Anyone overclocking a Core 2 Duo E4xxx?

All about them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Anyone overclocking a Core 2 Duo E4xxx?

Post by smilingcrow » Wed Apr 25, 2007 10:57 am

I’m thinking of buying an E4400 and changing the FSB from 800 to 1066 which will increase the clock speed from 2 GHz to 2.67 GHz. i.e. an E6700 on the cheap but with half the cache.
But I remember that certain motherboards would automatically boost the VCore if the FSB was run above stock. This was with the Core 2 Duos that ran at a stock 1066 FSB, so I’m wondering if the same behaviour occurs with Core 2 Duos that run at 800 stock. Has anybody tried this?

It has a large implication for power consumption as I previously measured a difference of 23W at idle due to this effect; viewtopic.php?t=34167&highlight=

derekchinese
Posts: 255
Joined: Fri Mar 02, 2007 6:57 pm
Location: Nowhere.

Post by derekchinese » Wed Apr 25, 2007 11:39 am

But, don't you just manually set the VCore in the BIOS?

From what I have seen, AUTO Vcore is generally not very safe.

I would think that an overclock from 2ghz to 2.67ghz would require a very small boost of Vcore.

Derek

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Wed Apr 25, 2007 12:24 pm

derekchinese wrote:But, don't you just manually set the VCore in the BIOS?
From what I have seen, AUTO Vcore is generally not very safe.
I would think that an overclock from 2ghz to 2.67ghz would require a very small boost of Vcore.
I don’t want to set the Vcore in the BIOS; if you do that you lose the power saving feature of Speedstep dropping the VCore at idle. The whole point is to keep auto Vcore; reread the post I suggest.
Auto VCore is fine when overclocking provided you aren’t too ambitious.
Most Core 2 Duos seem to be good for at least 2.9 GHz at stock voltage, so 2.67 GHz isn’t at all ambitious.

jojo4u
Posts: 806
Joined: Sat Dec 14, 2002 7:00 am
Location: Germany

Post by jojo4u » Wed Apr 25, 2007 1:04 pm

I had my hands on a Gigabyte 965P-DS3 and if I remember correctly, vcore was constant. The BIOS even had a red caption saying you don't have optimal overclocking if you don't put Vcore to Auto instead of Default(?).

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Wed Apr 25, 2007 1:13 pm

jojo4u wrote:I had my hands on a Gigabyte 965P-DS3 and if I remember correctly, vcore was constant. The BIOS even had a red caption saying you don't have optimal overclocking if you don't put Vcore to Auto instead of Default(?).
When you say VCore is constant, under what circumstances?
I don’t know what the terminology means in Gigabyte speak! Default I can understand but what does Auto mean in this context?

derekchinese
Posts: 255
Joined: Fri Mar 02, 2007 6:57 pm
Location: Nowhere.

Post by derekchinese » Wed Apr 25, 2007 1:50 pm

Hey Smilingcrow, I just suggested the BIOS Vcore because I have heard alot of bad things about auto Vcore. I don't know if you have as well, but it really seems like when you do overclock, the mobo will always give too many volts. I guess this is why you were using clockgen though :). It's just that I have heard people had their CPU fried because AUTO gave them too many volts.

Just my thoughts though.

Derek

jojo4u
Posts: 806
Joined: Sat Dec 14, 2002 7:00 am
Location: Germany

Post by jojo4u » Wed Apr 25, 2007 6:09 pm

smilingcrow wrote: When you say VCore is constant, under what circumstances?
I don’t know what the terminology means in Gigabyte speak! Default I can understand but what does Auto mean in this context?
Vcore is constant when overclocking via FSB in BIOS and keeping the Vcore setting at Default or so. When you put the Vcore setting to Auto, there is some overvolting depending on the overclock. Please get some reinsurance about this since it's quite a time ago since I messed around with the board.

dragmor
Posts: 301
Joined: Sun Jul 10, 2005 7:54 pm
Location: Oz

Post by dragmor » Wed Apr 25, 2007 8:00 pm

I've got a E4300 on a GA-965-DQ6 rev1. The first motherboard I had was not stable at stock (just didnt like 9x200), but was fine when overclocked. It had other issues with on board raid so I ended up RMAing the board.

The 2nd board is stable at stock but the mounting holes were damaged. I cant get the ninja to click into place (I've already trimmed the bracket so its not touching caps). I could only just force the stock heatsink on (and I mean force). I would use a bolt through kit but that means removing the backplate that comes with the board. So I'm leaving it with the Stock HS and selling it when the K10 comes out.

