ASUS P5W DH - Problems + Fixes Thread

[scaramonga]

Thanks man, didn't completely get rid of the droop, but I got the load and idle to be with in .01v, which is all that matters.

What hardness of pencil did you use?......how many strokes?

[Krusty]

Hi, you've actually just run into a problem that windows xp has that's inherent to how it operates as an operating system. It was not designed to run as a symmetric multiprocessing (SMP) OS, but rather single core, single thread. As Intel released their hyperthreading technology, windows modified its code to allow for this, and tweaked it slightly to fit our dual and quad core environments of today. What you see in the OS is what you get, unfortunately.

I'm not sure the back-end logic and implementation of SMP in Vista, or if it will handle symmetrical processing better, but you have to be also using an application that is designed for it as well. All of the large, workstation class applications do have support for multiple cores, even running on XP (like Adobe Premiere, 3d Studio, etc), but you won't be getting as much benefit as you would on a similarly spec'd unix based workstation that was designed from the ground up with SMP in mind.

I know MS was going this route, as were all the game developers, so I'm sure we'll see better utilization and better scaling of games and applications upon multiple cores in the near future.

So in the meantime, if you want to only stress one core at a time, you'll have to manually set affinity in your task manager. However, from what benchmarking I've done in various applications, forcing an application to use only one of the two cores actually cuts your overall processing potential in half. This may not be true for all applications, but just from what I've seen.

to nickfd:
Thanks for the explanation. I wonder does Linux handle multi-core processor better?

[Zeus]

I just did the mod and the droop has got a lot less.

Prior to the mod, i saw CPU-Z reporting Vcore going down as low as 1.392V while i had set it to 1.4375V in bios.

This was when running Orthos small fft's.

After i did the mod, i set Vcore to 1.425V (so, a tad lower) and now the lowest CPU-Z reports is 1.408V.

So prior to the mod i needed 1.4375V in bios to get Orthos stable at 3.6GHz due to Vdroop, now i can do the same with 1.425V.

It's just not that the Vcore doesn't fluctuate anymore, it still does but it does not get as low as it used to.

I see there's some misconception about the Vdroop pencil mod, let me explane how to do it:

Locate the resistor in the picture, it's somewhere at the northside of the board.
Take a Digital multimeter and measure resistance of this resistor, mine was 50Kohms.
Just take a soft pencil and stroke the resistor a number of times, now measure again.
Repeat untill it reads 25Kohms.

Make sure there's no graphite particles elsewhere on the board before firing up.

This mod is done at your own risk.

Sit back and enjoy.

[J33pG33k]

to nickfd:
Thanks for the explanation. I wonder does Linux handle multi-core processor better?

Most Linux distros are built by default to handle SMP (up to 256 cores!). I'm on Kubuntu and its purring right along...

[scaramonga]

I see there's some misconception about the Vdroop pencil mod, let me explane how to do it:

Nice one Zeus!

Thanks for a decent explanation at last.

[roger_h99.nor]

...After i did the mod, i set Vcore to 1.425V (so, a tad lower) and now the lowest CPU-Z reports is 1.408V...

Repeat untill it reads 25Kohms.

Glad it worked! But you haven't found your mobos 'sweet-point' yet. Mine dosen't droop at all running Orthos. The only stresstest where I notice droop is TAT. Try lower it to about 20Kohm.

[Shpoon]

Blimey that seems worse than mine. I have a wk 24a ! What are your BIOS settings ?

Over the weekend I got mine 9 hours Orthos stable (CPU and Mem test, blend?) using CPU Vcore of 1.55V (actual 1.52-1.53) and FSB of 389 (3.5Ghz). Memory at 2.1V. After 9 and bit hours Windows blue screened with MACHINE_CHECK_EXCEPTION, but I never use my PC for 9 hours!

FSB: xxx
Memory Voltage: 2.15V
CPU Vcore Voltage: xxx
FSB Termination: 1.5v
MCH Chipset Voltage: 1.60
ICH Chipset Voltage: 1.2
Freqs are 33.33 and 109.

