ASUS P5W DH - Problems + Fixes Thread

[G40RoCkEt]

vcore settings work down to 1.225v?

Yes, at the first time that you boot the bios (after a good clear cmos) , but in the second boot bios change the value to 1.400v.

The P5W DH revision 1.02g not fix vcore lower than 1.400v

But my 1.400vcore in windows is variable between 1.36-1.40

And with my E8600 i can get 3.850Ghz with all voltages in auto and vcore too.

To get 4Ghz and more vcore is necessary set to 1.400v and more.

The important thing is that the bios 2801 accept all CPU revision E0.

[Daimao]

Just noticed something, with my board not matter how i try it I can´t get past 890 MHz stable. To get there I had to put 1.34V Vcore when I try 400 MHz it posts but but it ain't stable.
Tried puttin' Vcore to 1.40 and even 1.50 but no dice. The only conclusion I can take is that the P5W doesn't like 400MHz FSB on Quads.

[Daimao]

Just a quick question what happens when I change the PCI Express Frequency? Does it make any interference OC wise?

[Falkentyne]

Just noticed something, with my board not matter how i try it I can´t get past 890 MHz stable. To get there I had to put 1.34V Vcore when I try 400 MHz it posts but but it ain't stable.
Tried puttin' Vcore to 1.40 and even 1.50 but no dice. The only conclusion I can take is that the P5W doesn't like 400MHz FSB on Quads.

Daimao:
You're using a 1.04g board right? (i assume you are since you can go down to 1.34v on your Q9450).

The highest stable FSB i can get on mine with a QX9650 is 380 fsb and that requires 1.85v mch or 1.75v mch+1.30 FSB (someone posted that FSB voltage affects mch voltage but i can't find the post right now, and search isn't helping). 390 FSB gives errors in memtest at 1.85v mch and 1.40 FSB so its unusuable, and 400 FSB won't post on mine. I'm sure it would post fine with a dual though. Note that with a divider my RAM can handle 450 mhz @ 2.1v, cas 4-4-4-12, but once again, to do that, I need 1.85v on the mch (or 1.75+1.30 vfsb); its the 2x2GB Gskill pc-7200.

The reason the P5W doesnt' do well on quads is not just because of lacking TRD and TRFC (Tomcug where are you??), but also there seems to be only one GTL/REF divider on the board, so the CPU has to handle the other one internally, which limits highest FSB. I just use an unlocked multi (333x12) for 4 ghz on mine (1.3625v).

PCI-E frequency MIGHT improve HDD loading benchmarks SLIGHTLY (I wonder if someone has done a HDtach or that other bench tool with different PCIE speeds), barely improves 3dmark, but some people say that it helps you overclock the core or memory on some video cards a few more mhz. Don't go above 110 if your using RAID (105 is safe) or 115 mhz for normal SATA.

[Daimao]

Thanks Falkentyne, but ain't 1.85V MCH a bit high. I normally use 1.50 and my temps are 20 (cpu) and 40 (board).

[mikepaul]

Yes, at the first time that you boot the bios (after a good clear cmos) , but in the second boot bios change the value to 1.400v.

The P5W DH revision 1.02g not fix vcore lower than 1.400v

But my 1.400vcore in windows is variable between 1.36-1.40

OK, but even 1.4Vcore didn't stabilize my Q9450 at 333 FSB. I did move up to 295 successfully at AUTO.

Any advice on the other voltage settings for a 1.02G revision using a 45nm CPU?...

[Falkentyne]

I'm really not sure what to do about the 1.02 version.
Perhaps try to pick up a cheap 1.04 or grab a P45 board? Did you try to call Asus and demand to speak to level 3 support about the 1.02g boards and the vcore bug with those 45nm cpu's? (not the grunts at the basic level?). A 1.04 should get you 360-380fsb, and 360fsb should be doable at 1.55 mch.

