Watercooling Results!

[cl0ck_ed]

Hi, building my water cooling rig has taken months due to uni work etc. But I can happily say that the setup is complete.

I'm posting my results to let you guys know how it got on. There are loads of posts about what setup to get but very little feedback as to results.

Ambiet 24c.

Ok, here it goes. Intel E4300 @3.4ghz 1.48v cpu-z with no Vdroop.

Before wc with Tuniq tower on high fan speed the cores peaked at 78c in the occt stress test temp reader.

After wc same cpu ,vcore and clock but with d-tel fuzion, thermochill pa120.3 and 120.1 , laing ddc 18w and a mixture of 3 xinruilian fans and 1 yate loon high. The temp is peaking at 63c.

15c lower load temps I'd call a success. And using 'Real temp' the max temp recorded when occt stress testing is 48c. Which I believe to be the correct temp and the idles on real temp are close to, if not ambient.

I'll post a pic of the machine in a bit.

Also the arctic cooling mx-2 thermal paste has only just been applied for 5mins!!! I was previously running the pc with NO paste and it was idling at 24c when the ambiet was 20c, but load would crash within 2min.

Can anyone else add to these results with there set-ups so that we can embrace the power of water cooling. It would be good to include clock, vcore and cpu used along with the water cooling gear to get a comparison.

 

[FarFarAway]

Nice mate, now crank that CPU up some more

 

[anklebiter]

Nice mate, now crank that CPU up some more

Just goes to show the difference that can be achieved fairly easily, although when will temp become an issue to performance,

If ur going to OC ur cpu, then awesome. but would someone say if having ur system OCed running at peak load at 80c is worse that 40c because arent most systems designed to operate at 80c anyway?

Regards. Ps sorry for posting here, but thought u would know

and Id love to show you how my PC performs, but will have to wait till it arrives *rubs hands with Glee* and then Ill have to quiz u about where the benchmarking software is and which version to use,(internet) and how to run it probably

 

[cl0ck_ed]

Nice mate, now crank that CPU up some more

Lol, having fun overclocking atm. First time i've tried overclocking without temperature being the limiting factor.

Atm stress testing at 3.51ghz, had to increase Vcore to 1.512v in cpu-z. And fsb term maxed at 1.55v bios.

Lol the max temp has only increased by 1c Peak. 1.48>1.512v and 3.4>3.51ghz. Only 1c greater.



24mins left of occt then upping the fsb. These E4300 have a fsb wall at about 400fsb. So the limit with my setup will be the cpu.

I built this excessive cpu only loop for a quad really, I'm prob one of the only people with an effective pa120.4!! just for a dually. haha.

EDIT: 12min left!! of occt, i'll be so

ed if it fails in the last few mins.

 

[cl0ck_ed]

STable at 3.51ghz yay. Now lets pump the fsb.

 

[coffeejunky]

I needs me one of those setups. I darent go above 1.425v (3.2ghz on my E2180) for fear of reaching 80*C.

Looks like a nice clean setup glad it worked out great for you

 

[cl0ck_ed]

I needs me one of those setups. I darent go above 1.425v (3.2ghz on my E2180) for fear of reaching 80*C.

Looks like a nice clean setup glad it worked out great for you

Are, cheers mate. I had a few 85c experiences on air. Don't worry about the temp too much as the chips will just shut themselves off when they get to the t-junction temp. For my e4300 this is 85c and you get a blue screen.

On air the temp rises very fast when you reach high vcore. As the air coolers arn't efficient enough to take away the heat.

I'm aiming for 3.6ghz, but knowing that there is a fsb wall around there, it is doubtful. Don't think I'll upgrade cpu till nehalem comes out.

 

[coffeejunky]

Are, cheers mate. I had a few 85c experiences on air. Don't worry about the temp too much as the chips will just shut themselves off when they get to the t-junction temp. For my e4300 this is 85c and you get a blue screen.

On air the temp rises very fast when you reach high vcore. As the air coolers arn't efficient enough to take away the heat.

I'm aiming for 3.6ghz, but knowing that there is a fsb wall around there, it is doubtful. Don't think I'll upgrade cpu till nehalem comes out.