Memory using SPD
DRAM ECC Mode: Disabled
Hyper Path 3: Disabled
DRAM Throttling Threshold: Disabled

Advanced CPU settings:
Modify Ratio Support: Disabled
Microcode Updation: Enabled
Max CPUID Value Limit: Disabled
Execute Disable Function: Disabled
Enhanced C1 Control: Disabled
CPU Internal Thermal Control: Disabled
Virtualization Technology: Disabled
Intel (R) SpeedStep (tm) tech: Disabled

[AMD_Deluxe]

hi m8s !

i have a question ..

what is the real vcore when i set the vcore in bios to 1.485?
i have readed that the real vcore is at 1.485 -> 1.43V .. is that right ??
cpu-z shows me 1.213V ... seems to be a bug !!!
everest shows 1.49V .. i dont think that this is the right vcore

[triggerc]

I just did the mod and the droop has got a lot less.

Prior to the mod, i saw CPU-Z reporting Vcore going down as low as 1.392V while i had set it to 1.4375V in bios.

This was when running Orthos small fft's.

After i did the mod, i set Vcore to 1.425V (so, a tad lower) and now the lowest CPU-Z reports is 1.408V.

So prior to the mod i needed 1.4375V in bios to get Orthos stable at 3.6GHz due to Vdroop, now i can do the same with 1.425V.

It's just not that the Vcore doesn't fluctuate anymore, it still does but it does not get as low as it used to.

I see there's some misconception about the Vdroop pencil mod, let me explane how to do it:

Locate the resistor in the picture, it's somewhere at the northside of the board.
Take a Digital multimeter and measure resistance of this resistor, mine was 50Kohms.
Just take a soft pencil and stroke the resistor a number of times, now measure again.
Repeat untill it reads 25Kohms.

Make sure there's no graphite particles elsewhere on the board before firing up.

This mod is done at your own risk.

Sit back and enjoy.

I have my droop pretty well under control now, but i've no idea how to use my DMM to measure resistance. Did you put both probes on the resistor or just one?

[Zeus]

Put the probes on both sides of the resistor and select >50Kohm on your dmm.
Put the probes on the silver parts right on the edges of the resistor.

[triggerc]

and i do this with the system off but psu on right?

[Zeus]

Everything off!!

Do not poke around with the probes with the PSU turned on.

You'd best pull the powercord.

[triggerc]

alright, i just took a look and i'm getting 18 or so k ohms, maybe that's a little too low?

[Zeus]

I dunno, haven't tried that low.

Take a q-tip and remove some graphite if you want it a bit higher.

[triggerc]

22 seems to be my sweet spot. the voltage doesn't drop anymore, thanks for all the help.

[Psitech]

FSB: xxx
Memory Voltage: 2.15V
CPU Vcore Voltage: xxx
FSB Termination: 1.5v
MCH Chipset Voltage: 1.60
ICH Chipset Voltage: 1.2
Freqs are 33.33 and 109.

Memory using SPD
DRAM ECC Mode: Disabled
Hyper Path 3: Disabled
DRAM Throttling Threshold: Disabled

Advanced CPU settings:
Modify Ratio Support: Disabled
Microcode Updation: Enabled
Max CPUID Value Limit: Disabled
Execute Disable Function: Disabled
Enhanced C1 Control: Disabled
CPU Internal Thermal Control: Disabled
Virtualization Technology: Disabled
Intel (R) SpeedStep (tm) tech: Disabled

How do you disable "Virtualization Technology"? I'm using bios 1707 and it is grayed out.

I asked about this last week and didn't get any response.

[Shpoon]

How do you disable "Virtualization Technology"? I'm using bios 1707 and it is grayed out.

I asked about this last week and didn't get any response.

Mine just...isn't? And maybe you don't want it disabled look at my OC.

[Seb R]

Peeps...

Are any of you using vista? I've had the RTM installed for a while now but can't be bothered with the fact that loads of software packages I use just won't work with it - however I've decided to give it another go seen as it's just been publically released, and I have a really irritating problem that's been present since some of the early betas:
Every now and then (this could be 1 minute or 20 minutes) my mouse/audio/general experience starts to stutter (lasting for up to a couple of seconds)... are any of you using vista and having the same problem?

I've got all the latest drivers installed also...

S

[NghtShd]

Hi,

You can actually run 4 sata hdd's off your intel ICH7 southbridge, and I would highly recommend this over trying to get the JMicron and/or the SiliconImage crap(ez-backup) working.

Plug your drives into SATA 1 (red), SATA 3(actually#2 black), SATA 4 (black), remove the jumpers off the ez-backup area, and plug into the right yellow ez-raid port (actually SATA2). I don't remember the physical marking of the right ez-raid port, so please forgive me.

