Another P4GD1/Dothan 740 setup with some interesting details inside

**mrlobber** · 08-21-2005, 08:12 AM

Well, finally set up my P4GD1 rev1.03, Dothan 740 and 2x512 Mb Mushkin BH-5, Voodoo3 PCI and PCP&C 850 SSI

With a completely unmodded board and memory at 215 Mhz

, got a 26 sec SuperPi:

It think, not too bad a result, considering also that P4GD1 can give even 1sec lower Pi times than P4C800 Deluxe (at least that's what Charlie once said

)
Then again, it was a completely fresh WinXP install with some of the services wiped out as well.

Anyway, my 740 hits a wall at around 2800 Mhz as well, just as the other bunch of Dothans, except the dreamy 780's, of course. Using a PCI card so far, until my 7800GTX arrives. Then it will be 3D time on DI/LN2

However, now the most interesting part follows:

I'm using stock cooler @ 24 ambient, except a different fan mounted on it, a 70mm 6000rpm screamer from Thermaltake Volcano 7. The temps were sky high though, noticed around 57 C in BIOS IDLE

Found the cause for it as well: the actual VCore the board reports is 0.8v higher than set in bios. For example, if 1.4v in bios, it gives 1.48-1.488 actual, if 1.45 in bios, it gives 1.528 actual

Anyone else experienced such a behaviour??? Especially if Kamerat wrote that 1.45 is the max voltage if using P4GD1 with the CT-479 adapter...

**toledo** · 08-21-2005, 08:29 AM

For 215 Mhz it sonds good . What is the magic ?

.

Any issues with PCI frequecies ?

**mrlobber** · 08-21-2005, 10:23 AM

Originally Posted by toledo

For 215 Mhz it sonds good . What is the magic ?

.

Don't know. Maybe the 2800 Mhz?

Any issues with PCI frequecies ?

Since I have only BH-5 and UTT ram available, and the board is unmodded, I cannot run high fsb on ram without the voltage boost... I did clockgen up to 230 fsb and it didn't freeze, I was able to look around and cpu-z there as well, but, of course, things like superpi crashed instantly there.

Clockgen displays weird numbers, although not as offline as MrBean described here . Regardless of PCI-E frequency set in bios, Clockgen shows it to be 100 (PCI - 33.33), and then it goes up with fsb overclocked... This is regardless of booting under 200fsb, 200 or 200+ FSB.

**MrBean** · 08-21-2005, 10:27 AM

Good going, MrLobber.....but yeah, I note the same issues with your board wrt the PCI/Ex frequencies....not locked

I just posted my 1st deadstock 2.0 Dottie/7800 GTX BFG benchies over here: http://www.systemshock.co.za/forums/...st=0#entry9401

Just put it this way, I was really surprised it stuck in with FX55 in 3dM05, scoring 7922 vs 8047, and actually beating my FX55 in 3dM03, 17128 vs 16600.

So, that should give you an baseline idea of where you should be, at around 2gig. Bear in mind this was done at 133FSB, which is very low.....

3dM01 gave me 26400+, and Aquamark around 80800.....where the FX is stronger, with 85k at defaults.

So yes, sorry for off-topic, just giving you some ideas

Clockgen displays weird numbers, although not as offline as MrBean described here . Regardless of PCI-E frequency set in bios, Clockgen shows it to be 100 (PCI - 33.33), and then it goes up with fsb overclocked... This is regardless of booting under 200fsb, 200 or 200+ FSB.

To be honest, mine also shows correct freq at 133fsb, when I up the FSB, it just follows, same as yours. Just wish we could get working multiplier from within bios for this board. Asus? You care to fix it?

Keep well,
MrBean.

**toledo** · 08-21-2005, 10:35 AM

@mrlobber

Since I have only BH-5 and UTT ram available, and the board is unmodded, I cannot run high fsb on ram without the voltage boost

OCZ booster works in this mobo .....no Vdimm mods required

Clockgen displays weird numbers, although not as offline as MrBean described here . Regardless of PCI-E frequency set in bios, Clockgen shows it to be 100 (PCI - 33.33), and then it goes up with fsb overclocked... This is regardless of booting under 200fsb, 200 or 200+ FSB.

This clockgen is not for this mobo , that is why are you getting weird numbers .

**alexio** · 08-21-2005, 11:10 AM

Good to hear that it does just as good as at the P4P800SE

When I run superpi at the edge of cpu stability the screen always freezes for a few seconds before the data on the screen is refreshed. This happens a few time each run and I was wondering if you have the same at the P4GD1.

I think this cpu will kick some ass at dry ice or LN2

**mrlobber** · 08-21-2005, 11:30 AM

Originally Posted by toledo

OCZ booster works in this mobo .....no Vdimm mods required

I had looked around previously and googled as well but came up empty... How do you know that? Moreover, I'm a little careful about that OCZ state the DDR Booster is not compliant with 24 pin PSU's. Haven't investigated on this matter but it doesn't sound good anyway.

