do i go for a chipset block, or an air cooled one?

[veryhumid]

Next week I will hopefully be buying an Apex kit, or if not the kit, the various main parts (apogee, mcp-655, mcr-220). My main goal is to be able to run my CPU colder and overclock it, or run it for silence. I'd like to do a GPU block, but it currently stays very cool on the air heatsink and would require some work by viperjohn to make room for it (the block). For the sake of keeping this simpler I won't do that yet. So for now I am definitely getting a cpu block, but I have one more problem. The chipset fan on my board is notorious to die, and it has. I've gotten by passively but I won't be able to with watercooling and overclocking. Do I go with the swiftech chipset water block (http://www.sidewindercomputers.com/swmcchco1.html) or the air cooler (http://www.sidewindercomputers.com/swmcchco.html) ?

the air cooler is totally silent as far as i'm concerned, and for comparison, assume they cost the same amout. it comes down to whether there would be any significant pressure drop/increase in temps by putting the NB in the loop. thanks for the advice guys!

[dylskee]

I'm actually wondering the same thing, i just put the Swiftech H20-220-APEX ULTRA kit in with a gpu block too. I noticed the NB was scorching hot so i bought a 40mm fan and taped it on there but I'm not sure if that's doing the trick. I was wondering if i added a chipset block to the loop if i would see an increase in temps from the added heat. I would think if your not going to put a block on the video card i would definitely put one on the chipset.

[veryhumid]

eventually, i do plan on doing a gpu block, but not for a while. I'm thinking the 875 chipset is maybe 20 watts or something? hopefully someone will post who knows what happens. the pressure drop seems minor, I'm more worried about a significant increase in CPU temp I could easily avoid by going with air.

[batboy]

With Intel systems, I would be worried about raising the temp of the N/B if you have a block on there. I went with N/B air cooling to simplify things. If your current N/B air cooling is not enough or if you are trying to go quieter, put a bigger sink on the N/B.

[veryhumid]

so you are thinking a waterblock just wouldn't be as effective dissipating heat as an air cooler?

[Aidenswarrior]

i have a chipset block on my system. Keeps the chipset around 29 degrees celc. With an air cooler, it was running around 36-38. The chipset didnt change the temps of the other items in my loop at all.

[dylskee]

How can i check the temps of my NB without a probe contact on it? /thread jack

[voigts]

watercooling the chipset is going to lower its temps far more than air cooling just as it does for CPU and GPU. Most people however simply do not need the increased cooling of using water on the chipset as the chipsets usually don't get hot anyway. The little bit of heat put out by the chipset is not even a burp in the loop as far as heat dump is concerned, but it can make tube routing more difficult. The Swiftech block is about as unrestrictive as it gets for a chipset waterblock.

Having said that, I have seen some people report with certain motherboards high chipset temps when overclocking. If your chipset is too hot, you can always just go with a larger air heatsink, or if you want silence or better temps, you can always add for example the swiftech or another chipset waterblock. Its just that trying to plumb tubing to a chipset block results in some very sharp turns.

[batboy]

I just use the "system" temp to sort of get an idea of N/B temp. The system thermister is on the mobo close to the N/B. Doesn't give you actual N/B chip temp, but if you improve N/B cooling, you'll notice system load temp will go down. Unplug the N/B cooler fan and you'll see system load temp go up.

With my system, water temp is higher than my system temp on air. In other words, my system temp would go up if I watercooled. I'm running my N/B passive, so can't beat that for silence.

[dylskee]

With my system, water temp is higher than my system temp on air.

I noticed the same thing so i put a fan on my NB and a couple fans on the fets and memory and i saw a 6°c system temp drop. It's getting a little loud now that i've added some fans to the mix so i just bought a fan controller to keep things quiet when i'm surfing and i can crank em back up when gaming or encoding.

[batboy]

watercooling the chipset is going to lower its temps far more than air cooling just as it does for CPU and GPU.

That might be true with most AMD systems, but not always true with some of these newer Intel systems.

[speed bump]

It really depend on your chipset I have heard people got decent gains WCing the 865/875. The NF2 was the quintessential example of chipset that should be WCed. On my Asus p5wd2-p WCing the cipset was worth about 3more MHZ FSB compared to stock with a Nidec beta 5 blowing over it.

[AlanD911]

I have heard from the dfi-street forums that overclocking a dfi motherboard is kind of useless since air cooling will give you low temps allready. THe main reason why people watercool their chipset is so that they can get a silent computer.

[veryhumid]

there are air coolers with silent fans, passive sinks which are naturally silent. water doesn't guarantee silence because it depends what is on the rad.

sounds pretty much like a toss up so far. I'm not worried about the curve in the tubing because i'm going to try the cool sleeves.

