|
Burningissues has received an unexpected amount of information & comment following our recent report on the MSI 694D-Pro A dual
motherboard.
Surprisingly, none of this feedback is from MSI: MSI's indifference to widespread problems with one of their products is a real danger-signal to anyone contemplating purchase of this
motherboard, or to any owner facing problems. If MSI can't be bothered to address the serious concerns we have raised then published - after giving them a chance to comment - what chance of support do you imagine the average consumer has?
A constant theme running through serious discussion of this motherboard is the disturbing contrast between published reviews based on MSI-supplied "freebies" & the experiences of real-world
customers [burningissues purchased their board anonymously through normal retail channels].
One private communication from a reliable source in the possession of a pair of unlocked FC-PGA 933's [each tested capable of 1.1GHz stable @1.80 vcore] states they will not work stably on this motherboard at
1GHz; this source also - as with many others - confirms the vcore instability & the need to constantly jumper-clear the CMOS.
Tweaks: 1 - fansink on the Northbridge:
Anyone concerned at the real or potential effects on stability of the very high temperatures on the VIA133a chipset Northbridge - up into the 50's °C running dual [120's -130's °F], which is a) too hot to
touch, & b) around the level even the most gung-ho AMD overclockers see as a sane limit for their CPU's - can easily stick on [literally] an old 486 fansink or a good quality purpose-built chipset-cooler.
|
|
|
The upper readout from both these dingy images show temperatures taken from the central mass of the old 486 fansink used - we chose this one because its strange form allowed easy attachment of the 4 radial arms down onto
the motherboard with blobs of silicone sealant [the assembly left to set overnight under a weight] - the base of the fansink was lapped dead flat & heatsink grease applied between it & the Northbridge. The temperature with no heatsink was 39°C [102°F] at 19.5°C ambient with the same
setup of only one PPGA CPU at 100FSB. In both shots the lower temp is ambient. As you see, a heatsink alone has limited effect - & the one used here has a great deal more mass & surface-area than the typical "greenie" manufacturers apply to Northbridges.
|
Tweaks: 2 - hardwired vcore control:
An infuriating & dangerous [to your CPU's] problem occurring again & again to users of standard retail 694D's is the strange way the vcore of each CPU is controlled through the soft-BIOS - or not
controlled, which is the problem.
The analysis & answer [scary, unless you are unusually competent with a soldering-iron] published below is the work of a correspondent from Japan: this modification has been carried out by him &
others, & apparently works - unlike the MSI solution - but is obviously a very tricky process & only to be carried out by those confident of their ability so to do & fully accepting all responsibility for & consequences of their actions. The
vcore-control problem with the 694D falls into three parts, all of may be cured by the modification below - since it takes all vcore settings away from the soft-BIOS:
1: [improved, not cured, by later BIOS'] You can't set the vcores for CPU's #0 & #1 to be closely similar or the same - this is
obviously important for stability.
2: [improved, not cured, in later BIOS'] Adjusting the vcore upwards for one CPU - typically #0 - will affect the vcore [upwards] of
CPU#1 in a non-linear way. For instance, a gentle nudge upwards of , say, 0.05 vcore to CPU#0 may boost #1's vcore by 0.1v or 0.2v - taking it to a level at which that CPU may cease to run stably, or overheat the fansink used, or enter the voltage danger-zone for that model of CPU at which its
working life is shortened.
3: [basic & unimproved - possibly not able to be improved - by any BIOS to date] On a warm reboot the vcore of either or both
CPU's may momentarily leap up as far as 2.05v before the soft-BIOS resets it to the selected vcore - this appears to be a stability-enhancing strategy used by MSI for overclockers or those using different stepping/multiplier CPU's - & in numerous reported instances may "stick" at or
near this "temporary" level, or cause one CPU to be "lost" - perhaps due to a feedback confusion. This first is obviously a very serious flaw indeed - the absolute maximum you can run a FC-PGA CPU at for any length of time is around 2.0vcore, & prolonged running at this
level is emphatically discouraged by Intel: the second effect - the "lost CPU" - is common when overclocking.
Our correspondent's modification is as below - a 4-way switchblock is needed for each E5HNK F244/74F244 :
|
|
pin11
|
pin13
|
pin17
|
pin15
|
|
|
Vid0
|
Vid1
|
Vid2
|
Vid3
|
Vcore
|
|
On
|
On
|
On
|
Off
|
1.65
|
|
Off
|
Off
|
Off
|
On
|
1.70
|
|
On
|
Off
|
Off
|
On
|
1.75
|
|
Off
|
On
|
Off
|
On
|
1.80
|
|
On
|
On
|
Off
|
On
|
1.85
|
|
Off
|
Off
|
On
|
On
|
1.90
|
|
On
|
Off
|
On
|
On
|
1.95
|
|
Off
|
On
|
On
|
On
|
2.00
|
|
Tweaks: 3 - safe BIOS-flashing:
With a motherboard of dubious basic stability, requiring frequent firmware updates in order to see if any of its problems have yet been addressed by MSI, you want to use the safest possible technique when
flashing the BIOS. The [sensible] technique packaged with v1.50 BIOS - only available, for no good reason we can see, from the Korean importers of MSI products - is to prepare a DOS boot-floppy with the minimum system-files required,
then to copy across three files: the correct Awardflash executable, the flash-file, & a "biosup.bat" [supplied with this version] .bat file to give the correct instructions to the Awardflash utility.
Flash-files for this board are named: 6321vms.xxx - where xxx=the version number [ie v1.50=6321vms.150; 1.62beta=6321vms.162].
The "biosup.bat" file - we name them: "up150.bat", "up162b.bat" etc may be prepared by opening a new text-file; then copying to it the line:
AWDFL770 W6321VMS.xxx /SN /PY
- where xxx=the BIOS version you wish to use. Rename the file "upxxx.bat" - copy it to the floppy you've prepared, then, after booting to this DOS-floppy, type at the A:> prompt [replacing xxx
with the correct number]:
upxxx
- the flash-process will then occur without further input from you. Do not on any account touch anything while the BIOS is being flashed - you can lose the machine permanently if the process is interrupted.
Given the advisability of the above process, we are amazed MSI are adapting their terrifying "Live-BIOS" utility for this motherboard. This utility flashes the BIOS whilst in Windows [including,
gawd 'elp us, Windows9x]. Don't use this silly thing - certainly not in W9x should you have a boot-partition with this OS: physically replacing a BIOS after use of this utility on this motherboard will be a time-consuming hassle & doubtless a nice little earner for providers of said BIOS
< 694D-ProA report
> 694D-ProA Tweaks 2
>> 694D-ProA Tweaks 3
|
