Sign of the times: power supplies, 950W!

G

Guest

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

Yousuf Khan wrote:
> X-bit labs - Hardware news - 950W Power Supply Units Sneak into Market
> http://www.xbitlabs.com/news/other/display/20041126101213.html

Ridiculous - especially considering that those PSU's are
targetted at desktops and not 4P or 8P servers.

I've built a few dual-Opteron workstations and a good
550 or 600 W PSU is plenty for a pair of Opty 250's,
a high-end workstation or gaming graphics card, 8 GB
of RAM, a DVD burner, and four SATA hard drives, plus
a few other little things like video capture cards,
sound cards, etc.
 

cjt

Distinguished
Mar 30, 2004
440
0
18,780
Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Yousuf Khan wrote:
> X-bit labs - Hardware news - 950W Power Supply Units Sneak into Market
> http://www.xbitlabs.com/news/other/display/20041126101213.html

Assume for a moment you actually use something close to 950W in a
24 X 7 machine. At 70% efficiency, you'd use .95 / .7 * 24 * 365 =
11888 KWH in a year. At 10 cents / KWH, that's almost $1200 worth
of electricity.

Personally, I'm trying to move the opposite direction.

--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form che...@prodigy.net.
 

mygarbage2000

Distinguished
Jun 5, 2002
126
0
18,680
Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Sat, 27 Nov 2004 15:56:47 -0500, Yousuf Khan <bbbl67@ezrs.com>
wrote:

>X-bit labs - Hardware news - 950W Power Supply Units Sneak into Market
>http://www.xbitlabs.com/news/other/display/20041126101213.html

Today's P4 already doubles up as a space heater. The same animal,
dual-core, with bigger caches, and possibly higher clock... Throw in
2x high end video card (SLI), some 10k (maybe even 15k) rpm
harddrives, DVD burner...
And don't forget the compressor for freon-filled cooling system...
Wait, this one would have a separate power supply!
;-)
 
G

Guest

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

On Sat, 27 Nov 2004 15:56:47 -0500, Yousuf Khan <bbbl67@ezrs.com>
wrote:

>X-bit labs - Hardware news - 950W Power Supply Units Sneak into Market
>http://www.xbitlabs.com/news/other/display/20041126101213.html

It's actually just marketing. The units apparently are only capable of
some 750W at best and that might still be a marketing number :ppPP


--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me :)
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code
 
G

Guest

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

CJT wrote:
> Assume for a moment you actually use something close to 950W in a
> 24 X 7 machine. At 70% efficiency, you'd use .95 / .7 * 24 * 365 =
> 11888 KWH in a year. At 10 cents / KWH, that's almost $1200 worth
> of electricity.

Yeah, but a PS that is rated for a certain power is never expected to
actually move that much power through it. It's more like a potential
that can be accomodated but hardly ever used.

Even today's most inefficient processors (i.e. Intel Pentium 4's) don't
take up much more than 250W at the plug, so a good PS from several years
ago is still sufficient to power these systems today. I would assume
these 950W monsters are really designed for systems maybe three years
away too.

Yousuf Khan
 
G

Guest

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

Bitstring <hMmdneUbI657mTfcRVn-hA@rogers.com>, from the wonderful person
Yousuf Khan <bbbl67@ezrs.com> said
>CJT wrote:
>> Assume for a moment you actually use something close to 950W in a
>> 24 X 7 machine. At 70% efficiency, you'd use .95 / .7 * 24 * 365 =
>> 11888 KWH in a year. At 10 cents / KWH, that's almost $1200 worth
>> of electricity.
>
>Yeah, but a PS that is rated for a certain power is never expected to
>actually move that much power through it. It's more like a potential
>that can be accomodated but hardly ever used.
>
>Even today's most inefficient processors (i.e. Intel Pentium 4's) don't
>take up much more than 250W at the plug, so a good PS from several
>years ago is still sufficient to power these systems today. I would
>assume these 950W monsters are really designed for systems maybe three
>years away too.

They're probably 'designed' (using the term loosely) for bragging rights
for people with more money than sense. Along with gold plated silver HS,
and Dayglo fans.

--
GSV Three Minds in a Can
Outgoing Msgs are Turing Tested,and indistinguishable from human typing.
 
G

Guest

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

GSV Three Minds in a Can wrote:
>
[...]

> They're probably 'designed' (using the term loosely) for bragging rights
> for people with more money than sense. Along with gold plated silver HS,
> and Dayglo fans.

