GTX360 or PA120.3 ? which 1 to buy ?

[Marci]

As far as the TC sponsorship, not even sure about that, and have my doubts.

There is NO sponsorship deal between ThermoChill and XS. I was a member on XS long before ThermoChill ever existed, and have remained one since ThermoChill was founded.

I have seen those stats and I find it funny that they were made BY Thermochill

ALL testing on ThermoChill rads was done INDEPENDANTLY. The only testing done by myself was comparative of PA Model to PA Model and included NONE of the competitions' rads.

http://www.thermochill.com/PATesting - all tests done independantly by BillAdams of Thermal Management Testing - the guy who DESIGNED the Swiftech MCR series of Radiators, and the Coolingworks radiators... so highly doubt there's any bias going in ThermoChill's direction there.

Bottom line, find your own facts, don't trust what people say here. That's what I did and i'm glad that I did

Obviously the "facts" that you found weren't "facts" in the slightest *shrug*.

[masaville]

Thermochill supports this board and vise versa, so any opinions on who is better will be bias. I have never used the PA120.3 but I have a GTX360 and it gets better temps than any PA120.3 with the same setup (E6600+8800GTX) that I have seen on this board.

Bottom line, find your own facts, don't trust what people say here. That's what I did and i'm glad that I did

( Don't bother flaming, I won't read it )

I have a PA120.3 and a GTX 480. I've tried them both on my E6700 and 8800GTX in SLI. Although we aren't comparing apples to apples here (I don't have a GTX 360) the 480 beats the 120.3 hands down, no contest. If you're cooling a rig with dual 8800s and really want to get your temps down definitely go with 480 as the 120.3 does a mediocre job on that configuration. If you have a single 8800 it will probably be plenty, but not for SLI rigs.

Then again it depends what temps you're looking for. On the aforementioned hardware with no overclock I got temps up to 61c on the GPUs under load and it raised ambient room temps from around 21 to 25/26. On the 480 the gpu load temps would typically remain around 55c. There was an overall increase in CPU and MB temps as well, probably because of the hotter GPUs. Idle temps were a few degrees higher on the 120.3. Having that extra 120 on the end makes a difference with dual 8800s....would like to see a PA120.4 heh.

[IanY]

I have a PA120.3 and a GTX 480. I've tried them both on my E6700 and 8800GTX in SLI. Although we aren't comparing apples to apples here (I don't have a GTX 360) the 480 beats the 120.3 hands down, no contest. If you're cooling a rig with dual 8800s and really want to get your temps down definitely go with 480 as the 120.3 does a mediocre job on that configuration. If you have a single 8800 it will probably be plenty, but not for SLI rigs.

Let me get this straight. You are passing an overclocked Core 2 and two G80s through a single radiator? Good luck.

Incidently, from your empirical point of view, what's the differential between the GTX480 and the PA120.3 ?

[masaville]

Let me get this straight. You are passing an overclocked Core 2 and two G80s through a single radiator? Good luck.

Incidently, from your empirical point of view, what's the differential between the GTX480 and the PA120.3 ?

No overclock - just stock. Now that I have a 480 and a 120.3 I was thinking of getting an extra pump and using both. One for the CPU and one for the GPUs. Is it worth it do that or not really?

Btw, it wasn't empirical. I recorded idle and load temps using the standard apps, but I don't have them with me which is why I can't post them. The point of the post though is just to make a note that a 480 is a better choice for dual 8800s though, not to knock the 120.3, which I'm sure is fine for other configs. If you want lower temps on dual 8800s, especially if you're overclocked, then you really need a 480, two 120.3s, or something along those lines.

[IanY]

No overclock - just stock. Now that I have a 480 and a 120.3 I was thinking of getting an extra pump and using both. One for the CPU and one for the GPUs. Is it worth it do that or not really?

Of course its worth every penny. Putting two G80s together with a cpu on one radiator is not the best of ideas, no matter how good the radiator. You will do well to seperate into two independent loops.

You haven't answered my question though... how much better is the GTX480 compared to the PA120.3 ?

[masaville]

Of course its worth every penny. Putting two G80s together with a cpu on one radiator is not the best of ideas, no matter how good the radiator. You will do well to seperate into two independent loops.

You haven't answered my question though... how much better is the GTX480 compared to the PA120.3 ?

Don't have the 'official' numbers in front of me I can roughly tell you. About 6c on load with the GPUs, 4-5c on the MB, and 4-6c on the CPU. That's based on an ambient room temp around 21c. That's what I discovered this weekend. 480 is quite large though and unless you have a monster case it's difficult to mount, whereas the PA120.3 fits nicely on the top or inside many cases.

[IanY]

Don't have the 'official' numbers in front of me I can roughly tell you. About 6c on load with the GPUs, 4-5c on the MB, and 4-6c on the CPU. That's based on an ambient room temp around 21c. That's what I discovered this weekend. 480 is quite large though and unless you have a monster case it's difficult to mount, whereas the PA120.3 fits nicely on the top or inside many cases.

Thank you for the information. That was all I was after. Honestly, I would much prefer multiple Thermochills rather than relying on a GTX480. I have seen some people mount the GTX480 with a RadBox at the back, but it is quite unsightly.

[Cathar]

Don't have the 'official' numbers in front of me I can roughly tell you. About 6c on load with the GPUs, 4-5c on the MB, and 4-6c on the CPU. That's based on an ambient room temp around 21c.