BTW i got mine to post at 400 FSB with the QX9650, with 2x1GB gskills, 5-5-5-15, and 2.2v but didn't even try to enter memtest... 1.85vmch+1.30 vfsb gave a "slow" post then cpu ucode error (i was surprised it let me enter BIOS without everything crashing or corrupting the chip), 1.40 vfsb was normal speed POST without that error, but was too unstable to even enter memtest86+

Even with 6 GB RAM installed in all 4 slots, 380 FSB+1.85 vmch was stable except some weird things happening when I had the 6 GB of RAM in there (2x2 GB and 2x1GB filled), PCIE link was being set to 8x; set vfsb to 1.30 and it was 16x again; power off, set it back to 1.20 and 16x again O_o....guess its something to do with that much FSB with a quad and all 4 dimms filled... fully stable tho. But since I don't have windows 7 yet (i was too slow to register) and am on XP/32, I pulled the 2x2GB dimms (4-4-4-12) so I could run the 2x1gb gskill d9's at 3-3-3-8.

And yes 1.85vmch is high; even with fan mounted on NB I don't know if that will cause eventual damage or degrade, so I don't keep it at those settings; I usually use 333x12, 1.3625v (1.375v for linpack) if I need 4 ghz for a game or quick stress test. Board is pencil vdroop modded.

[crishan]

I mailed ASUS' German tech support, and guess what:

There'll NEVER be support for Core2Duos with the E0 stepping. Never.

Off to the Gigabyte GA-EP45-UD3P then.

[mikepaul]

I'm really not sure what to do about the 1.02 version.
Perhaps try to pick up a cheap 1.04 or grab a P45 board? Did you try to call Asus and demand to speak to level 3 support about the 1.02g boards and the vcore bug with those 45nm cpu's? (not the grunts at the basic level?).

Nope, didn't escalate. I noticed over the weekend that the email I sent with my specs was in November. I guess I'll have to ask about that.

And I don't WANT to spend more money on a board for this CPU. That's for the i7 I want to get next year. This year's Top Of The Line will be cheaper then...

[Falkentyne]

I mailed ASUS' German tech support, and guess what:

There'll NEVER be support for Core2Duos with the E0 stepping. Never.

Off to the Gigabyte GA-EP45-UD3P then.

Don't email them.
All you're doing is reaching grunts.

You need to reach their taiwan office--on the phone. That's the ONLY way.
And you will have to *fight* to pass your way through the grunts of level 1 support until you can reach someone at level 3 / bios area. Once you do that, then you will be in business and be talking to someone who actually knows something. This won't come to you free--you have to fight-hard- for it.

If you're able to do this with the taiwan office, also inquire about "TRFC" and "TRD" memory settings in the BIOS (remind them that the P5B board has this, and basically ALL of their boards they made except the P5W deluxe has it and ask them WHY they didn't include this--which is EASY to add, for the P5W). Also for the benefit of mikepaul, inform the taiwan office, that there is a bug with the 1.02g version boards and voltage selection on 45nm cpu's; (1.04g boards are fine); you can't select lower than 1.4 or 1.45v core with 1.02g boards with 45nm).

[tomcug]

Crishan, E0 stepping is fully supported with latest 2801 BIOS. You can find this info even at ASUS site. I bet that all those grunts' job is to tell people to buy better motherboard. I already experienced that when I asked them about timings support. I'm still fighting with them but I don't think I'll win . It's really hard to reach someone at higher level who knows something .

[Falkentyne]

Tom, you called their Taiwan office, right?

Can you also try emailing one of the BIOS developers? I remember that someone in another thread posted the email for one of them, because they were trying to get a serious problem fixed and it took a _lot_ of phone calls and emails, but eventually a beta bios came out...

It was either in the Rampage 2 extreme thread (the core i7 board) or the P5Q Deluxe discussion thread (for pre-1406 BIOS); I think the latter, but you can check both; I know there was contact with bios engineers in both threads though...