I wouldn't upgrade 'till then.

The highest I have got out of my e2180 is 3.5ghz.

And I have been hot enough to trigger the thermal trip at 1.565v (some unsuccessful insane suicide runs to get 3.7ghz)

I only have a Big Typhoon as cooling, although both that and my CPU are nicely lapped now.

Hopefully most of the parts of your WC system can be taken over to nehalem (barring the CPU block)

 

[cl0ck_ed]

Hopefully most of the parts of your WC system can be taken over to nehalem (barring the CPU block)

Yeah, i'm also hoping that the cpu block will fit the new sockets, don't see why intel should change it. amd have the same for their am2 and 939.

Just tried 3.6ghz, no luck fails in a few seconds i upped the vcore. no luck. the chip makes a error before it even gets hot.

Knowing that there is a fsb wall with my cpu, I tried to jump over it by going for 3.69ghz. no windows boot.

So, think the magical 400fsb is out of my reach with this chip.

 

[anklebiter]

actually Im excited just reading this

Can you tell me how u change the voltage, is it all in the bios setup, or are you using ubber OCing software. Im a complete beginner, so any help for something that would work on my system and allow me to carefully dabble would be ace, thanks

Regards Ollie

 

[cl0ck_ed]

actually Im excited just reading this

Can you tell me how u change the voltage, is it all in the bios setup, or are you using ubber OCing software. Im a complete beginner, so any help for something that would work on my system and allow me to carefully dabble would be ace, thanks

Regards Ollie

Refering to your setup in your sig, you should get 4ghz easily with a nice air cooler. And possible higher clock with water. As you have an nvidia motherboard some of the bios options are worded/named differently.

All alterations are done in bios.

I'd start by:

Setting overclock to manual.

Slacken off your memory so that it is as slow as possible so that it doesn't limit your overclock.

Up the vcore set to about 1.35v to start with.

leave the fsb at 333mhz

change the cpu muliplier to 11

This should give you 333x11=3.66ghz easy.

As you have a lovely extreme cpu you can use the high 11 multiplier. You may need to adjust the north bridge voltage, fsb(front side bus) voltage. To make it stable.

When this is all stable up the fsb from default 333 to say 350, which should yeild 3.85ghz. This will almost certainly need more voltage.

Just do some more research on the net, there will be plenty of guides for quads on your board.

PS. you get into bios by pushing 'Del', before you boot windows.

 

[cl0ck_ed]

If ur going to OC ur cpu, then awesome. but would someone say if having ur system OCed running at peak load at 80c is worse that 40c because arent most systems designed to operate at 80c anyway?

There is no effect on performance if your pc is running at 40c or 80c. But chips have a shorter life span when run at high temperatures. But this still isn't really an issue as cpus last YEARS and i'm sure most people upgrade every few years. So will never experience a cpu die.

If your cpu is running at 80c on load and on a hot day it reaches 85c its thermal limit, it will crash and turn itself off or restart, so you really want to be a safe distance away from the cpu trottle-ing temperature.

 

[anklebiter]

Hi, building my water cooling rig has taken months due to uni work etc. But I can happily say that the setup is complete.

I'm posting my results to let you guys know how it got on. There are loads of posts about what setup to get but very little feedback as to results.

Ambiet 24c.

Ok, here it goes. Intel E4300 @3.4ghz 1.48v cpu-z with no Vdroop.

Before wc with Tuniq tower on high fan speed the cores peaked at 78c in the occt stress test temp reader.

After wc same cpu ,vcore and clock but with d-tel fuzion, thermochill pa120.3 and 120.1 , laing ddc 18w and a mixture of 3 xinruilian fans and 1 yate loon high. The temp is peaking at 63c.

15c lower load temps I'd call a success. And using 'Real temp' the max temp recorded when occt stress testing is 48c. Which I believe to be the correct temp and the idles on real temp are close to, if not ambient.

I'll post a pic of the machine in a bit.

Also the arctic cooling mx-2 thermal paste has only just been applied for 5mins!!! I was previously running the pc with NO paste and it was idling at 24c when the ambiet was 20c, but load would crash within 2min.