1st board, load temps with Ninja.
2.4ghz @ 1.25v TAT = 55c
3.0ghz @ 1.38v TAT = 72c
3.6ghz @ 1.50v TAT = 85c

2nd board, load temps with stock Intel HS
1.8ghz @ 1.00000v TAT = 50c
2.4ghz @ 1.11825v TAT = 72c

I'm sure the CPU could go below 1v and the motherboard lets me select down to 0.8v but auto sets the voltage to 1.2v when I go below 1v.

Matija
Posts: 780
Joined: Sat Mar 17, 2007 3:17 am
Location: Croatia

Post by Matija » Thu Apr 26, 2007 12:10 am

I have an E4300 on a P5B Deluxe. Auto voltage.

I don't have a clue about the voltage on the default frequency (9x200), but overclocked to 9x266=2400, the board sets the voltage to 1.28V. With EIST and 6x266=1600, the board sets the voltage to roughly 1.12V.

I *think* the board set 1.35V when I OC'd the little thing to 2700 MHz for a short while, but I'm not sure.

So yeah, ASUS does that. However, can't you use RMClock or CrystalCPUID to control the voltage after booting?

Willy Higinbotham
Posts: 102
Joined: Sun Mar 11, 2007 12:39 pm

Post by Willy Higinbotham » Thu Apr 26, 2007 4:18 am

dragmor,

I remember reading somewhere that someone have managed to put a backplate on top of the existing cooler plate that comes on the DQ6 by using some plastic spacers. That was an easy DIY.

That way he was able put a tower HSF without removing the "Crazy Cool" plate. And he was more than happy with the results.

dragmor
Posts: 301
Joined: Sun Jul 10, 2005 7:54 pm
Location: Oz

Post by dragmor » Thu Apr 26, 2007 5:11 pm

Willy Higinbotham wrote:dragmor,

I remember reading somewhere that someone have managed to put a backplate on top of the existing cooler plate that comes on the DQ6 by using some plastic spacers. That was an easy DIY.

That way he was able put a tower HSF without removing the "Crazy Cool" plate. And he was more than happy with the results.
Link? I've looked at the Ninjas mounting system and I figure with some nuts & bolts and a couple of washers I could mount the system I'm just not sure how good it would be and it would require destorying the pushpins of on the ninja. The other hassle would be getting some sort of brack to spread the pressure around the motherboard rather than on the mounting holes.

Willy Higinbotham
Posts: 102
Joined: Sun Mar 11, 2007 12:39 pm

Post by Willy Higinbotham » Fri Apr 27, 2007 6:30 am

dragmor wrote:Link? I've looked at the Ninjas mounting system and I figure with some nuts & bolts and a couple of washers I could mount the system I'm just not sure how good it would be and it would require destorying the pushpins of on the ninja. The other hassle would be getting some sort of brack to spread the pressure around the motherboard rather than on the mounting holes.
OK, after checking my browser's archive, I've managed to find the link.

Here you go:

http://www.gam.net.au/pcmods/dq6ttbt/

He has installed a Thermaltake Big Typhoon on the DQ6.

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Sun Apr 29, 2007 2:56 am

Matija wrote:I have an E4300 on a P5B Deluxe. Auto voltage.
I don't have a clue about the voltage on the default frequency (9x200), but overclocked to 9x266=2400, the board sets the voltage to 1.28V. With EIST and 6x266=1600, the board sets the voltage to roughly 1.12V.
I *think* the board set 1.35V when I OC'd the little thing to 2700 MHz for a short while, but I'm not sure.
So yeah, ASUS does that. However, can't you use RMClock or CrystalCPUID to control the voltage after booting?
Thanks for the info. If the BIOS leaves the Vcore at stock when you overclock then RMClock should still work for voltage control. If it tweaks the VCore automatically even when it’s on default then it probably will not work.

The BIOS in the S3 works as I hoped; with the VCore on stock and the FSB increased from the stock 200 to 266, the VCore stays within the same range – 1.136 to 1.28V.
RMClock can adjust the VCore but with such a narrow range to work with it’s not such a big deal. There’s plenty of room for undervolting in the BIOS though.