If you have the jumpers off and boot with the drive attached, you'll see it in the list of the ICH7 area as your box posts, if it's in the wrong one, you won't see anything. Also, you may want to set the DH Features Change option in the bios to Enabled when you take out the jumper and attach the drive to have it tell the SiliconImage crap to turn off.

Bingo. You have 4 hdds on ICH7 and can turn off JMicron in bios.

That's a great tip if it really works. One of my biggest gripes with this board's design is the use of the SiliconImage port multiplier on the ICH7R. Have you verified that it is really turned off and not running in RAID 1 (which I believe the manual says to use if you are running a single drive on an EZ port)? Does NCQ work on a drive connected this way?

[BenchZowner]

I'll have the site ready tomorrow. and will I publish the related article and Asus's Tech Support evaluation results article.
Gonna publish the article on TheInquirer ( sigh ) aswell
Hopefully that'll grab Asus's attention.
Else, I will grab the details of a high ranked Asus employee ( with the help of my friends in the industry ) and will have a chat with him.

Finally! The time has come!

http://www.benchzone.com/page.php?al=asustek_evaluation

[ awaiting for my friend Ian @ TheInquirer now ]

[NghtShd]

Is anyone using a FireWire hard drive on this board's FireWire controller? I have and admittedly cheap FireWire/USB to IDE enclosure. I was getting delayed write failures and other error messages and lost a lot of data. Switching to a USB 2.0 connection for the enclosure resulted in better performance and no more errors. Now I've installed a PCI FireWire card and went back to using the drive enclosure's FireWire connection with none of the previous problems.

I wonder if my MB's firewire controller is bad or if it could simply be the FireWire to IDE chipset and my motherboard don't get along.

[iadstudio]

I use a firewire backup drive and it works fine. Nice transfer rates too. I actually ran out of USB. I only have SATA drives in my machine though. Did you say you have IDE? Shouldn't matter which type is in the enclosure though.

[grudgelord]

Is anyone using a FireWire hard drive on this board's FireWire controller? I have and admittedly cheap FireWire/USB to IDE enclosure. I was getting delayed write failures and other error messages and lost a lot of data. Switching to a USB 2.0 connection for the enclosure resulted in better performance and no more errors. Now I've installed a PCI FireWire card and went back to using the drive enclosure's FireWire connection with none of the previous problems.

I wonder if my MB's firewire controller is bad or if it could simply be the FireWire to IDE chipset and my motherboard don't get along.

I've seen delayed write failures occur frequently with cheap USB enclosures. In most cases, if you change only one component (either the computer/motherboard, the usb/fw enclosure, or the drive that lives in the enclosure) the problem ceases.

Case in point: I have a cheap USB enclosure in which I installed an old 200GB Maxtor drive. When I tried to use it with my laptop I'd get DW failures. I replaced the drive with another Maxtor 200GB (identical model) and it functioned fine. Fearing that the first drive must be failing, I installed it into a different USB enclosure and tried it with the laptop... it worked fine.

During my troubleshooting of the issue I discovered that compatibility issues between drive, controller and interface can occur, especially with cheaper chips.

I know that you are using FW instead of USB but I believe the issue is the same as it comes back to the IDE controller chip. So, no, I don't think it's your firewire port.

[Phoenix_TT]

FSB: xxx
Memory Voltage: 2.15V
CPU Vcore Voltage: xxx
FSB Termination: 1.5v
MCH Chipset Voltage: 1.60
ICH Chipset Voltage: 1.2
Freqs are 33.33 and 109.

Memory using SPD
DRAM ECC Mode: Disabled
Hyper Path 3: Disabled
DRAM Throttling Threshold: Disabled

Advanced CPU settings:
Modify Ratio Support: Disabled
Microcode Updation: Enabled
Max CPUID Value Limit: Disabled
Execute Disable Function: Disabled
Enhanced C1 Control: Disabled
CPU Internal Thermal Control: Disabled
Virtualization Technology: Disabled
Intel (R) SpeedStep (tm) tech: Disabled

Perhaps increase the MCH v to 1.65. Also what memory do you have ? What's your cooling like and do you have an aftermarket HSF ? You may need to pump more CPU vcore, if you are prepared to do so, but make sure you have some decent cooling.

What are you using to test stability ? Also you say "crashes doing the weirdest things" can you expand ?

[OneyedK]

Finally! The time has come!

http://www.benchzone.com/page.php?al=asustek_evaluation

[ awaiting for my friend Ian @ TheInquirer now ]

Great article, hope the bots at ASUStek will read it