ASUS P5W DH - Problems + Fixes Thread

[w2richwood]

ok lawry i got your bios now i need a g6600 to try it out with be a week or so will post back thanks
Rich

[ajm786]

The Paperclip solution

I first came up with this solution back in the days of the Abit IC7-G motherboard. My thread is well documented in the Abit forums. Exact situation happened to me, and the paperclip solution worked perfectly.

[P5WDHdeluxe]

When trying to o/c my E6600 (steping B2) to 3GHZ my system looks pretty unstable... I guess that the problem is on FSB, ICH & MCH voltage where i leave the auto option cause i couldn't find proper values on o/c situations....
So there which are the correct values during a 3GHZ o/c?

[Skoogra]

I just got my Q6600 installed with Tuniq Tower. B3 Stepping
I am currently at 3Ghz 333MHz*9Multiplier and it seems stable with Prime95 running for almost 4 Hours now.
Core temp 0.95 says 52* 51* 58* 54* one second, then 61* 57* 49* 49* C
What is the highest temperature the cores can reach safely? And what is a good way to stress all 4 cores 100%? I am on 2004 bios, should I upgrade?

Side Note: should I use the 2004 G0 edition even though I have a B3?

[lawrywild]

If those temps are under load, they are fine. Mine can reach up to 75'C under Prime SmallFFTs. (1.2625v)

Your choice, I would stick with 2004 official if I was you.

[Skoogra]

Ok, I tried going like 350 fsb at 9x multiplier. But windows would either freeze or not go past the login step. I changed to 333 fsb and up'd the voltage of just the cpu to 1.40v instead of 1.30v and it seems to be stable at 3Ghz. It would be sweet to go 3.2GHz or something if I can keep the temps low and if it were perhaps stable. Kind of new to the whole extreme over clocking. Some day I will be as good as you guys though I hope!
I am using the small ffts and it has been going for like 5 or 6 hours now with no problems, but I have no idea why the temps fluctuate so rapidly.

[Skoogra]

I got up to 98.2% CPU usage using 3xOrthos running small fft's, 1 super-pi 1.5 running 8M figures, 1x Prime95 running small fft's.
The thing that is weird is temps are 68*C, 66*C, 61*C, 60*C and cores 0 and 1 seem to be straight 100% while cores 2 and 3 seem to jump a little every millisecond(from 90-99%?). Is this normal?
I doubt I will ever actually use all 4 cores and get 100% out of them if it takes this much just to get to 98.2%...

[asapreta]

My cousin has this board and he is using the 2301 Asus Bios.
His overclock is a bit unstable on the E6750 Intel processor.

I know that within a certain bios version it is not possible to downgrade it.
But I also remember sending a friend a link about how to do it and he made it.

I dunno if the right command is:

AFUDOS /i2004.rom /o2301.rom /n /pbnc

Is it?? Or Should I do it enabling downgrading on Asus Update Utility?

Thanks in advance.

[michaellundmark]

I got up to 98.2% CPU usage using 3xOrthos running small fft's, 1 super-pi 1.5 running 8M figures, 1x Prime95 running small fft's.
The thing that is weird is temps are 68*C, 66*C, 61*C, 60*C and cores 0 and 1 seem to be straight 100% while cores 2 and 3 seem to jump a little every millisecond(from 90-99%?). Is this normal?
I doubt I will ever actually use all 4 cores and get 100% out of them if it takes this much just to get to 98.2%...

Wprime?

[Skoogra]

Ok, so I downloaded a Prime95 that has multiple apps running, one for each of the four cores. It is now hitting 100% cpu usage and the temps are still under 70*C. This is with small fft's though. (No errors)
When I go to blend where the ram is tested, I get errors after 1-3 minutes per core. Do I have to alter the memory voltage or timings or is the memory bad?

[Grendel66]

BIOS 2302 now official -- ftp://ftp.asus.com.tw/pub/ASUS/mb/so...eluxe-2302.zip

Edit: According to fc /b it's exactly the same as the 2302 beta that was available earlier here.

[lawrywild]

No need to upgrade for me, my 2004 mod works for me like a charm.

[Skoogra]

What is the advantages/disadvantages of 2302 vs 2004? It must be something in the positive direction since they wouldn't work at a zero effort..

[w2richwood]

well im not using asus new. bios 2004 is fine came across a e6700 es so i dont need to upgrade and if i did it would be lawrywilds moded bios
Rich

[ajm786]

I'm having a major problem. I just bought another 2GB of Corsair TWIN2X2048-6400 (2 x 1GB). My computer does not want to run AT ALL with 4x1GB. It crashes all over the place, even when running the memory at default speed. If I take out 2GB, the other 2GB works fine just as it always has.

What can I do to run all 4GB at once?

[crishan]

I'm having a major problem. I just bought another 2GB of Corsair TWIN2X2048-6400 (2 x 1GB). My computer does not want to run AT ALL with 4x1GB. It crashes all over the place, even when running the memory at default speed. If I take out 2GB, the other 2GB works fine just as it always has.

