Sign in with
Sign up | Sign in
Your question

401 FSB on P5WDG2 WS Pro...What should be my next step?

Last response: in Overclocking
Share
October 10, 2006 9:58:59 PM

Looking for advice on what my next approach should be.

Running 6600 on SI-128 with 102 CFM Sanyo Denki with a 38mm spacer to reduce the dead-spot effect.
6600 and SI-128 were both lapped.
Using AS5.
ASUS P5WDG2 WS Pro Mobo (unmodded)
4 gig Mushkin DDR2-800 4-4-3-8 @1.9-2.1v
x1600 temporary video card

Reading 3611 MHz in CPU-Z

Dual Prime Torture Mode Large FFTs for 9 hours no error. (@ approximately 60c)
Dual Prime Torture Mode Small FFTs for 2 hours no error. (@approximately 60c)
2.5 hours OTComp memtest 86+ no error (I have also run OTComp memtest 86+ at no overclock for 48 hours to quality check the RAM.)
Dual 32M Pi runs fine
3Dmark2006 CPU Score: 3181

Idle Temp: 39c
Load Temp (torture mode): 60c (might be closer to 58 actually, I had high ambient on those 60c numbers).
MOBO Load Temp: 34c

AS5 hasn't set yet so I expect a degree or two lowering of temps. In a nice chilly air conditioned room (where this is going) after AS5 has set I expect this load temp would be 55 or 56c.

BIOS: 0408
BIOS LOG:

CPU Frequency: 401
DRAM Frequency: DDR2-802
Performance Mode: Standard
PCI Express Frequency: 100
PCI Clock Synchronization Mode: 33.33 MHz
CPU Clock Speed Spectrum: Disabled
PCIE Clock Spread Spectrum: Disabled

Memory Voltage: 2.10v
CPU vCore Voltage: 1.5125v (*edit* can now adjust up with BIOS switch)
FSB Termination Voltage: 1.50v
MCH Chipset Voltage: 1.80v (current BIOS max)
ICH Chipset Voltage: 1.20v (current BIOS max)

Default CPU Ratio: Enabled (can’t change…)
VID CMOS Setting: 62 (can’t change…)
Microcode Updation: Enabled
Max CPUID Value Limit: Disabled
Execute Disable Function: Disabled
Single Logical Processor Mode: Disabled
Enhanced C1 Control: Disabled
CPU Internal Thermal Control: Disabled
Virtualization: Disabled
Speedstep: Disabled

CAS Latency: 5
RAS Latency: 5
RAS to CAS Delay: 5
RAS to Activate: 15
Write Recovery Time: 4
Read Delay: 7T

DRAM ECC Mode: Disabled
Hyperpath 3: Disabled
DRAM Throttling Threshold: Disabled

BIOS Power Pull-Down Readings:
vCore: 1.560v
3.3v: 3.280v
5v: 5.043v
12v: 12.091v



It seems stable at 3.6GHz. (Running a longer prime torture session tonight)

My biggest concern is squeezing out every last milimeter of 24/7 performance.
I am unstable at BIOS CPU frequency 412 though I can calculate 1M Pi and Windows comes up fine (prime torture fails in just a few seconds however).


Where do I go from here?

Should I be bringing down some of the MOBO voltages?

Is vCore too high for 24/7, gaming, and the like? I will possibly be moving to a RAID-0 set-up with an Areca card in the PCI-X slot at some point in the not-too-distant future, plus I will be installing a real graphics card and losing the x1600. Is there a greater RAID-0 destabilization risk with my board voltages? How about the PCI-e frequency? Should I be easing that up for graphics card performance?

Should I perhaps look at water cooling rather than air? I would rather not mess with water but I want to KEEP the 3.6GHz overclock for 24/7 at the least. If I can’t squeeze past my current overclock, so be it, but I want to avoid falling backward.

Should I focus on memory timings and speed?

Perhaps I need to use different BIOS?
October 11, 2006 1:39:58 AM

A Voltage that high is not going to create a very long lasting chip. I am amazed it boots, and I doubt you need it... Sounds like a major problem area, and one that may lead you to be without a chip. If you fry it at that level it should SMOKE and I would be supprised if Intel replaces it.

You may end up cooking the MoBo, too.

Are you sure your running two instances of Prime with the "-t" and "-t -An" as the switches? This makes a big difference in how Prime runs, and might start resulting in errors.