[Spade]

That might be true with most AMD systems, but not always true with some of these newer Intel systems.

that makes no sense.

why?

your saying watercooling will not help newer intel chipsets. why?
those chipsets are not intel cpu's notorious for heat right now and
40+ on water. i fail to see why it wouldnt lower chipset temp any
worse just because youre using intel gear rather than amd..

[batboy]

At 4.5 gig with elevated vcore, my CPU load temp is 45 degrees. With prolong usage, the water temp will be about that too. My system temp is 30 degrees or less. I know that's not exactly N/B temp, but it's not too far off. Why would I want 45 degree water running through a chipset block to raise N/B temps higher than what I get with passive air cooling? With these new Intel systems, the mosfet temps are higher than N/B chip temp. So, you need some case ventilation regardless. Like Humid said, it's probably 6 of one and half dozen of the other. Personally, I like simple water cooling systems. CPU and maybe GPU if you have a high end vid card.

[veryhumid]

that is an excellent point! I forgot to mention that this rig has been folding 24/7 for almost 2 years, and it's going to keep doing it *knock on wood*. Running at load 24/7 would definitely be putting warmer water through the block than the approx. temp of the chipset.

I am leaning towards the swiftech air cooler now.

[batboy]

Oh yeah, I'm using those Cool Sleeves. I highly recommend 'em if you have tight bends. They work great and look good too.

[speed bump]

At 4.5 gig with elevated vcore, my CPU load temp is 45 degrees. With prolong usage, the water temp will be about that too. My system temp is 30 degrees or less. I know that's not exactly N/B temp, but it's not too far off. Why would I want 45 degree water running through a chipset block to raise N/B temps higher than what I get with passive air cooling? With these new Intel systems, the mosfet temps are higher than N/B chip temp. So, you need some case ventilation regardless. Like Humid said, it's probably 6 of one and half dozen of the other. Personally, I like simple water cooling systems. CPU and maybe GPU if you have a high end vid card.

Get a thermometer and check that water temperature actually it will be about 20°c not even 30°c water. Your CPU temp will not be the same as your water temp, in fact after you hit 20+ °c your cpu temps will go up quite abit.

Also depending on your board you sys temp could be further away from you NB than you think. This is the reason why if you take ventilation away from that area your temps go way up but if you add in more ventilation the temps drop siginifcantly.

[ƒÓÒl]

Just to chime in..my NB was running at 56-60C though my system temp was substantially lower (DFI nf3).
Too toasty for my liking, so it's watercooled now with one of those Gemini blocks that are on sale.

I don't overclock, so there's no comparison to give you there, but I'm happier none the less now that it's much nearer to the case temp.

[veryhumid]

hmmm, okay so what is the water temperature like in a loop? is it near what the cpu is loaded at? or around ambient?

[batboy]

I'll guarantee my water temp is more than 30 degrees when I'm gaming for a couple of hours. But, you are right about water temp being actually somewhat less than CPU temp. On the other hand, water temp depends on which side of the rad and/or which side of the CPU block you measure. I was thinking, what if you plumb the N/B block before the CPU block? That way the water would be cooler, before the CPU block heats it up.

[veryhumid]

lol that is an interesting idea. but my rad is on the roof of my case and I think things would get complicated. I think air will be simpler, at least for now.

[clocker2]

At 4.5 gig with elevated vcore, my CPU load temp is 45 degrees. With prolong usage, the water temp will be about that too. My system temp is 30 degrees or less. I know that's not exactly N/B temp, but it's not too far off.
You have made a couple of false assumptions here.
Firstly, the temp of the water in the loop will be very close to room ambient.
Second, there is no way to estimate the temp of the northbridge based on a "system" temp derived from a sensor located who knows where on the motherboard.

Before I added the Maze chipset block, the NB was the hottest component on my motherboard.
Adding in the block (and BTW I completely agree with voights about the added complexity that the extra block will introduce...don't count too much on CoolSleeves solving all the issues) did not impact the CPU/GPU temps whatsoever.

[batboy]

Again, it depends on which side of the rad and which side of the CPU block you measure water temp. I would hope the water temp would be cooler as it exits the rad. But, as the water exits the CPU block that is sitting on top of a dual core 3.2 Presler running at 4.5 gig with 15% more vcore than default, water temp ain't going to be near ambient. If it was, you wouldn't need a radiator then would you?

Admittedly, my experience with comparing system temps with N/B temps are limited to Abit motherboards. Perhaps other manufacturers have the system thermister located farther away. But, on motherboards I've used, the system temp is very comparable to N/B temp. Like I said earlier, improve N/B cooling and system temp goes down. Unplug the N/B fan and system temp goes up. How is that a false assumption?

I can place my finger on my passively air cooled N/B sink even at top O/C speed under load. It feels cooler than when I place my finger on the CPU block. On the other hand, I burn my finger when I touch the mosfet sink.

