Name that motherboard

  • Thread starter Thread starter Dave
  • Start date Start date
D

Dave

MUST BE:
NOT Asus/Asrock/Gigabyte
AM2+ with native support for 125W or more Phenom processors, without a BIOS
flash needed! (because I won't have another processor to fit this)
2 or more PCI slots
A PCI-Express 2.0 X16 slot that runs at X16 with no extra hardware required,
and the slot must not be in position 1 (counting from CPU side). This is
important as my video card is passively cooled and has a heatsink on the
BACK. So PCI Express X16 being the first expansion slot won't help me.
No legacy connectors (ie, parallel/Com ports)
Supports 10 or more USB 2.0 ports, with 6 or more on the I/O panel
5 phase voltage regulator
PASSIVELY cooled (no fans)
heatsinks of some type on Northbridge, Southbridge, and voltage regulators
Good overclocker!!!

Near as I can tell, the above does not exist. Somebody please prove me
wrong. -Dave


Bonus points:
Four DDR2 slots
AMD brand Northbridge
Less than $150 from the usual better-known U.S. vendors
 
Bill said:
You're right, it doesn't exist. What's the deal with the 5 phase
voltage regulator?

Well it's nice (more stable, better MTBF) if each core has its own voltage
regulator. Phenoms have up to four cores each, at the moment. But if there
are just four voltage regulators, then one of the cores has to share a
voltage regulator with the memory controller. Thus, 5 voltage regulators
are required (not just a good idea, but required). SOME mainboards have
this. Others come close to meeting all my requirements, EXCEPT that they
have 3 phase voltage regulators. I don't even want to know how 3 voltage
regulators power 5 components with any semblance of stability. (???)

Have you ever used NewEggs power search?

http://www.newegg.com/Product/PowerSearch.aspx?N=2010200022
&SubCategory=22&GASearch=3

Yup. I've found a few that are close to what I want, but learned later that
they had 3-phase voltage regulators, or legacy connectors (plus not enough
USB 2.0 ports), which is just plain stupid. -Dave
 
I don't
even want to know how 3 voltage regulators power 5 components with any
semblance of stability. (???)

Connect together the loads that need the same voltage... all of the
current required is shared by the connected loads. It's not a big
deal. Google "Series & Parallel circuits"...
 
Dave said:
Well it's nice (more stable, better MTBF) if each core has its own voltage
regulator. Phenoms have up to four cores each, at the moment. But if there
are just four voltage regulators, then one of the cores has to share a
voltage regulator with the memory controller. Thus, 5 voltage regulators
are required (not just a good idea, but required). SOME mainboards have
this. Others come close to meeting all my requirements, EXCEPT that they
have 3 phase voltage regulators. I don't even want to know how 3 voltage
regulators power 5 components with any semblance of stability. (???)



Yup. I've found a few that are close to what I want, but learned later that
they had 3-phase voltage regulators, or legacy connectors (plus not enough
USB 2.0 ports), which is just plain stupid. -Dave

It doesn't work that way.

AM2 uses a single plane for power distribution. Whether there were three
phases or four phases, they're all tied to, and pumping into, the same
single Vcore copper plane.

Phase1---+
Phase2---+ --- CPU_socket_Vcore
Phase3---+
Phase4---+

What AM2+ adds to the mix, is separate power for the memory interface.
If you see N+1 phases next to the CPU socket on an AM2+ board,
then the N phases are tied to the Vcore plane, and the 1 phase is
tied to the memory interface. So, again, there is no issue with
even or odd numbers of phases. The "N" portion can be any number
sufficient to provide the current necessary. I could use N=3
or N=5, if the total current was enough.

Phase1---+
Phase2---+ --- CPU_socket_Vcore --- Phenon_all_cores_common
Phase3---+
Phase4---+

Phase5-------- CPU_socket_memory_interface_power

The "count of phases" is not a complete characterization of a motherboard.
I can build four "big" phases, using big MOSFETs (cost a bit more each).
Or I can build eight "small" phases, where the current each offers is
less. It isn't possible to just eyeball the design, and say "oh, that
is good for 100 amps".

But it is possible to make some general observations. The Anandtech article,
where they had problems with the highest power processors stuffed into
some AMD boards, offers some insight. They noted the number of phases
the designers used, and which boards failed. You could get some idea
from that, that three phases of the type used, weren't cutting it.
And Anandtech forgot to check the CPUSupport chart for the product,
before doing the experiment. If you look carefully at a CPUSupport
chart, if only 95W processors are listed, then you'd know that 125W
or 140W isn't possible with the Vcore design.

I like this Asus page, not for the advice at the top of the page,
but for the fact that they listed motherboards at the bottom
of the page, where the board is 140W capable. Counting MOSFETs
is for the birds.

http://event.asus.com/mb/140w/

Paul
 
Back
Top