Archived from groups: alt.comp.periphs.mainboard.asus (
More info?)
On Sat, 11 Jun 2005 13:12:45 -0400, Paul wrote:
> In article <1mzjs5uhzdzve.1df87203pywpt$.dlg@40tude.net>,
> signmeuptoo@earthlink.net wrote:
>
>> I did a quick look in my manual (paper one included with my ASUS A8V
>> Deluxe, Ver. 2 Circa December, it has a BIOS "1009" on it I think, because
>> that is what it says on the EEPROM sticker) and I only saw one reference,
>> on page 4-28 where it says DDR options: "[auto] [2.6] [2.7] 2.8]". Does
>> that mean that I can't raise my memory voltage to 3 volts, or higher?
>>
>> Does that option change with newer BIOS versions?
>>
>> What about my CPU max voltage, what is it, and is that controlled by the
>> BIOS version (my understanding is that it is with some limitation).
>>
>> If I update to version 1013/1014 beta, then what will my max voltages be
>> for my system RAM and my CPU?
>>
>> Can someone help illuminate this for me?
>
> To start with, a voltage setting cannot go higher than the hardware
> design allows for. For example, the DIMM voltage on some Asus boards,
> is done with an op-amp based linear regulator, and in fact there aren't
> even three settings in the hardware (perhaps there are two). So,
> while the BIOS may offer a comforting set of values, the hardware
> doesn't necessarily have to match. In your example above, maybe both
> the 2.6 and 2.7 setting result in 2.6V output - only observation with
> any hardware voltage monitoring, is going to tell you what is happening.
> Since Vdimm is not one of the monitored voltages on your board, you
> would need a voltmeter to measure the actual voltage being used.
>
> The Vcore situation is a bit different, because a standards compliant
> switching regulator is used. If AMD defines 0.8 to 1.55V as the range
> of voltages for the K8 family, then that is the range that will be
> supported by the chip. If Asus offers voltages outside that range,
> it could be achieved by means of fooling around with an offset pin
> on the regulator chip, to adjust the voltage upward. Again, use the
> voltage monitor in the BIOS or Asus Probe, to know whether anything
> is actually happening when a higher voltage is selected. (I must
> confess that I'm puzzled how all this can work, because you would
> want programs like clockgen, CPUZ, NVtune, all sorts of Windows
> utilities to return valid voltage values, and I don't know how
> you can extend a Vcore circuit voltage range, and allow utilities
> like that to accurately reflect the setting.)
>
> So, to start with, you need to reverse engineer the chips on the
> motherboard, to understand the hardware limits. The BIOS is just
> another layer of confusion added on top, as the BIOS may or may
> not be reflecting what the hardware can do.
>
> As a general rule, don't expect a BIOS update to offer new voltages.
> I think I have seen one BIOS released by Asus, that added a voltage
> value to the BIOS interface, so it can happen, but it is not a
> common occurrence.
>
> The hardware itself has limits. 2.8V is probably considered a
> safe generic overvoltage for your average DIMM, so Asus doesn't
> consider that they will be liable if any user burns out a DIMM.
> If you had a DFI board that offered 4.0V, there is every
> possibility that either through a BIOS bug (value gets set to
> 4V by a BIOS bug), or a noob user jamming the control to max,
> will result in the killing of DIMMs. I'm sure DFI has plenty of
> disclaimers, that whatever you do, it is your fault.
>
> Asus is more conservative, and tends to pay attention to the spec
> limits for hardware. For example, the Athlon64 has a max DIMM voltage
> of 2.9V, so don't expect Asus to provide voltages above that
> value. (I don't know of a way to predict the headroom that is
> actually available on Vdimm - it really depends on I/O protection
> structures, and it can also depend on how the various power
> rings are connected together by parasitic structures inside the
> chip. When boosting a voltage to artificially high values, you could in
> fact be causing current to flow through paths in the chip,
> that is wasn't intended to go through - parasitic paths
> are ones that are created by using conditions outside the operating
> voltage specs.)
>
> Various private forums on the net, offer volt mods for Vdimm and
> Vcore. If you are determined to "go to the wall", that is one
> way to get what you want. Or, perhaps a DFI or Abit motherboard
> is the way to go.
>
> Wire tricks (placing a wire between certain pins in the socket),
> only helps if the Vcore defined for the processor has some range
> left in it. If the K8 VID table only goes to 1.55V, then a wire
> trick cannot offer more than 1.55V. An actual volt mod can do it,
> but is dangerous - this thread proves you should do mods to your
> own board, not let someone else do it:
>
>
http://www.xtremesystems.org/forums/showthread.php?t=40378&highlight=a8v
>
> The purpose of Asus boards, is to offer broad compatibility with
> stuff like DIMMs, and easy out-of-the-box auto-everything setup
> by the user. Asus boards are not the optimum solution, if you
> also want to overvolt (or undervolt) stuff.
>
> HTH,
> Paul
Paul, you read my mind, and you answered the questions in my mind perfectly
and then some.
Since this will be my first Post PIII build (if and when I get a good job
to pay for the CPU, RAM, PS, etc., it will probably be good that I cannot
go off the wall anyways. But I do hope that I get around to building
another system in the future where I can do more advanced setting work. By
then maybe I will know more about what I am doing, I hope!
In any event, it now seems to me that I will be best off with TCCD RAM
rather than BH-5. I am only worried that I won't be able to get any before
they stop selling the stuff!
OT: You sound like you either work/worked in either the semi-con or
network or electronics industries, or all of them. I really envy your
knowledge, if I can be anyone, I would like to aspire to be like you, for
you are a real asset here, please don't ever leave.
Tell me, do you participate in any of the other Usenet/forums?