[veryhumid]

Firstly, the temp of the water in the loop will be very close to room ambient.


Even if it is running at full load indefinitely?


here we go, guys, some basic data showing water temps: http://www.ocforums.com/showpost.php?p=4046147&postcount=2

http://i8.photobucket.com/albums/a28/SewerBeing/shrouds-BIX.gif

[batboy]

Less than 1/2 degree difference from water block in vs. water block out? I'd like to know what the system specs were and level of overclocking. I suspect those are idle temps. Regardless, that's still a bit more than ambient... unless your room temp is like 85 degrees F.

[veryhumid]

I think the fact that there is a Watt calculation means that the processor is loaded? If so, the temps are far closer to ambient than they are to CPU temp. But who knows how long the loop was running...

[speed bump]

Batboy just for you I decided to do some water temp measurements.

The loop I tested was as follows.

Computer: one in sig.
Cooling loop actual: Eheim 1250>Swiftech MCW50-T@7v>DD Maze 4>temp measure point>Dtek JR-120 radiator with Nidec Beta V fan@7v. All tubings is 1/2".

To test I ran ATItool artifact scanning becuase in my experiance it loads up my system just like a game does. As opposed to something like superpi or prime that just hits the CPU. Besides with the tec my GPU blocks dumps in most of the heat anyway. I ran this overnight.

Temps
Ambient was 18°c
Rad air was 19°c
water was 20°c
CPU temp was 39°c
GPU temp was bugged below 25°c.
Sys temp was 27°c
NB temp measured with a temp probe 30°c

[batboy]

I appreciate you taking the time to run those tests. How did you take the water temp? Did you actually have the probe in the tube? I loaned my thermometer probe to a friend and never got it back. Now he don't know what happened to it. These two tests still aren't loading the system like a toasty Intel Prescott or Presler under full load, but I'm willing to admit water temp is probably a little cooler than what I had originally thought.

[Sneaky]

i was gonna WC my chipset but picked up a NB-1C from the classifieds, and its doing its job quite well - there was no way tubing would allow me to route a chipset block as cleanly as i wanted


i'd say just pick up one of the new swiftech copper NB heatsinks w/ a 40mm fan on it

[Stratus_ss]

on a related note I am going to be watercooling an overclocked p5wd2-P should I get a chipset block in the loop or will the passive heatsink be enough?

[speed bump]

Batboy on my 550@4.3ghz 1.55v with an x850xt pe in the loop my water temps were ussually about 20°c but that was without a radiator and just a bucket of water with a fan on it for cooling. Once that water hit about 23+°c I knew it though becuase the CPU would go from an idle of 38°c to an idle of like 45°c and load temps aproaching 60°c.

@tratus_ss I mounted a chipset block on my p5wd2-p and it worked well but for what I was doing and unless your going for absolute max FSB I wouldn't worry to much about it on the i955 chipset it seems much cooler than my 875 or 915 ever were as well as anything I have used with the name AMD on it. Lapping the stock cooler using AS5 and putting a fan over it for me was only a loss of 2mhz FSB on my overclock compared to using 1°c water.

[Lancelot]

My system is on 24/7 full load running F@H. I once came home to find my chipset waterblock fallen off onto my AGP card. I was actually checking my emails and stuff before I noticed! One of the lugs was torn from the mainboard and the system was running absolutely fine with an exposed i875P chipset!!! After that I simply mounted a passive heatsink on the chipset and didn't even bother with my current setup in my sig...

[Javascripterror]

I think the chipset blocks are very handy for the hotter boards like the DFi types. The DFI ultra D here sits with a 48 degrees chipset temperature even on days with an ambient of 25degrees.

[veryhumid]

is that one of the NF4 boards? I heard those are quite hot.

[Soong]

I didn’t read the entire thread but I’ll offer this; I have a Pentium D 2.8GHz@3.5GHz on an Intel 955 chipset. I discovered to my great horror that the north bridge chip’s heat sink was over 100 C when running passively. I have no idea why Gigabyte would even provide software to turn it off if it can’t cool passively. When the little atrophied 40mm fan going it can still go over 70C (keep in mind this is the heat sink not even the chip). Unfortunately, the 40mm fan is going bad in a hurry and is currently louder than all 6 120mm fan in the case and on the radiator.

The 955 chipset is still pretty new and this (http://www.polarflo.com/index.asp?PageAction=VIEWPROD&ProdID=119) thing (http://www.polarflo.com/index.asp?PageAction=VIEWPROD&ProdID=58) is the only water block I have been able to locate that will fit the expansive 2.75 inch, center-to-center mounting distance for this north bridge chip.

[Rev]

NF4's are very toasty from what I read and WCing them may provide a gain (then again maybe not as most good AMD boards can go much farther than a CPU anyway). As for an 875, I think you've picked the right choice in going with air. Simple WC loops are best.