I have an Alpine 7 installed at the moment and I don’t think the contact is good enough according to the temps that I’m seeing so I haven’t stress tested it for stability. Speedfan’s CPU Core temps don’t match those of TAT and CoreTemp not that it’s important as it doesn’t allow fan speed control as far as I can see so far. There are quite a few BIOS variations that I could try but I’ve tried the obvious ones with no joy.
The clearance around the socket is poor so I’m not sure if one of the Scythe towers will fit; looks like I’ll try a different motherboard.

There doesn’t seem to be an obvious choice for an LGA 775 motherboard and heatsink combination for Core 2 Duo. Scythe’s new mounting method for the Ninja+ sounds a bit hit and miss in terms of clearance and effectiveness; a backward step for LGA 775 boards.
Motherboard support for Speedfan fan control seems to be rare, which leaves me with these four candidates:

Asus P5B – low power consumption, unknown speedfan support - £62
Asus P5B-E Plus – low power consumption, okay Speedfan support - £90
Asus P5N-E SLI – probable medium to high power consumption as its SLI, Speedfan reported as working but no details given - £63
Asus P5W DH Deluxe – medium power consumption, excellent speedfan support - £111

The prices are the best I can find and include shipping. I’d prefer optical audio out on the back plate which rules out some boards; the P5N-E SLI doesn’t have it but it’s cheap enough to budget for an adapter.

With support for the new Scythe mounting mechanism also not clear it’s hard to know which board to go for!

Matija
Posts: 780
Joined: Sat Mar 17, 2007 3:17 am
Location: Croatia

Post by Matija » Sun Apr 29, 2007 4:47 am

In case you're interested: I've downloaded RMClock and played around a bit. It works even with automatic voltage control set in the BIOS, but only for downvolting. It won't let me set more than 1.28V, which is what the P5BD had chosen for 2.4 GHz. I might have messed something up, as at one point, I had the processor downclocked to 1.6 GHz at 1.28V, and it wouldn't budge until I somehow did something ;) Not a very user-friendly app, I must say...

I can't help you with information whether SpeedFan works or not, as I still have only one controllable fan in the case, on the stock CPU cooler. The ASUS Q-Fan thing does its job well enough for me, but if you want me to test it, I can try.

Oh, one fun thing: I downvolted the CPU to 1.05V on 2.4 GHz and it worked okay for half an hour. The core temperatures under full TAT load went down from 71 to 63C, but I didn't like how the CPU fan sounded (it's worse on ~1100 rpm than on ~1400), so I stopped my experiment and set the voltage back to 1.28V.

As for the cooler, I'm thinking about the Noctua UH-12F, but I'm having a hard time justifying its cost/benefit. The stock cooler works okay, I'm not doing anything CPU-intensive unless I'm playing a game (which isn't THAT often), and then the sounds drown-out the CPU fan (not so with the GPU fan on my x1950 Pro, I definitely need a proper cooler for it).

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Sun Apr 29, 2007 7:04 am

Matija wrote:In case you're interested: I've downloaded RMClock and played around a bit. It works even with automatic voltage control set in the BIOS, but only for downvolting. It won't let me set more than 1.28V, which is what the P5BD had chosen for 2.4 GHz. I might have messed something up, as at one point, I had the processor downclocked to 1.6 GHz at 1.28V, and it wouldn't budge until I somehow did something ;) Not a very user-friendly app, I must say...
I looked at RMClock version 2.25 for the first time over the weekend and it seems to have gone backwards in the usability stakes, but it does have more features. I couldn’t get it to open minimized which is a minor annoyance.

RMClock and other utilities can show false VCore readings if you change the VCore in the BIOS from stock with certain BIOSs. The only way to test this is to measure power consumption or CPU temps. I always confirm with power consumption tests as it’s quicker and more accurate I imagine.
Matija wrote:I can't help you with information whether SpeedFan works or not, as I still have only one controllable fan in the case, on the stock CPU cooler. The ASUS Q-Fan thing does its job well enough for me, but if you want me to test it, I can try.