If you ARE running prime correctly, and you can lower your voltages, you will have a very fast machine, and a processor that will be on the top of its game.
October 11, 2006 2:09:16 AM

Hmmm...The system is rock-solid stable now after almost continuous testing for 24 hours. Not even the slightest hickup. I have been running prime correctly as well. The P5WDG2 WS Pro is a board built for overclocking.

The chip can't be replaced anyway. I lapped it.

In any event, I AM going to slowly pull back the voltages. After one more day of Prime testing. Shooting for vCore just a shave below 1.5v. Then I'll burn it there for a while before I drop a little further. Ultimate vCore goal is 1.48 range while maintaining the 400 FSB.
October 11, 2006 3:25:25 AM

Quote:
Memory Voltage: 2.10v
CPU vCore Voltage: 1.5125v (current BIOS max)
FSB Termination Voltage: 1.50v
MCH Chipset Voltage: 1.80v (current BIOS max)
ICH Chipset Voltage: 1.20v (current BIOS max)

You've gotta be kidding me...
Make sure you're using SPD as well, it's on the same BIOS-primary as P5W DH.

You really need that 1.60V to have a chance at reaching 4Ghz, 445Mhz FSB.
Your current vMCH is enough for 445Mhz as for all your other voltages.
Just the extremely weak vcore.

vCore @1.55V is when 65nm is at its intended lifetime operating limit (set by Intel), after that the lifetime gets cut down in an accelerated fashion. Starting at ~4yrs @1.60V down to under one month @1.75V.

I'll need to add that into my guide as there are too much FUD and BS flying around I've noticed recently regarding to those voltages.

Either e-mail ASUS to see if they're going to add in higher VID values or else mod it.

I was surprised too. I have a beta bios for this board that unlocks the CPU multiplier down. It may also enable higher vCore. I will be doing the bios flash thing tomorrow when I get to work and see what it gets me.

There may be another bios that opens up vCore a bit, not that I really want to punch much above my current CPU speed. The possibility of the multiplier going down to enable higher FSB though is interesting.

I will be bringing the board voltages down gradually as I guarantee stability at 3.6.
October 11, 2006 6:14:11 PM

New BIOS installed.

Processor multiplier unlocked down.

vCore goes to 1.7v+


I will start new overclocking efforts tonight after work.
October 13, 2006 1:43:45 AM

Playing with multipliers didn't help. I am up against a FSB wall just over 400. Lowering multi just more heavily overclocked my north bridge and prevented post.

Dropped FSB from 401 to 400 (not that it matters much).

Dropped board voltages:
FSB Termination Voltage: 1.50v is now 1.4v
MCH Chipset Voltage: 1.80v is now 1.6v

(Also disabled legacy USB and firewire.)

Running Orthos for 4 hours stable. It will go until noon tomorrow.

My vCore is what it is. I seem very very stable. My temps are hanging at 58*C full load on orthos. Normal use is below 50*C (at most) and idle is 39*C.

Totally stable 3.6GHz/400 FSB on air isn't bad really.
October 13, 2006 1:53:53 AM

Quote:

Make sure you're using SPD as well, it's on the same BIOS-primary as P5W DH.

Is this for stability reasons or something else? I'm running 5-5-5-15 for the stability of loose timings plus the fact that tighter timings don't really gain me much except in memory benchmarks.

Quote:
Your current vMCH is enough for 445Mhz as for all your other voltages.

I can't punch past about 410 FSB no matter what I do. It's like a brick wall. I know there is a dynamic at work here but I'm at a loss to figure it out.
October 13, 2006 2:00:49 AM

you have go over 1ghz of overclocking, thats awesome
October 13, 2006 2:05:39 AM

I would say that is a very decent overclock...
October 13, 2006 5:26:11 AM

Quote:

It's for setting a hidden sub-timing value.

Could you expand on that please?

Quote:
I assume you've replaced the TIM gunk with thermal paste for all the heatsinks you could find on mobo right? (And have ripped off all the ASUS logo plates covering HSs)

Asus plates are off, I am scheduled to do AS5 on the mobo this weekend.

I haven't been registering much heat on the mobo up to this point...though I don't trust my temp number sources.