I'm 100% sure the email was posted; see if that helps...

BTW we also need a new BIOS (not just to fix mikepaul's problem, but also for new cpu's) for those 1066 FSB wolfdales that just came out recently....

[tomcug]

Tom, you called their Taiwan office, right?

Can you also try emailing one of the BIOS developers? I remember that someone in another thread posted the email for one of them, because they were trying to get a serious problem fixed and it took a _lot_ of phone calls and emails, but eventually a beta bios came out...

It was either in the Rampage 2 extreme thread (the core i7 board) or the P5Q Deluxe discussion thread (for pre-1406 BIOS); I think the latter, but you can check both; I know there was contact with bios engineers in both threads though...

I'm 100% sure the email was posted; see if that helps...

BTW we also need a new BIOS (not just to fix mikepaul's problem, but also for new cpu's) for those 1066 FSB wolfdales that just came out recently....

Negative, I've only called Polish office where I live, but I can try to call to Tainwanese office. Just one question, do they speak English? About writing to BIOS developers, I can do that if you give me their e-mails. This shouldn't be a trouble, I'll explain all problems and write why they should fix that. Now I'm waiting for you answer.

[crishan]

Crishan, E0 stepping is fully supported with latest 2801 BIOS. You can find this info even at ASUS site. I bet that all those grunts' job is to tell people to buy better motherboard. I already experienced that when I asked them about timings support. I'm still fighting with them but I don't think I'll win . It's really hard to reach someone at higher level who knows something .

Um NO, it ISN'T.

Check for yourself by going here:
http://support.asus.com/cpusupport/c...Language=en-us

Choose Motherboard, socket type (775) and name (P5WDH);
The list you'll be getting only incorporates these cpus:

Core 2 Duo E8200 (2.66GHz,1333FSB,L2:6MB,rev.C0) ALL 2504

Core 2 Duo E8300 (2.83GHz,1333FSB,L2:6MB,rev.C0) ALL 2704

Core 2 Duo E8400 (3.00GHz,1333FSB,L2:6MB,rev.C0) ALL 2504

Core 2 Duo E8500 (3.16GHz,1333FSB,L2:6MB,rev.C0) ALL 2504

There is no, repeat NO, REPEAT, NO SUPPORT FOR E0-STEPPING CPUs
(sorry for shouting, but you are just making assumptions. ASUS clearly
don't support the E0-stepping, they say so on their own CPU support page).

And yes, for some time after the 2801 Bios was released, they had the
E0-Stepping CPUs in the list. However, they took them out again, for
whatever reason.

[tomcug]

Crishan, stop writing such bull*****! I know that they removed E0 revision CPUs from their site but 2801 BIOS fully supports them! Many people have E8600 with P5W DH Deluxe and they don't have any troubles! Check this out man:

Will support: E8400 revE0, E8500 revE0, E8600, Q9400 revR0, Q9550 revE0, Q9650 revE0

ftp://ftp.asus.com.tw/pub/asus/mb/so...eluxe-2801.zip

[Falkentyne]

Negative, I've only called Polish office where I live, but I can try to call to Tainwanese office. Just one question, do they speak English? About writing to BIOS developers, I can do that if you give me their e-mails. This shouldn't be a trouble, I'll explain all problems and write why they should fix that. Now I'm waiting for you answer.

Yes most of them do speak some English. Patience is needed, though. They may answer the phone in chinese of course.

Not sure where their taiwan number is; pretty sure you can find it. The emails may be hard to find; I remember at least 1 email was in either the R2E thread or P5Q Deluxe discussion.

Crishan : what motherboard revision do you have?