Oh, one fun thing: I downvolted the CPU to 1.05V on 2.4 GHz and it worked okay for half an hour. The core temperatures under full TAT load went down from 71 to 63C, but I didn't like how the CPU fan sounded (it's worse on ~1100 rpm than on ~1400), so I stopped my experiment and set the voltage back to 1.28V.
It would be great if you could test Speedfan with your CPU cooler. You’ll need to disable Q-FAN in the BIOS usually.
Your experience with auto BIOS control of fan speeds is why I prefer to use Speedfan if I want to adjust them automatically. I’m happy to set case fan(s) manually but I like to have an adjustable CPU fan ideally.

charlaph
Patron of SPCR
Posts: 32
Joined: Mon Jan 31, 2005 1:20 pm
Location: Dundee, Scotland

Post by charlaph » Wed May 02, 2007 4:07 am

I'll try this for you tonight. I have an E4400 running on a DFI Infinity 975X, but I've set the vcore and fsb manually in the BIOS and disabled speedstep. I did this because I have a Scythe Infinity and Yate Loon running at ~6V and they keep the CPU cool and it's quiet enough for my needs at the moment, but I'll try putting everything apart from the fsb back to stock and see what happens.

FYI, I've got mine running at 300Mhz (3Ghz) and vcore is at 1.36v. Under full load this drops to 1.328v, which results in temps stabilising at 52°C, and at idle it's around 32°C. I haven't messed around with it very much as I only got it set up on saturday and 3Ghz was the figure I was aiming for which it hit first try with minimal tweaking and okay temps. This is my gaming setup so silence isn't top priority, but I'm very happy with the performance of the chip in all areas.

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Wed May 02, 2007 12:05 pm

charlaph wrote:I'll try this for you tonight. I have an E4400 running on a DFI Infinity 975X, but I've set the vcore and fsb manually in the BIOS and disabled speedstep. I did this because I have a Scythe Infinity and Yate Loon running at ~6V and they keep the CPU cool and it's quiet enough for my needs at the moment, but I'll try putting everything apart from the fsb back to stock and see what happens.

FYI, I've got mine running at 300Mhz (3Ghz) and vcore is at 1.36v. Under full load this drops to 1.328v, which results in temps stabilising at 52°C, and at idle it's around 32°C. I haven't messed around with it very much as I only got it set up on saturday and 3Ghz was the figure I was aiming for which it hit first try with minimal tweaking and okay temps. This is my gaming setup so silence isn't top priority, but I'm very happy with the performance of the chip in all areas.
Thanks. What are you using to monitor the CPU temps?

charlaph
Patron of SPCR
Posts: 32
Joined: Mon Jan 31, 2005 1:20 pm
Location: Dundee, Scotland

Post by charlaph » Wed May 02, 2007 11:18 pm

It didn't make a difference on the DFI. If I set everything to auto and enabled speedstep it dropped the voltage to 1.16v when it dropped the multiplier, and upped it to 1.33v under load (which is still under the recommended vcore, but I think that's down to vdroop being so bad on this board)

If I set the fsb to 266 but left the voltage setting at auto it still swapped the voltage between 1.16v and 1.33v, no different to above.

If I set the fsb manually to 300Mhz and the Vcore manually to 1.35v¹, but left speedstep enabled, it still reduced the voltage to 1.16v when idle, although obviously the PC was running at 1.8Ghz, then under load it went up to 1.31v - this is now my 24/7 setting.

Thanks for posting this thread though, I wouldn't have bothered messing around with speedstep if you hadn't prompted me as I wasn't impressed with it on my P4 setup. I've now reduced my idle temps by 8°C, and of course my power draw is down as well. It was amusing watching it all running last night because I was encoding music to ogg and it only actually sped up a couple of times - there's so much headroom at 1.8Ghz that normal usage doesn't trouble it.

Oh, and to answer your question I'm using Speedfan and SmartGuardian to monitor temps. Unfortunately CoreTemp makes the board crash, I'll try downloading TAT tonight. I was using CpuZ and SmartGuardian to monitor voltages.



¹I set the vcore to 1.35v in the BIOS but it never actually runs at that. Without speedstep it droops to 1.31v under load, same as with it enabled.

kaange
Posts: 275
Joined: Thu Dec 21, 2006 6:58 pm
Location: Melbourne, Australia

Post by kaange » Sun May 06, 2007 9:26 pm

Matija wrote:Oh, one fun thing: I downvolted the CPU to 1.05V on 2.4 GHz and it worked okay for half an hour. The core temperatures under full TAT load went down from 71 to 63C.
I got the vCore down to 0.98725V stable on my E4300 (std clocking) the other day and it has made a huge difference to temps - both idle and full TAT load.

Below 0.925V, the bios would not post while under 0.9750V, XP would not load. While XP loaded from 0.9750V, testing using Orthos generated errors until 0.98725V.

I'm now using this for 'normal' operations and clocking up for intensive processes.

Post Reply