Asus has all of those large passive heatsink assemblies. Should I lose them and go active?
October 13, 2006 5:37:16 AM

Honestly, what I really wanted to do was crack 450 FSB while staying at 3.6GHz on the CPU via an x8 multiplier (or perhaps even higher with a x7 multi). I was getting some crazy stuff happening as I pushed up past 410 though. Very unexpected. Even at 3.4GHz on the CPU, that FSB wall seems unscalable. In fact, I seem MORE unstable at lower multipliers even staying below 400 FSB.

Note post number 3 by freecableguy in this thread:
http://www.bleedinedge.com/forum/showthread.php?t=23803

What he is describing sounds familiar...

So....if lowering the multi isn't the golden path, then perhaps cooling the mobo better is, as you are indicating.

BUT....I'm not so sure. I noticed that I would go from fine to not even posting with my changes. It wasn't an issue of increasing heat destabilizing my system, it was going from fine to changing the BIOS numbers to not even running. No post. Nothing. I don't think that is heat. The comp didn't even have time to break a sweat. Not to say heat isn't an issue. I have no doubt it is at high mobo volts, but the lack of post leads to me think the effective overclock is just so high it stalls.

I am not comfortable pushing vCore any higher than I have it now with air. So I have to make this happen with an 8x multi if I am going to make it happen at all....I think....
October 13, 2006 2:19:30 PM

I think you got a great overclock. Someone said in another thread that Asus is doing some funny (or creative) things to get past 400mhz FSB.

Anyways, you may be running into the processor capabilities, and/or the 800mhz rating of your memory (assumptions).

I would back down a bit on your overclocking, so that you have a chip that runs for many years to come. Then again, there is no data on the OC lifespan of the chips, and it is probably chip specific.

I talked to Intel about OCing, and they stated what I suspected... That A) it voids your warrenty B) that it will decrease the lifespan of your chip. I dont expect Intel to provide much more information then that. They must love the fact that their chips are so good at OCing, for it makes people like me happy, and quick to recommend the chip to my friends and family; however this is totally different from endorsing it.

You just got to think, why they didnt offer higher clockspeeds on their chips right from the get go. Its probably because of the reliability/lifespan issues.

Anyways, most reports I see is of people with 400fsb. I get my new mobo today, so I will see what the extreme can do. Hopefully I got a good chip. I would be happy with a 400fsb, or a 13/14 multiplier... I hope that I can increase the Memory Multiplier up one step if I got to go the FSB route. I would be extremely happy with the numbers your pushing. I would like to use my 1066mhz memory at the numbers it was made for, but it also gives me that additional room and freedom.

Mike
October 13, 2006 2:21:47 PM

To what extent could the presence of 4 gig of RAM be destabilizing me?
October 14, 2006 2:57:55 AM

Quote:
Using a x86-64 OS will decrease the extent you can overclock when compared to using normal x86 OS, that's a known fact.


How much of an impact does it have... Any idea?

Im not looking for the highest OC, but I love my Windows 64bit, and will be happy with Vista. Right now, most applications dont even run 64bit code, unless it is encoded into the DLL libraries. But I dont care, it is a great OS, and I have no problems with it.
October 18, 2006 5:42:18 AM

It turns out above 1.4 volts FSB can be destabilizing on this board. With 1.4v FSB and vCore to 1.575 BIOS I was able to boot into Windows with 423 FSB and run successful superPI, but it rebooted doing Orthos after a few seconds.
1.55 vCore just crashed on Win logo screen (see the progression?). I suspect pumping vCore a little more would make 3.8 GHz (423 FSB) stable but I am not going to do it. I won't run my volts over 1.6 for any extended period.

In any event, I am Orthos stable at:
389 FSB/3.5 GHz/1:1
3/3/3/4 RAM timings @ 2.2 vDIMM (.1v over rated max)
1.4375 vCore set in BIOS (overvolts to 1.48/1.49v)
1.4v FSB
1.8v MCH
1.2v ICH

Yes, I lowered the overclock a hair to work back up with tighter memory timings. I paid $600 for the stuff and I'll be damned if I am going to run it at crap speeds AND crap timings.

Did the board mod, replacing all of the ASUS thermal material with AS5. Lowered board temps on ASUS probe by an average of maybe 2 to 3*C. Also modded my SI-128, adding kapton tape around the outside to better direct air straight through. Removed and reseated SI-128 heatsink with thinner layer of AS5.

Currently 555/56*C at Orthos load (21*C ambient) with my above listed air cooled set-up. Buying a tuniq tower tomorrow and I'll maybe try that out at some point. I will also be replacing all of my stock board heatsinks for aftermarket.