What can I do to run all 4GB at once?

Try to increase your memory voltage to 2.2 Volts, also increase MCH Chipset Voltage to 1.65 Volts (while only having 2 of the memory sticks installed). Then switch off your computer, put in the remaining 2 memory sticks and switch on your machine. Also try a memory voltage of 2.25 Volts...


On another note:
The 2302 BIOS is still crap. They changed nothing from the beta versions;
The Realtek HD Audio Controller still stays enabled, as does the Serial Port Address (on 3F8/IRQ4). Intel Speedstep remains Disabled

The real kicker comes when in the BIOS you try

TOOLS
-> ASUS Music Alarm
    -> ASUS Music Alarm Option
        -> Detect CD:

Pressing ENTER here yields the result of a BIOS crash within the BIOS,
while the BIOS' language gets switched to Chinese BIG5

Bravo, ASUS, really, Bravo
Maybe they're screwing up this bad intentionally, or they have monkeys code
their BIOSes now.

It's back to 2103 for me, which is the last BIOS without all the issues.

[ajm786]

Try to increase your memory voltage to 2.2 Volts, also increase MCH Chipset Voltage to 1.65 Volts (while only having 2 of the memory sticks installed). Then switch off your computer, put in the remaining 2 memory sticks and switch on your machine. Also try a memory voltage of 2.25 Volts...

I tried EVERYTHING, including that. Nothing works at all. Crashes galore.

I did some further research and found out that it's hard to run 4x1GB with any OC at all. The only way I could get it to run stable was by lowering the speed to DDR600, which puts me under 100Mhz from the FSB of the CPU, and, thus making my memory a bottleneck.

I went out and bought some 2x2GB sticks. This will solve the problems. I'll report back.

[MacClipper]

fyi, my mobo needed 1.75v for vMCH to run 4 x 1GB stable at 400FSB when I was running my dualcores (E6750@3.2, E6300@2.8 both Auto Vcore) while vDIMM then was just 2.00v for 4-4-4-8 on my Crucials. Now Q6600@333FSB, all volts auto except Vdimm at 2.05v 3-3-3-8 Crucials@667.

HTH.

[shaithis]

I would love to know why adding a quad core severely limits the FSB on this motherboard.

Haven't heard of any other mobos doing the same =S

[lawrywild]

All motherboards clock lower on the FSB using a Quad vs Dual.

[BulldogPO]

Yes, All.

[crishan]

I would love to know why adding a quad core severely limits the FSB on this motherboard.

Haven't heard of any other mobos doing the same =S

I have found an answer to that question at The Tech Repository:

http://www.thetechrepository.com/showthread.php?t=87

specifically here:

http://www.thetechrepository.com/sho...91&postcount=2


where freecableguy talks about Adjusting [Advanced] Gunning Transceiver Logic (A/GTL+) Voltage Levels
for Increased Front Side Bus (FSB) Signaling Margins and Overclocking.

To quote him:

We can confidently say to you that Asus designs are so similar to Intel designs in this area of the motherboard that there is no doubt in our minds that Intel must supply them with some sort of "reference design." We have shown below modifications (Figure 1) for the Asus P5W DH Deluxe motherboard, a 975X chipset solution. We've also shown a picture that details the trace routes on the motherboard for one of the VTT to GTL reference voltage dividers (Figure 2) so that those that want to look for this on their motheboard at least know where to start looking.

Interesting to note, we cannot locate more than two (2) GTL reference voltage dividers on this motherboard, which could perhaps explain why this motherboard is rather poor at overclocking a quad-core (QC) CPU. This suggests that Intel's QC processors will supply Core 0/1 and Core 2/3 GTL reference voltages (Address) with their respective Data values if there is no external bias applied to pins F2 and G10. Please let us know if you can manage to locate the other two...

[Grendel66]

The 2302 BIOS is still crap. They changed nothing from the beta versions;
The Realtek HD Audio Controller still stays enabled, as does the Serial Port Address (on 3F8/IRQ4). Intel Speedstep remains Disabled

Very odd, I never had those problems. 1st gen board, went through all BIOSes that came out, always used EZ-FLASH followed by a Load Defaults. Afterwards I just disable the audio crap and enable EIST. Never tried to disable the SIO ports tho..

[crishan]

Very odd, I never had those problems. 1st gen board, went through all BIOSes that came out, always used EZ-FLASH followed by a Load Defaults. Afterwards I just disable the audio crap and enable EIST. Never tried to disable the SIO ports tho..

Apparently, this behaviour also differs between those who had the Realtek audio enabled before flashing to the 2205/2206/2301/2302 BIOS, and those who did not.

Also, I of course always clear the cmos and load setup defaults after upgrading the BIOS etc. Still, no-go with 2302.
I don't understand why noone at ASUS has stumbled upon this?

[biri36]

Hi

Can i disable Enhanced C1 Control and Intel SpeedStep with P5W-DH-ASUS-Deluxe-2004-G0-45nm-Edition ?When no wich Bios can i us?



thx