Can anyone else add to these results with there set-ups so that we can embrace the power of water cooling. It would be good to include clock, vcore and cpu used along with the water cooling gear to get a comparison.

Sorry but whats a Vdroop? 1 yate loon high? thanks for the patience with nooby questions,

Ollie

 

[cl0ck_ed]

Sorry but whats a Vdroop? 1 yate loon high? thanks for the patience with nooby questions,

Ollie

Yate loon high, is a high speed fan made by yate loon.

Vdroop, is the amount of voltage the Vcore drops by when under load eg. idle 1.5v load 1.47v for example.

 

[anklebiter]

ah Brilliant, being ultra thick, I dint even know that the volts dropped under load. really appreciate your answers.

So I guess you can safely bring the system back up to top voltage when under load by OCing, or are there more things I should be aware of, How high do you go? are you just looking at the temp guage?

\---

\\\\\\\have just seen your reply Cl0ck_ed, I shall look online for some OCing instructions with my board/cpu,

when u say slack off the memory I presume thats meaning in the bios, to slack off the Ram setting or something. I guess untill Im on the bios screen I will be asking stupid questions.

hehe yeah pressing del to enter setup is one thing I did know

Thanks again,

Ollie

 

[cl0ck_ed]

ah Brilliant, being ultra thick, I dint even know that the volts dropped under load. really appreciate your answers.

So I guess you can safely bring the system back up to top voltage when under load by OCing, or are there more things I should be aware of, How hi do you go? are you just looking at the temp guage?

Vdroop you have to except, it was always happen, but I pencil modded my motherboard. Which is dangerous and could have broken my motherboard. But the mod stops Vdroop.

Vdroop isn't a problem. Just means that while idle the cpu has more voltage than it needs a runs a LITTLE hotter. But this is NO problem.

'bring the system back up to top voltage when under load by OCing', when youy overclock the cpu it will be running faster then normal therefore needs more voltage for it to run. You ONLY give the cpu enough voltage to be stable at the increased clock. You can kill your cpu by putting in the max voltage that your board will allow.

Temperature, is what kills cpu's, voltage is generally fine. BUT there are general guide lines to how much voltage you should use. In general your 45nm chip should NOT get any more than 1.4-1.45v. MAX.

Again Temperature is an early warning sign that your cpu doesn't like the amount of voltage being applied by getting too hot.

 

[anklebiter]

ah righty, thanks again.

I really dont want to fry my lovely new cpu as it costs as much as anything else..

 

[coffeejunky]

One thing to be aware with 45nm range is that the 'accepted' maximum voltage is 1.365v. Above this chips tend to start to degrade (die). Below this they are perfectly safe. With 45nm (which yours is) voltage becomes the limiting factor long before temperature.

All overclocking settings are found in the BIOS. Also you may make your system unstable by overclocking too far on too little voltage - to correct this you will need to clear the CMOS - this is done using a jumper (plastic switch thing) or a button on some higher-end boards. You will need to refer to your mobo manual to find out where this jumper/button is located.

Finally don't be afraid, I've done some pretty extreme overclocks with no real issues...It is a great experience I highly reccomend.

 

[GavX]

45nm can take more than 1.45 I had mine at 1.55 to get a stable 3.84. I say stable, it wasn't... damn thing crashed after 20min! I have heard of people raising ther VCore to te likes of 1.7 but I really would not risk that one, not on a 45nm unless L2 is involved

 

[coffeejunky]

45nm can take more than 1.45 I had mine at 1.55 to get a stable 3.84. I say stable, it wasn't... damn thing crashed after 20min! I have heard of people raising ther VCore to te likes of 1.7 but I really would not risk that one, not on a 45nm unless L2 is involved

Yeah, but would you run a 65nm above 1.5v 24/7.

For 45nm the abslute limit in my opinion is 1.365v. Sure you may be able to go higher but its a risk not worth taking IMO. Many people have run into problems of electromigration and degredation. However, above 1.365v is fine for a benchmark run but I wouldn't risk it for 24/7 usage.