I will probably try pushing back to 400 FSB since I was there before, I am already very close, and I have even better stability now than previously.


Can I run a small airconditioner IN my computer room recirculating inside air? Ambient below 20*C would be swell...
October 18, 2006 7:17:22 PM

Quote:
Can I run a small airconditioner IN my computer room recirculating inside air? Ambient below 20*C would be swell...


Just to be clear, you know you have to exhaust the heat out of the room, right? I ask because you capitalized "IN". Not long ago I worked with a Ph.D engineer that didn't know, claiming that he thought the 110V A/C line carried the heat away. I was giving him the hairy eyeball, just knowing he was pulling my leg but he was serious. That's what drinking too much vodka in your pre-teen years will get you apparently.

The only problems I've heard of when people blow air conditioned air onto a PC is condensation. If you are in a high humidity area, you will want some kind of humidity sink between the output of the air conditioner and the PC. A louvered vent made of metal will work OK usually. Maybe two if ambient humidity is real high. I live in a dry place and have my game box right above the air conditioner vent. Works like a charm with zero condensation.
October 18, 2006 8:58:02 PM

Quote:
Can I run a small airconditioner IN my computer room recirculating inside air? Ambient below 20*C would be swell...


Just to be clear, you know you have to exhaust the heat out of the room, right? I ask because you capitalized "IN".
Yes, I know. Any additional air conditioning would have to exhaust to another (unused) room rather than outside as this room doesn't have outside windows. It would be recirculating air in the apartment (raising one area to lower another) because I can't duct it out.

Quote:
The only problems I've heard of when people blow air conditioned air onto a PC is condensation.

Don't need to blow it ON the PC (especially since I often sit next to it :wink: ), just want to drop ambient.

It's a non-issue anyway as I was mostly just musing out loud.
October 19, 2006 2:22:10 AM

There is a Air conditioned Case already built...

Its on frozencpu... let me see....

damn, now I cant find it...

Why dont you just make a case out of a freezer. Just make sure its got auto-defrost:) 
October 19, 2006 6:39:39 AM

Quote:
Great, Crashman told me about it so I went out and searched.
Every ASUS board made for Intel on this planet still needs a vdroop mod. It's a design error in the VRM11.0 spec which has been correct by every manufacturer(including ECS...) baring ASUS.
Some are worse than others, out of all of their C2D boards PB5 Deluxe seems to be the least affected while i975X series are heavily bugged.

Help yourself to it
You can just pencil that resistor than remeasure its resistance. It has to be less than half of the original resistance.



What is with ASUS and these voltage problems! :?
October 19, 2006 8:12:45 AM

C2d will die in a month with 1.75vcore?
October 19, 2006 3:36:34 PM

ummm... what have I been saying about Asus?

I have a strong feeling that the corporate culture of Asus, does not promote the highest quality product for the money (aka Quality). These types of issues, and the ones I have experianced lead me to the conclusion that the corporation is not in good shape. Asus has some of the highest costing products on the market; yet their quality doesnt seem to be of the same caliber as other companies.

If I owned Asus stock, I would sell it....

You got a very good overclock there. I hope the voltage issues are not bad enough to cause damage to your chip. My X6800 would heat up so quickly on my Asus Mobo (the P5B Basic) even at stock settings. Now that I have a Gigabyte my temps are 10-20c degrees lower, even if I overclock it by 20%... My Asus couldnt overclock the X6800 by 1-MHz without becoming unstable.

I dont want to overclock my machine as much as you do. 1.5v seems REALLY high, and I just dont feel like having a toasty chip. I wish you well, and a long life on your processor.
October 19, 2006 3:41:18 PM

I got a good question... Ive always assumed that the E6600-X6800 come from the same die, and are labeled via a quality inspection/selection process. Therefor shouldnt the overclocking vCore for a selected Ghz be around the same on all the chips?
October 19, 2006 3:45:09 PM

Quote:
ummm... what have I been saying about Asus?


You mean that you buy before doing product research?

Quote:
Now that I have a Gigabyte my temps are 10-20c degrees lower, even if I overclock it by 20%...


How many variables were changed to result in that temp difference? I see you are also trying to pawn off the water cooling junk that you had on the Asus setup, right? Put it this way: with the same Vcore and same cooler, FSB, etc., I don't think a CPU will run 20C hotter just by changing the mobo. How were these temp differences determined? Onboard sensors? Get a clue.
October 19, 2006 4:39:05 PM

Are you kidding me... Im talking about a manual setting of vCore 1.35 on the Asus P5B with all other settings manually set to stock settings. When C2E was disabled, my temps would run 60c with PRIME on my CURRENT watercooling setup.