Now if Hipro came out with his DDR Maximizer, we would be on gas...

This clockgen is not for this mobo , that is why are you getting weird numbers .

I saw that in the other thread. Not that it bothers me much (as long as it is working and so far it had

), however, this indicates we need to find other reliable source how to tell the correct PCI-E/PCI frequencies the board is running (a quick thought doesn't reveal me any

). This knowledge might turn out to be critical when more xtreme benching is involved.

Originally Posted by Alexio

This happens a few time each run and I was wondering if you have the same at the P4GD1.

No, I don't have such a behaviour. As soon as SuperPi demands too much, Windows just locks up. Had BSOD's a couple of times as well.

But what do you guys think about the 0.8v additional vcore weirdness I emphasized at the bottom of the 1st post?

**toledo** · 08-21-2005, 11:52 AM

Originally Posted by mrlobber

I had looked around previously and googled as well but came up empty... How do you know that?

Dothan @ CT-479 - howto and fixes

First page (for P4GD1)

OCZ DDR Booster Works (tested 3,3V-3,4V)

I really don`t know who did the test . I`m assuming it is true

**Mr.x3m** · 08-21-2005, 01:26 PM

I tested up to 4V whit p4gd1 and booster.

**toledo** · 08-21-2005, 01:31 PM

Originally Posted by Mr.x3m

I tested up to 4V whit p4gd1 and booster.

24 pin PSU ?

**alexio** · 08-21-2005, 01:37 PM

I tested up to 4.1v (DDR booster reading) with the P4P800-SE with a Tagan 480W which is 24 pins but the motherboard need a 24->20 pin adaptor anyway.

It works fine

**mrlobber** · 08-21-2005, 01:46 PM

Good, now we got that offtopic DDR Booster thing sorted out

(I wonder myself why I missed that Booster info on the first page of the CT-479 thread) Obviously it was too obvious to notice

But what about the Vcore thing? Has anybody noticed anything like that?

**alexio** · 08-21-2005, 01:49 PM

I think you need a multimeter to be sure.

Does the board overvolt at any vcore set in the bios?

**toledo** · 08-21-2005, 01:50 PM

Originally Posted by alexio

I tested up to 4.1v (DDR booster reading) with the P4P800-SE with a Tagan 480W which is 24 pins but the motherboard need a 24->20 pin adaptor anyway.

It works fine

I hope your bh-5 (or UTT ) sticks are not fried

.

Why OCZ does not officially support 24 pin PSU .... ?

Anyway , glad to know it works .

**k|ngp|n** · 08-21-2005, 01:50 PM

Originally Posted by MrBean

Just put it this way, I was really surprised it stuck in with FX55 in 3dM05, scoring 7922 vs 8047, and actually beating my FX55 in 3dM03, 17128 vs 16600.

3dM01 gave me 26400+, and Aquamark around 80800.....where the FX is stronger, with 85k at defaults.

Dothan kicks the

out of an fx in aquamark, in fact dothan kicks the

out of an fx in all 3dmarks, am3, pifast, and of course superpi. This is fact.

**MrBean** · 08-21-2005, 02:24 PM

Hi Kingpin,
Yes, I think this would be quite true, although in my FX55 vs Dottie tests, all things being equal, stock BFG for both, and 2.6gig vs 2gig, the FX beat Dottie by 5k points in Aq3, and about 1k5 in 3dM01. Dottie takes the lead by 500 points in 3dM03, and draw level in 3dM05.

But, I suspect in Aq3 it is the memory bandwidth that makes for this big gap, as my setup was running 133fsb only. I believe at 2oomHz FSB, the Aq3 gap would close considerably, even at 10x200.

Soz for off-topic, mrlobber

**mrlobber** · 08-21-2005, 02:34 PM

Originally Posted by MrBean

Hi Kingpin,
Yes, I think this would be quite true, although in my FX55 vs Dottie tests, all things being equal, stock BFG for both, and 2.6gig vs 2gig, the FX beat Dottie by 5k points in Aq3, and about 1k5 in 3dM01. Dottie takes the lead by 500 points in 3dM03, and draw level in 3dM05.

Kingpin probably meant it clock-for-clock. Indeed, a 2.6 Ghz Dothan would probably smoke the FX...

Originally Posted by MrBean

Soz for off-topic, mrlobber

Nevermind, it's always nice if it leads Xtreme legends posting in the thread you created

Originally Posted by Alexio

I think you need a multimeter to be sure.
Does the board overvolt at any vcore set in the bios?

I'll get it out as soon as I can. However, seeing those 50+ idle temps, either it's complete non-contact with the heatsink (which isn't the case, I checked), or the overvoltage might be there or all the motherboard sensors are telling absolute crap to me.

And yes, it happens at any vcore set in the bios.