And on a lighter note,
I think I can do 4.1 ghz with my QX9650 relatively prime stable with 1.4v (didn't try, no point in risking degradation), as 1.3875v at 4.1ghz passed 8k FFT but when i went straight to 40k, in-place, core 4 failed shortly (seems like 8k, 10k and 40k are most stressful). Linpack is 50 loops stable at 1.375v and 4 ghz (for some odd reason, with 2x1GB, Linpack needed 1.3875v for 50 loops, but for 2x2GB, 1.375v? temps only 2C higher at 1.3875v), and fully prime stable at 1.3625v. The cores were screaming for mercy at 4.2 ghz and 1.4v in Prime (cores 1, 2 and 4 didn't last more than 5-20 seconds without one of them erroring out). I can probably game safely at 1.425v but that's already too high to risk degradation (even tho the TRUE keeps temps down), and I'm sure it won't prime very long. Seems like the most realistic MAX goal is "possibly" prime stable at 4.1 ghz/1.4v (1.392v load-which should be 100% stable in games) or possibly game stable 4.1 ghz 1.3875v (1.384v load), and a safe goal is 4 ghz 1.35v for normal stability and 4 ghz/1.3625v for max stability in everything except Linpack.

enough numbers I guess...things start getting hard after 4.0 ghz 100 mhz to 4.1g doesn't seem like its worth .0375v though

[Ragdd]

Crishan: I have 2 Asus P5W DH Deluxe motherboards with E0 C2D cpu's.
The only thing you must have is the latest BIOS 2801
My Pc : Asus P5W DH Deluxe revivion 1.04g with Intel C2D E8600 E0 stepping
Second pc: Asus P5W DH Deluxe revision 1.02g with Intel C2D E8500 E0 stepping
Hopefully the C2D E8700 will be supported.

[Falkentyne]

What's the voltage ranges?
The 1.02g can't select less than 1.4v on any 45nm cpu.
Can your 1.04g do it with that E8600 ? Ragdd ?

[crishan]

Crishan, stop writing such bull*****! I know that they removed E0 revision CPUs from their site but 2801 BIOS fully supports them! Many people have E8600 with P5W DH Deluxe and they don't have any troubles! Check this out man:

Oh, nice. When you're out of arguments, you try ad hominem attacks, right? .

Since when is simply STATING what the manufacturer writes on their CPU
support page considered, to quote yourself,

writing such bull*****!

Hmm?
Or is that the way one treats one another where you come from?
Never learnt any manners, have you?

AGAIN, I DO HAVE a E8400. E0-Stepping, on a revision 1.04G board, with Bios 2801. Yes, it works, BUT, I am unable to change vCore, the Bios just displays 'Auto'; It also undervolts the CPU slightly (I get 1.168 Volts under load, while my VID is actually 1.2125 Volts).

Many people have E8600 with P5W DH Deluxe and they don't have any troubles!

Not having troubles isn't equal to being unable to change vCore which is my issue with my combination of board and CPU. Note that I didn't write of having troubles, but that I am unable to set vCore.
Through the undervolting, I am e.g. unable to even run LinX for more than a minute; Games, benchmarks etc. run fine, though. I am just very annoyed that such a fine board like the P5WDH is hampered by bugs in its Bios.

Maybe they screwed up at ASUS, putting the E8600 in, while not doing this for the E8400 (stepping E0). This so far seems to be the most logical conclusion. G40RoCkEt, with a revision 1.02G board, at least gets vCore displayed, Ragdd's E8600 works on his 1.04G revision board, although he doesn't write if he's able to change vCore.

[tomcug]

Crishan, once again, with 2801 BIOS all E0 CPUs are fully supported. Didn't you read that man?

Will support: E8400 revE0, E8500 revE0, E8600, Q9400 revR0, Q9550 revE0, Q9650 revE0

ftp://ftp.asus.com.tw/pub/asus/mb/so...eluxe-2801.zip

About your problem, it happens usually. Many people had such troubles with E7200 (M0) and E8500 (C0). So it's not because of E0 revision of your CPU. Send me some shots of your BIOS settings, so I'll be able to compare them with my settings. Sorry for being a little angry but you were just writing fake information from ASUS. They just removed those CPUs from their site because they want people to buy newer boards.