I switched to the Gigabyte and am running a vCore of 1.45 with a FSB of 320mhz and it is now over an hour running prime with a temp range of 50-53c.

Yes... My temps have gone WAY down simply by switching MoBo's... In addition I am able to overclock the chip. My old Asus P5B would fail OCing at a single digit rise on the FSB. 266->267

I am currently running over 3.5ghz on 1.45 vCore, and am on 1 hour plus in Prime95...

As for my watercooling "Junk"... the upgrade to my current watercooling system has decreased temps about 3c... I blamed my watercooling setup for the heat problems I was receiving, so I went and spent over $200 for a very high pressure, best of the best, watercooling system. Turns out that Asus was once again to blame. I would have never changed my watercooling setup if I knew that Asus was failing me once again.

The P5B basic is not a X6800 compatable mobo dispite what Asus Says.

The more I read about Asus, the more I see why it is great that I moved to the Gigabyte. My experiance with Gigabyte is amazing. This is the second Asus MoBo that had this same behavor. Simply put, my Asus MoBo isnt dead, it is just poorly designed. My Gigabyte costs the same as the Asus, and has big solid state compacitors, all over it.

Lastly I got stuck reading your stupid message, cause I can only ignore you on a per forum basis.

Clueless lives up to his name once again. I will now push the wonderfull IGNORE button... AGAIN!

Mike
October 19, 2006 7:07:31 PM

Quote:
When C2E was disabled, my temps would run 60c with PRIME on my CURRENT watercooling setup. I switched to the Gigabyte and am running a vCore of 1.45 with a FSB of 320mhz and it is now over an hour running prime with a temp range of 50-53c.


60-53 = 7

60-50 = 10

You said 10 to 20C. Where did the 20C difference go?

Did it fly on the monkey's back on the way outta yer ass?

And once again, how were these temps measured? Do you trust onboard monitors without question?

Quote:
As for my watercooling "Junk"... the upgrade to my current watercooling system has decreased temps about 3c... I blamed my watercooling setup for the heat problems I was receiving, so I went and spent over $200 for a very high pressure, best of the best, watercooling system. Turns out that Asus was once again to blame. I would have never changed my watercooling setup if I knew that Asus was failing me once again.


So you bought not only a mobo that wouldn't meet your needs but also a WC rig that was inadequate... Maybe the old-school method of getting your info straight BEFORE buying woud help you towards enlightenment. No, that won't work cuz then you couldn't come here to sing the blues and say nasty stuff about this or that part maker. Go ahead with your buy-first/think-second plan - at least it provides entertainment.

Quote:
Lastly I got stuck reading your stupid message, cause I can only ignore you on a per forum basis.


You're drowning in the Sea of Incompetence. You're obviously quite at home there.
October 20, 2006 1:49:44 AM

Ive never had a CPU die on me... Everything else seems to break after awhile, but CPU's seem to be immortal. However, I also havent never Overclocked the hell of them either. This latest build is the largest overclocking I have done. I feel safe, only because I run a water system. The Core2Duo E6600+ series is so fast that it really has no need for overclocking. The program I use that requires the beef is Cinema Craft Encoding, and I have gotten that process down to under an hour at default speeds.

No games currently need the CPU speed. They probably wont really need it for about 2-3 years.

I should just run my CPU at stock speeds, until I need more from it... but its just so easy to want to go and get something for free.

My father has a bin he places all the old CPU's in... hes got a hundred or so in there, and I can promise you that every single one of them has been used for a VERY long time, and that every one of them still works.

Ill wait before I push my CPU any more then 1.45v... I will see how other E6600, E6700's hold up under 1.5v before I decide to do it... or wait until the day I actually need the speed:) 

Too bad more people dont visit these sites who own and OC X6800's... I would like to actually talk with some people about this chip. Very little information is even out on this chip. It is very expensive, and I understand why people wouldnt spend the money on it. You could almost buy 3 E6600 for the price of one X6800's...

If they sold multi-processor motherboards, I would have bought 4xE6600's. I wish clustering was possible in the non-linux, non-server world...

Then I would be in E6600 heaven:)  ...and power bill hell:) 
!