**k|ngp|n** · 08-21-2005, 02:54 PM

Originally Posted by MrBean

Hi Kingpin,
Yes, I think this would be quite true, although in my FX55 vs Dottie tests, all things being equal, stock BFG for both, and 2.6gig vs 2gig, the FX beat Dottie by 5k points in Aq3, and about 1k5 in 3dM01. Dottie takes the lead by 500 points in 3dM03, and draw level in 3dM05.

But, I suspect in Aq3 it is the memory bandwidth that makes for this big gap, as my setup was running 133fsb only. I believe at 2oomHz FSB, the Aq3 gap would close considerably, even at 10x200.

Soz for off-topic, mrlobber

133 vs 200 is not all things equal, so dothan doesnt "get beat"....ever.

NOt trying to start

, just dont want anyone to misconstrue your statements. I do realize that you are pobably comparing out of box to out of box....BTW-- nice results with the p4g mr. Lobber, these are really nice boards(i really like the azalia high def. sound) and I still feel privleged to have one for daily pc.

**alexio** · 08-21-2005, 02:58 PM

Try the 400fsb jumper or no jumper at all (this defaults to 400fsb). I heard about the temsensor reporting a 10 degrees lower temp compared to the 533fsb jumper when using the 400fsb jumper at the Asus P4*800 boards.. I'm not sure which value is right and I've not tested if I had the same drop. I'm sure this temperature drop is tempsensor related and not real temp related but the question is: which reading is the closest to the real temp?

**MrBean** · 08-22-2005, 12:17 PM

I do realize that you are pobably comparing out of box to out of box....

Exactly what I meant. Just debating why it would perform so good in 3dM03/05, even at 133fsb vs FX55 stock, and being left a little behind in Aq3.

My best bet is memory bandwidth, as Aq3 relies heavily on that for good results, so, if I could up my fb to 200, drop multi to 10, this gap would probably close completely. I saw a 1sec drop in SuperPI, keeping cpu speed the same by decreasing multi, and increasing fsb to 143. 2007mHz (133fsb) vs 2002mHz (143fsb) , improves SuperPI from 37.78 to 36.65 sec.

But, enough off-topic, I am getting worried looking at MrLobber's avatar

@ MrLobber: You experiencing any POST issues with PCI-Ex GFX card installed?

**jrw** · 08-23-2005, 01:18 AM

Nice there mrlobber. I have a multimeter and will try foguring out how the vcore is acting on this thing.

I get my P4GD1 today, so now i just need a new GFX card. Btw, a PCI graphics card should be working if plugged in, right? I mean, it doesnt take any "tricks" or something? Bought an old one which seem to be dead.

**MrBean** · 08-23-2005, 11:31 AM

Hello Stryg,
One thing you may need to do 2 or 3 times would be holding down the "delete" key, and power on. Keep the key depressed until GFX init.

Other than that, it should require no input/setting adjustments, even if GFX init order is set PCI-Ex/PCI in Bios, if no PCI-Ex card found, it would boot PCI card.

**jrw** · 08-23-2005, 02:28 PM

Originally Posted by MrBean

Hello Stryg,
One thing you may need to do 2 or 3 times would be holding down the "delete" key, and power on. Keep the key depressed until GFX init.

Other than that, it should require no input/setting adjustments, even if GFX init order is set PCI-Ex/PCI in Bios, if no PCI-Ex card found, it would boot PCI card.

Ok Ill try that.

Im just all

since the guy I bought it from confirmed that it worked before shipping it. Nobody would cheat someone for 8$

Sorry for jacking mrlobber, Ill post some comparetive stuff when I get my 730/P4GD1/6600GT combo running... (I envy your 2.8er - mine's outta steam at 2739MHz for 1M :/)

**mrlobber** · 08-24-2005, 01:10 PM

Originally Posted by stryg

Sorry for jacking mrlobber, Ill post some comparetive stuff when I get my 730/P4GD1/6600GT combo running... (I envy your 2.8er - mine's outta steam at 2739MHz for 1M :/)

Do it so that more P4GD1 results keep coming

**Kane** · 08-26-2005, 02:43 AM

Here is my setup. Been up and running for about 3 days. Low FSB is becasue I've heard that the locks don't work. I'm using this machine 24/7 and don't want to loose a harddrive or four

Asus P4GD1
Dothan 740 Cooled by Zalman 7700Cu
2x512MB BH-5 tried with four sticks, but BSOD's all the time

7800GTX
1x74GB Raptor
3x250GB WD SATA

I think there is something wrong with the temp readings. I'm getting close to 50's idle and right over 60 load. Seems a bit to high. Oh, and mrlobber. My card overvolts just as yours do.

Had it priming overnight @ 2.8GHz with 1.475v in bios. Gives me 1.525v in Windows. Also haven't tried 3DMark2001 jet. I'll let the pictures speak for them self.