Re: A8N-VM CSM Memory speed

On Fri, 20 Jan 2006 21:12:06 -0500, Squizzy Taylor
<Fri_Manday@xxxxxxxxx> wrote:

>You can defend AMD/ASUS or whomever you like but this board has been a
>disapponitment in so many ways and it is nice to have a place to come
>to to say the board sucks and ASUS took my money and AMD helped them
>and I don't appreciate it.
>As stated
>Live & Learn

You don't list a single one of the ways in which you've been
disappointed.
If it's the board's inability to run four DIMMs at 400 MHz, what board
would choose which can do that?
What feature(s) do you find to be different from what Asus and AMD
claimed in their manuals?

>On Fri, 20 Jan 2006 05:14:50 GMT, "Simian Dyson"
><[nospam]jhansman@xxxxxxxxxxx> wrote:
>
>>
>>"Paul" <nospam@xxxxxxxxxx> wrote in message
>>news:nospam-1901062259210001@xxxxxxxxxxxxxxxx
>>> In article <sqXzf.13254$PL5.5795@xxxxxxxxxxxxxxxxxxxxxxxxxx>, "Simian
>>> Dyson" <[nospam]jhansman@xxxxxxxxxxx> wrote:
>>>
>>>> OK, so I poplulate all 4 memory slots on this board with PC3200 DIMMS
>>>> (Patriot Signature 4x512mb), and when I boot the POST screen show the
>>>> memory
>>>> running not at 400 (and yes, the memspeed setting in BIOS is 200), but at
>>>> 333. So I call Asus, and what does the 'tech' tell me is th reason for
>>>> this?
>>>> Wait for it...........................
>>>>
>>>> All 4 slots are populated!! And, if I remove two modules, it *should* run
>>>> at
>>>> 400. Are they fucking serious? Anyone else seen this behavior in this or
>>>> any
>>>> other NF4 board?
>>>
>>> The reason they are "****ing serious" is because of a certain AMD
>>> document.
>>>
>>> http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/26094.PDF
>>>
>>> If you look at Table 46 "Unbuffered DIMM Support For Revision E
>>> 939-pin Lidded Micro PGA Package", there is a table of DIMM
>>> configurations versus memory settings. You have to be a little
>>> careful with this table. Column 1 and Column 3 are on one memory
>>> channel, while Column 2 and Column 4 are on the other memory
>>> channel. Basically, when a channel has two double sided DIMMs
>>> on it, AMD recommends operation at DDR333 command rate 2T.
>>>
>>> I believe under these circumstances, Asus uses DDR333,
>>> but I'm not sure if they use command rate 1T or 2T. As the end
>>> user, your realistic options are DDR333 1T or DDR400 2T, with
>>> something like 4x512MB. If the BIOS won't let you do this at
>>> first, start it out with two DIMMs, set the BIOS to DDR333 2T,
>>> add the second two DIMMs, and try for DDR400 2T. (DDR400 2T
>>> is the better of the two options.)
>>>
>>> Once you have the setting in there, and the computer will POST
>>> without a problem, run memtest86+ to be sure that the memory
>>> controller on the processor is happy. Make sure that Vdimm is
>>> boosted a bit (2.7V), to help the memory controller do its
>>> job. 2.9V is the spec *** limit for the processor, so I
>>> would not try to go further than 2.9V.
>>>
>>> So, what is really happening, is Asus is selecting a setting
>>> that they hope will be guaranteed to POST for you, leaving it
>>> to you to make final adjustments to a more aggressive setting,
>>> followed by testing with memtest86+ and Prime95 in Windows.
>>> Think how you'd feel if the motherboard didn't POST at all
>>> when you plugged in your four DIMMs.
>>>
>>> Asus does not make the memory controller. The memory controller
>>> is inside the AMD processor. About the only thing Asus can
>>> change, is the track impedance used to route the memory
>>> traces. The rest of the performance is controlled by the
>>> BIOS, and it is really hard to say what parts of the BIOS
>>> that Asus has the source for, and is making changes to
>>> the BIOS. The core of the BIOS comes from Award or AMI,
>>> and I don't know if the memory controller code is something
>>> they give to Asus or not.
>>>
>>> HTH,
>>> Paul
>>
>>Well, alrighty then. I did know the memory controller was in the CPU; I also
>>realize this motherboard is set to run all stock settings, so neither the
>>memory or CPU can be goosed. Still, what's the point of advertising it as
>>PC3200 compatible when you can't even get that to run at spec speed? Sigh.
>>Live and learn.
>>

Ron
.