PSUs for new Intel chipsets?

J

John Smith

Hi,

Anyone know what PSUs will be required for the new Intel chipsets? I make
quiet PCs for people and usually recommend the Antec Sonata with a 380 watt
PSU but I wonder if it will be powerful enough for the new Intel stuff that
has just been announced.

Any ideas as to what PSUs will be needed as a minimum?

Thanks,


J.
 
M

Mavra Chang

For a stand alone PSU the Tagan TG480-U01 480 Watt Silent seems to be a good
bet. Reasonable and very quiet.

mc
 
G

Guest

John Smith said:
Anyone know what PSUs will be required for the new Intel
chipsets? I make quiet PCs for people and usually recommend
the Antec Sonata with a 380 watt PSU but I wonder if it will
be powerful enough for the new Intel stuff that
has just been announced.

Someone found that a comparable supply, a 350W Fortron (Sparkle,
Powerman, Hi-Q, Aopen, etc.), was good enough for a Pentium4 system
until it was overclocked to about 3.4 GHz, so my guess is that a 380W
Sonata will squeak through. There are power estimation worksheets,
with the best probably being at www.pcpowerandcooling.com and
http://takaman.jp. However the former assumes that the supply is
first-rate, which may be why it says 275W is enough for almost
anything with a single CPU and not too many disk drives, while the
latter overestimates power needs, although not as badly as most. The
only way to really know what a computer consumes is by taking amp
measurements, the most practical way being with a clamp-on AC/DC
ammeter, but beware that most clamp-ons measure only AC amps.
 
D

Dorothy Bradbury

There are 2 specs for the VRM on Intel motherboards/chipsets,
with it likely to exceed the 110-120Amp mark eventually. It's
not so much the chipset, but graphics card people may add.

High-end graphics card wattage is continuing to scale higher,
at a faster-rate, than the processors - which have slowed up.
Motherboard cooling, not just CPU cooling, due to the VRM
for memory & processor is the main issue on new chipsets.

A 350W PSU will still power a top-end CPU & motherboard,
it just comes down to what HDs, optical drives & graphics card.
Mobile Athlon on a desktop board shows how horrible Prescott is.
 
J

John R Weiss

Dorothy Bradbury said:
There are 2 specs for the VRM on Intel motherboards/chipsets,
with it likely to exceed the 110-120Amp mark eventually.

I hope not! That's 330-360 watts alone at 3 volts, and up to 1440 watts at
12V! You'd need some awfully heavy-duty wiring for that kind of amperage...
 
G

GSV Three Minds in a Can

from the wonderful person said:
I hope not! That's 330-360 watts alone at 3 volts, and up to 1440 watts at
12V! You'd need some awfully heavy-duty wiring for that kind of amperage...

I'd assume that the 110 Amps is for the VCore supply .. i.e. somewhere
under 1.5v. It's still a pretty ludicrous power consumption for a Usenet
reader.
 
J

Johannes H Andersen

John said:
I hope not! That's 330-360 watts alone at 3 volts, and up to 1440 watts at
12V! You'd need some awfully heavy-duty wiring for that kind of amperage...

That's why there is also a switching circuit directly on the motherboard next
to the CPU. The CPU Voltage is about 1.5V. The PSU has 12V and 5V lines.
 
D

David Maynard

John said:
I hope not! That's 330-360 watts alone at 3 volts, and up to 1440 watts at
12V! You'd need some awfully heavy-duty wiring for that kind of amperage...

I don't know how you're deriving your numbers but they don't add up.

The current she's talking about is Vcore, which would be 1.5V, or
significantly less for a 'future' processor. Actually, even for a 1.5V core
P-4 it's already significantly less as Vcore drops with current draw and at
70 amps it's supposed to be down by .15V. So, we can safely presume only
1.35 volt, or (more likely) less, Vcore at 110 amps for 149 watts.

Now, Vcore is feed by the 12 volt line but through a switching regulator
so, to get current draw on the 12 volt rail, ignoring regulator efficiency
for the moment, we use the voltage ratio as power is constant (power in HAS
to equal power out or else something blows up, which adds to the power out
and we again have power in = power out). That is 1.35/12 for 0.1125 and
then times the 110 amps for 12.375 amps on the 12 volt side. If we assume
regulator efficiency is 95% (the rest goes to heat) then the 12V current
draw becomes 12.99375 amps; might as well say 13 amps.

This, btw, is why they switched to using the 12 volt rail for the core
regulator instead of the 3.3 or 5 volt rails: to lower the current on the
PSU wires/mobo connector as they have resistance that introduces losses
proportional to the current.

Note that the 12 volt wattage is the same, plus the regulator loss.
 
D

Dorothy Bradbury

I'd assume that the 110 Amps is for the VCore supply .. i.e. somewhere
under 1.5v. It's still a pretty ludicrous power consumption for a Usenet
reader.

Correct - VCore.

Agreed - power consumption is getting out of hand, the sooner we get
P-M architecture in the desktop the better vs Prescott "3-bar-fire". We
can have it now, as Mobile-Athlon & Desktop boards show clearly.

BTX improves matters a little, but only so they can jack heat a bit higher.
I think 150W will remain as the limit for desktop - server may exceed it,
at least in "theoretical output" terms re 64-bit, dual-core and such like.
 
G

GB

Dorothy Bradbury said:
Correct - VCore.

Agreed - power consumption is getting out of hand, the sooner we get
P-M architecture in the desktop the better vs Prescott "3-bar-fire". We
can have it now, as Mobile-Athlon & Desktop boards show clearly.

BTX improves matters a little, but only so they can jack heat a bit higher.
I think 150W will remain as the limit for desktop - server may exceed it,
at least in "theoretical output" terms re 64-bit, dual-core and such like.

I had never really thought about how much heat is being generated by the
CPU. You're saying these things generate 150 watts inside a bit of
semi-conductor the size of a penny. They make that work!!!?????
 
D

David Maynard

GB said:
I had never really thought about how much heat is being generated by the
CPU. You're saying these things generate 150 watts inside a bit of
semi-conductor the size of a penny. They make that work!!!?????

Well, currently (no pun intended) it's more like 85 watts for Northwood and
Athlon XP with Prescott getting into the > 100 watt range but yes, that's
what she's saying.

Now you know why that CPU heatsink/fan is needed =:O)
 
G

GSV Three Minds in a Can

Bitstring <[email protected]>, from the wonderful
person GB said:
I had never really thought about how much heat is being generated by the
CPU. You're saying these things generate 150 watts inside a bit of
semi-conductor the size of a penny. They make that work!!!?????

Not today; today they can only manage a miserable 80w or so (and you
have to run some rather cute software to get enough of the CPU switching
at once to achieve that level). Still pretty impressive (or dumb,
depending on whether you believe in global warming or not, and how cold
a climate you run your PC in).

That's what the big chunk of Cu/Al, and the jet-engine fan is for.
 
T

Tim Auton

GB said:
I had never really thought about how much heat is being generated by the
CPU. You're saying these things generate 150 watts inside a bit of
semi-conductor the size of a penny. They make that work!!!?????

Not yet. As has been mentioned 80W or so is more typical of today's
machines. But crank up the Vcore and clock speed a bit, as some people
like to do, and you can see why water cooling, heatpipes and huge
copper sculptures are required.

Apparently it takes 11 minutes to cook an egg on an Athlon XP 1500:

http://www.phys.ncku.edu.tw/~htsu/humor/fry_egg.html


Tim
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top