[G40RoCkEt]

I use the E8600 with Revision E0.

No problem here.

[InterMurph2]

I have had my P5W DH Deluxe for over two years. I had my X6800 running reliably at 3.6GHz (360MHz * 10) on air (Arctic Freezer Pro 7). I have 2x2GB Patriot RAM running at 4-4-4-12 on 2.1V.

I only now realize how great the unlocked multiplier was on my X6800. I put a new Q9650 (Revision E0) in, and got it working at the defaults (333 * 9 = 3.0 GHz). I started my overclocking by upping the FSB to 366MHz * 9 = 3.3GHz, and I started running into problems. Windows starts fine, but under load it would restart.

I ran Memtest86+ on these settings, and it ran for 8 hours with no problems.

Then I started tinkering with various voltages, but couldn't make it any more reliable.

Can somebody suggest some baseline settings to start with? And an idea of what I might be able to achieve?

Thanks.

[Falkentyne]

What board revision is that? 1.02, 1.03 or 1.04g? This matters.
vmch, vfsb and what vcore in bios? What vcore is shown in cpu-z?
You should have 1.225-1.7v ranges. If you're getting 1.168v in cpu-z at idle, then you're not going to be doing 3.3 ghz if that board droops down to 1.09v load.

Try vcore 1.35v, as I don't think your board is vdroop modded, so that vcore will drop under load.
vmch 1.65v
FSB term:1.30v
Throw a small fan on the northbridge.
Disable automatic restart in windows in control panel/system/hardware/startup and recovery.

Your FSB isn't going anywhere with a Quad without at least a 1.03g board.

The highest realistic FSB I can be stable with on my QX9650 is 380. (380x9 1.225vcore or 380x10 1.275v), but that requires 1.75 vmch+1.30 vfsb or 1.85vmch and 1.20vfsb. 390 fsb can only be done with insane voltages (like 1.85 mch+1.40 fsb+RAM at 5-5-5-18 at 2.2v, as raising vFSB also raises the vmch slightly (1.85vmch+1.40 vfsb gives about 1.96v vmch which someone measured in a post I can't seem to find anymore...), and even that isn't stable enough to pass all stress tests.

With 333x12, lower on FSB but 4 ghz by raising multi, I can pass prime all day at 1.3625v and pass Linpack 50 loops at 1.375vcore with 2x2GB Ram, or 1.3875v with 2x1GB (huh....).

I found that I can go up to 365 FSB with a Quad fully stable without needing to increase vMCH or vFSB values past the lowest settings. Duals will obviously go much higher.

[InterMurph2]

I'm not sure which board revision I have; I can't find a sticker, and my searches don't turn up any clues about where the revision might be. But I bought it in August, 2006, so I'm assuming it's not a 1.03 or 1.04.

I have a North Pole fan on the Northbridge chip.

Right now, I am running everything at defaults (333 x 9 = 3.0GHz, all voltages set to 'Auto'), and CPU-Z says my Core Voltage is 1.200 - 1.216 V.

I'm going to start tinkering with your suggestions. I just wish that Intel didn't charge such a premium for the 'X'. I paid it 2.5 years ago for my X6800, but chose not to this time.

Thanks.

Edit: At stock everything, and idle, Asus PC Probe II says my CPU temperature is 28 to 30 Celsius, and my MB is 30 to 31 Celsius.

[InterMurph2]

Well there's a clue; I can't go below 1.400 on vCore. I'll try that.

So now I have:

FSB = 370 MHz (x 9 = 3.33 GHz)
vCore = 1.400
vMCH = 1.65v
vFSB = 1.30v

CPU-Z says the Core Voltage is 1.360v to 1.376v.

PC Probe says the CPU is idling at 32 Celsius.

I'm saving this message now because I'm about to put it under load...