Imagine the UK Department of Works and Pensions 80,000 PC's of 1KW each;
That alone would take an entire 80 MW power station :)
 
G

Guest

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

Bitstring <41AA2987.6D20B165@sizefitter.com>, from the wonderful person
Johannes H Andersen <johs@sizefitter.com> said
>
>
>GSV Three Minds in a Can wrote:
>>
>[...]
>
>> They're probably 'designed' (using the term loosely) for bragging rights
>> for people with more money than sense. Along with gold plated silver HS,
>> and Dayglo fans.
>
>Imagine the UK Department of Works and Pensions 80,000 PC's of 1KW each;
>That alone would take an entire 80 MW power station :)

Yeah, but think of the hot air output. Any reasonable spin-doctor would
have that publicised as a 10**4 productivity improvement, in a flash.

--
GSV Three Minds in a Can
Outgoing Msgs are Turing Tested,and indistinguishable from human typing.
 

keith

Distinguished
Mar 30, 2004
1,335
0
19,280
Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Sun, 28 Nov 2004 20:15:24 +0000, GSV Three Minds in a Can wrote:

> Bitstring <41AA2987.6D20B165@sizefitter.com>, from the wonderful person
> Johannes H Andersen <johs@sizefitter.com> said
>>
>>
>>GSV Three Minds in a Can wrote:
>>>
>>[...]
>>
>>> They're probably 'designed' (using the term loosely) for bragging rights
>>> for people with more money than sense. Along with gold plated silver HS,
>>> and Dayglo fans.
>>
>>Imagine the UK Department of Works and Pensions 80,000 PC's of 1KW each;
>>That alone would take an entire 80 MW power station :)
>
> Yeah, but think of the hot air output. Any reasonable spin-doctor would
> have that publicised as a 10**4 productivity improvement, in a flash.

....like a government bureaucracy needs more hot air? (not to mention that
80MW is a rather small plant).

--
Keith
 
G

Guest

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

> I've built a few dual-Opteron workstations and a good
> 550 or 600 W PSU is plenty for a pair of Opty 250's,
> a high-end workstation or gaming graphics card, 8 GB
> of RAM, a DVD burner, and four SATA hard drives,
> plus a few other little things like video capture cards,
> sound cards, etc.

Indeed, the article cites later some "small print" caveats:
o Figs are short-duration peak - beginning to sound like PMPO
o Figs are unremarkable 18A on 12V rail - rail grouping limits

Will be interesting to verify ATX compatibility re protection.
Price wise at 400$+ it better have had no corners cut in there.

Would be better if they spent the $ on better components:
o Stick 105oC or 125oC Mil-Spec electrolytics in there
o Most common failure in SMPS, and would cost << 400$

A more realistic industry maximum figure is 670W:
o Dual-Core Dual-CPU + 15.3k-rpm SCSI RAID + DDR2-667 1GB Dimms
---- multi-GB of memory gets very hungry re 10W/Dimm
o Dual-Core Dual-CPU + Very high-end Graphics-Card

In high-end servers you're more interested in N+1 than having
one super-big single-point-of-failure (ok, Gov't IT buyers aside).
Tend to see 950-1150W only in triplicate on multi-blade servers.

The multi-CPU/PCI backplane crowd are more interested in using
multiple ATX PSUs for Cost + Redundancy into a smart backplane.

Potentially you can hit a catch-22 of oversized PSUs re efficiency:
o SMPS tend to get more efficient as power drawn rises
o A very high output PSU is often somewhat inefficient at low load
---- efficiency can be 60% at the low end, 75-80% at the high end
o That low load may not be at the tail end of usage distribution
o On the contrary, it might actually be commonly encountered
---- the average PC draws around ~85W

Longer term P-M architecture will assist thermal dissipation.
o I wish AMD would push 25W Mobile Athlon on Desktop harder
o Mobile Athlon are pretty cheap, and use a cheap desktop board
---- a caveat is 25W Mobile Athlon seem rare (re P-M 21-28W)
---- even so £80+45 isn't too difficult to get something usable
---- cheaper than a Commell Skt370 VIA PL133 Tualatin + CPU
o P-M cpu are pricey, and use expensive proprietary boards
---- £160+189 is pretty steep, altho the board takes a 12V supply
---- easy to find fanless to 50oC DIN-rail single-rail PSUs

If poor low-70% efficiency for a 950W PSU, it's pulling 1350W:
o Ok, that is only be drawn if it were at 100% load
---- which is indicated to be short-term only
o Even so it would be 400W dissipated within the PSU
---- even under ideal conditions that needs 80cfm

Globally industry/regs/gov't have ensured few power plants built.
Economically it wasn't worth building them - until now, and thus
the easily manipulated upwards electricity pricing market. It is not
impossible to design SMPS of 85-90% efficiency, saves money.
Ok, it's not wasted money if you do want the room warming.

