to interface with
>each different kind of bus architecture to which it needs to connect,
>while keeping the latency down.
>
Itanium was (and is) SGI's chosen successor to MIPS. One might ask
what other choices SGI might have made, but they already have a
significant investment in Itanium.

RM

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

Robert Myers wrote:
>>But still, it must require some redesigns to get it to interface with
>>each different kind of bus architecture to which it needs to connect,
>>while keeping the latency down.
>>
>
> Itanium was (and is) SGI's chosen successor to MIPS. One might ask
> what other choices SGI might have made, but they already have a
> significant investment in Itanium.

Which brings it back to, if it was trying to save money on R&D costs,
why isn't it just using standard Itanium chipsets?

Yousuf Khan

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Tue, 05 Apr 2005 01:00:47 -0400, Yousuf Khan <bbbl67@ezrs.com>
wrote:

>Robert Myers wrote:
>>>But still, it must require some redesigns to get it to interface with
>>>each different kind of bus architecture to which it needs to connect,
>>>while keeping the latency down.
>>>
>>
>> Itanium was (and is) SGI's chosen successor to MIPS. One might ask
>> what other choices SGI might have made, but they already have a
>> significant investment in Itanium.
>
>Which brings it back to, if it was trying to save money on R&D costs,
>why isn't it just using standard Itanium chipsets?
>

You could have asked the same question of HP, which blew the
competition away with its zx1 chipset for Itanium. See my post under
the hurricane thread.

One possible explanation is that the Intel (and IBM) chipsets were
aimed at a different market. I'm sure that SGI would love to be
selling into the server market, but I don't think they do. Another
reason why my comment about the Altix wasn't intended to be taken
seriously.

RM