Bios AI Overclock DOCP

Asus P7H55D-M EVO mobo on my little desktop at home. Built and configured
"stock" in 2010, very stable ever since.

Normally I avoid fiddling Asus bios like the black plague, but today I
was trying to figger what it was doing with XMP ...

AI overclock tuner was set to auto. I tried to set it to DOCP and the
bios froze. Arrows, esc, etc had no effect. I had to cntl-alt-del to
get it back to auto. Replicated the problem once. Boots OK on original
settings.

Manual sez "select any of these preset OC config options". I select
DOCP and it freezes.

Any clue what was wrong?

P

"Law Without Equity Is No Law At All. It Is A Form Of Jungle Rule."

Puddin' Man wrote:
Asus P7H55D-M EVO mobo on my little desktop at home. Built and configured
"stock" in 2010, very stable ever since.

Normally I avoid fiddling Asus bios like the black plague, but today I
was trying to figger what it was doing with XMP ...

AI overclock tuner was set to auto. I tried to set it to DOCP and the
bios froze. Arrows, esc, etc had no effect. I had to cntl-alt-del to
get it back to auto. Replicated the problem once. Boots OK on original
settings.

Manual sez "select any of these preset OC config options". I select
DOCP and it freezes.

Any clue what was wrong?

P

http://vip.asus.com/forum/view.aspx?...Language=en-us

"You can use them via setting your DIMMs to operate under
X.M.P. Profile (Profile set out by DIMM vendor) or
D.O.C.P. (Profile set out by Asus) mode."

So that posting suggests Asus is attempting to guess the
setting for the RAM, at an extended frequency. It
implies taking SPD info (normal speed), and extrapolating
somehow. (For example, XMP tells the BIOS what DIMM voltage
to use, while for DOCP, using just regular SPD info, there is
no voltage spec, so they'd have to use their best guess,
like maybe 1.65V.)

*******

This is a bit ugly.

http://support.asus.com/Search/KDeta...BE944ED08D&t=2

"Q: The BCLK and CPU frequencies change after I enable
XMP or DOCP mode in the BIOS. Is this normal?

A: This is normal. Some CPUs provide native support
DDR3-1066 or 1333 only. When XMP or DOCP is enabled
in BIOS, the operating memory frequency (1600 MHz)
may be over the range supported by the CPU. When this
occurs, in order to make the memory run at the set
frequency, the BIOS will overclock the BCLK by the same
ratio. The CPU multiplier may also be reduced to make
the CPU frequency closer to the original value.
"

So there, you can see if they get carried away, it might cause
a stability problem. With the BCLK limits Intel has though,
they probably can't dial BCLK all that much.

*******

I thought maybe DOCP started dynamically testing or something,
but the first article suggests it's just a one-shot guess at
settings. Typically, a motherboard doing things like that
automatically, applies a bit more voltage (Vdimm)
to try to provide margin. That's one of their tricks.

What I'd try, is AI Overclock Tuner to Manual, and tune
from there. Just set the frequency to the desired value
(what RAM is rated for), and see if that stays up long
enough to run memtest86+. (Of course, if the frequency
is not offered, the implication is that the necessary
memory divider or synthesis setting, is not present
in hardware for some reason. You'd check the manual
again, to see what canonical values are listed.)

That's what I've used on Asus boards before, is find
the "major control" like AI Overclock Tuner, set it to
Manual, in order to expose more settings. That is handy,
even if all you want to do is verify what settings the
board proposes to use for you.

On older boards, they didn't make a distinction on settings.
When an 865PE board offered two memory settings above DDR400,
there was no "DOCP" setting for those. The board just
gave the impression the controls went all the way from
DDR200 to DDR500 seamlessly. The DOCP is making the
distinction that "hey, we'd doing something for you,
that isn't in the Intel official documents". Intel in
the past, has played a little game, where all possible
divider settings are not documented. The motherboard
makers soon figure it out, and offer the things they
learn, as BIOS options.

*******

Actually, the e5325_p7h55d-m_evo_v2.pdf manual shows
a bunch of information, which hints at what they're doing.
The entries which say "Auto" here, imply the processor
has a direct setting to get to that speed. While DOCP
implies they had to do something out of the ordinary
to get to that frequency (either use undocumented settings,
or screw around with BCLK, which is cheating). The text here,
even suggests adjusting BCLK after you're finished ? They seem
to be implying that divider settings are a function of processor
family and processor speed. I can understand the family
making a difference, but I don't get how the speed factors
in. You'd think the same ratios with respect to a BCLK of
100MHz would be there.

http://img7.imageshack.us/img7/4725/...vov2page78.gif

Lynnfield Clarkdale
2.66GHz 2.8GHz 2.93Ghz 2.8GHz Others
DRAM MHz
1333 Auto Auto Auto DOCP Auto
1600 DOCP Auto Auto DOCP DOCP
1600 DOCP DOCP DOCP N/A N/A

The manual says the configuration option values (above nominal)
are DDR3-1600, 1800, 1866, 2000, 2133, 2200. And those
are achieved (somehow), with multiplier/divider
settings in the clock synthesis for the memory clock output.

In summary:

1) Using purely XMP DIMMs, and the BIOS decides XMP conditions
(2 sticks) are valid, you get a working configuration
as defined by XMP in the SPD chip on the DIMM.

2) Using non-XMP DIMMs or XMP disabled, the BIOS uses
the SPD standard table section.

"Auto" settings cover the official Intel support (DDR3-1333).

3) Same as (2), only DOCP handles DDR3-1600 to DDR3-2200.
If the board crashes when you do this, set DOCP, then
flip over to the CPU section, set CPU things to "manual"
for a look, and see if it's been messing with BCLK,
Vcore_CPU, that sort of thing. There's got to be
a simple explanation for why it freezes.

As far as I know, DOCP should not freeze the thing
immediately. The clock settings should only be
changed, when the board next POSTs (after save and
exit from BIOS). It's not kosher to crash a board
in the middle of the BIOS screen :-)

HTH,
Paul

On Sun, 09 Dec 2012 23:24:38 -0500, Paul wrote:

...
As far as I know, DOCP should not freeze the thing
immediately. The clock settings should only be
changed, when the board next POSTs (after save and
exit from BIOS).

Quite so.

It's not kosher to crash a board
in the middle of the BIOS screen :-)

Hah! That qualifies as Understatement of the Century. Eh? :-)

I'm afraid of the damned thang. I'm gonna leave it alone.

"Just Another Glorious Asus Experience!". :-)

Thanks,
P

"Law Without Equity Is No Law At All. It Is A Form Of Jungle Rule."