Perhaps Enron should have gone into PSUs & CPUs...
--
Dorothy Bradbury
 
G

Guest

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

Bitstring <pan.2004.11.29.01.32.24.206706@att.bizzzz>, from the
wonderful person keith <krw@att.bizzzz> said
<snip>
>> Yeah, but think of the hot air output. Any reasonable spin-doctor would
>> have that publicised as a 10**4 productivity improvement, in a flash.
>
>...like a government bureaucracy needs more hot air?

It's their primary output, far as I can tell. Now =I= don't need it, but
apparently they believe they have to supply it. 8>.

--
GSV Three Minds in a Can
Outgoing Msgs are Turing Tested,and indistinguishable from human typing.
 

keith

Distinguished
Mar 30, 2004
1,335
0
19,280
Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Mon, 29 Nov 2004 01:57:08 +0000, GSV Three Minds in a Can wrote:

> Bitstring <pan.2004.11.29.01.32.24.206706@att.bizzzz>, from the
> wonderful person keith <krw@att.bizzzz> said
> <snip>
>>> Yeah, but think of the hot air output. Any reasonable spin-doctor would
>>> have that publicised as a 10**4 productivity improvement, in a flash.
>>
>>...like a government bureaucracy needs more hot air?
>
> It's their primary output, far as I can tell. Now =I= don't need it, but
> apparently they believe they have to supply it. 8>.

You may not need it nor want it, but you're paying for it in any case!

--
Keith
 
G

Guest

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

GSV Three Minds in a Can wrote:
> Bitstring <hMmdneUbI657mTfcRVn-hA@rogers.com>, from the wonderful person
> Yousuf Khan <bbbl67@ezrs.com> said
>
>> CJT wrote:
>>
>>> Assume for a moment you actually use something close to 950W in a
>>> 24 X 7 machine. At 70% efficiency, you'd use .95 / .7 * 24 * 365 =
>>> 11888 KWH in a year. At 10 cents / KWH, that's almost $1200 worth
>>> of electricity.
>>
>>
>> Yeah, but a PS that is rated for a certain power is never expected to
>> actually move that much power through it. It's more like a potential
>> that can be accomodated but hardly ever used.
>>
>> Even today's most inefficient processors (i.e. Intel Pentium 4's)
>> don't take up much more than 250W at the plug, so a good PS from
>> several years ago is still sufficient to power these systems today. I
>> would assume these 950W monsters are really designed for systems maybe
>> three years away too.
>
>
> They're probably 'designed' (using the term loosely) for bragging rights
> for people with more money than sense. Along with gold plated silver HS,
> and Dayglo fans.
>
They will not be maxed out with two o/c P4's, 4-8GB, and 4 big hard
drives and six or eight fans to keep the inside reasonable. Think four
15k RPM SCSI drives, and a power supply not loaded to spec.

--
bill davidsen (davidsen@darkstar.prodigy.com)
SBC/Prodigy Yorktown Heights NY data center
Project Leader, USENET news
http://newsgroups.news.prodigy.com
 
G

Guest

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

In article <Cp434iEcHjqBFANi@from.is.invalid>,
GSV Three Minds in a Can <GSV@quik.clara.co.uk> wrote:
>Bitstring <41AA2987.6D20B165@sizefitter.com>, from the wonderful person
>Johannes H Andersen <johs@sizefitter.com> said
>>Imagine the UK Department of Works and Pensions 80,000 PC's of 1KW each;
>>That alone would take an entire 80 MW power station :)
>
>Yeah, but think of the hot air output. Any reasonable spin-doctor would
>have that publicised as a 10**4 productivity improvement, in a flash.

....as if your average government agency doesn't have enough hot air already.
:p

_/_
/ v \ Scott Alfter (remove the obvious to send mail)
(IIGS( http://alfter.us/ Top-posting!
\_^_/ rm -rf /bin/laden >What's the most annoying thing on Usenet?
 

TRENDING THREADS