/me bites tongue

[IanY]

/me bites tongue

Appreciate if you could share your expertise. We can all benefit from your knowledge and insights. Please don't muzzle yourself.

[ramenchef]

IIRC according to Marci the GTX480 starts to outperform the pa120.3 at the 1000-1200rpm mark.

[Cathar]

LOL.

Okay. Here's my thoughts. Make of it what you will.

The only way that devices change in temperature by changing the radiator is through a change in the water temperatures. Flow rates also have an impact, but overall the radiator, especially a Thermochill, is going to be a very minor part of overall flow resistance. Low enough to be negligible in a loop with many blocks - so let's discount any changes in flow rates as being part of the factor, and focus on water temperatures alone.

Said person is reporting what appear to be a 5C delta (average) between the two radiators, corresponding to a 5C water temp delta. If we assume that the per-120.1 sectional cooling performance was equal between the two radiators, then this mathematically that the water temperature on the PA120.3 would have to be at least 20C above ambient.

At 20C above ambient, even with 3 x Nexus fans (38cfm) at 12v, we're talking a total heat load of around 650W required to generate such a large water temp delta, at the very least.

I cannot believe that there is 650W of heat load in such a system. I believe that other factors are at play.

I do not know what. I would be most curious to know the water temp rise above ambient air intake delta is.

[IanY]

Many thanks for the valuable insights.

[snowwie]

but cathar, would temp differences on the cpu, gpu, and nb between systems necessarily be equal to coolant temp difference between same two systems? the thermal resistance of each wb would then make coolant temp diffs less, no? or am I stupid at math.

[Cathar]

but cathar, would temp differences on the cpu, gpu, and nb between systems necessarily be equal to coolant temp difference between same two systems?

??

I thought we were talking about the same system here.

[snowwie]

by diff systems I meant diff rads. I thought if you change the rad (and rad alone) it's a diff cooling system.

edit: so does what I'm saying not make sense?

[Mr. Peanut]

by diff systems I meant diff rads. I thought if you change the rad (and rad alone) it's a diff cooling system.

edit: so does what I'm saying not make sense?

If it means anything, I sure don't understand what you're trying to get across. I'm sure it's a good point, I just don't understand it.

[Cathar]

by diff systems I meant diff rads. I thought if you change the rad (and rad alone) it's a diff cooling system.

edit: so does what I'm saying not make sense?

I already explained, and factored in, the differences in the system with respect to changing the radiator. Re-read what I wrote.

Aside from the minor differences in flow rate, about the only thing that should change is the water temperature, given a constant heat load.

I personally think that something else was occuring in that setup. The differences in water temperature alone simply cannot be that large between those two radiators, with that sort of heat load. Controlled tests in fact show that for weak fans, there'd be little to no difference at all. For stronger fans, maybe some difference, but certainly not 4-6C. Something else is occurring.

[serialk11r]

Serial do you have a PA120.3? Have you tested rads yourself?

No but what I meant was, do you think the barbs affect the rad's performance more than the rad itself? If so then people wouldn't be buying better rads, they would be buying better barbs (which doesn't make that much sense lol).

[phelan1777]

No but what I meant was, do you think the barbs affect the rad's performance more than the rad itself? If so then people wouldn't be buying better rads, they would be buying better barbs (which doesn't make that much sense lol).

I seriously doubt the barbs are affecting the over all performance of the rad.

[serialk11r]

I seriously doubt the barbs are affecting the over all performance of the rad.

Exactly, that guy was saying the testing was biased because it was done with different barbs....

[snowwie]

Cathar, all I was arguing was this point:

Said person is reporting what appear to be a 5C delta (average) between the two radiators, corresponding to a 5C water temp delta.

But now I think I'm wrong, nvm.

[snowwie]

i'd much rather understand WHY cpu temp difference corresponds to equal coolant temp difference than just be told that it's the case, even from cathar. if wc forums have taught ME anything...

[Cathar]

i'd much rather understand WHY cpu temp difference corresponds to equal coolant temp difference than just be told that it's the case, even from cathar. if wc forums have taught ME anything...

Ok, it's like this. You understand the concept of C/W right? For every wattage of heat that the CPU is putting out, the thing being cooled will increase by C temperature.

Ok, so if we've got that down-pat, what is all this relative to? What is doing the cooling? The water-block, right? What is it about the water-block that's doing the cooling? The water inside it. What about the water? It's temperature. If the water is hot (let's say we hooked it up to the hot water service and feeding it a never-ending stream of 60C temperature water), and the CPU is not putting out any heat at all (turned off) what will the CPU temperature be? Ultimately it'll get pretty damn close to the temperature of the water, right?

So we turn the CPU on. What will happen? It'll get hotter, because it's emitting heat. It started off at 60C (being the water temperature), and it'll increase in temperature by an amount proportional to the C/W of the waterblock.

i.e. the water temperature dictates the base-line above which the device's temperature will rise above.

Note: the above doesn't include secondary heat paths, such as the motherboard providing some measure of cooling based upon the ambient air temp, and heat from the CPU conducting across the MB and convecting into the air. For purposes of simplicity, we'll assume that it is a negligible proportion of the total cooling power provided by the waterblock when water temperatures are close(ish) to ambient. In practise, the motherboard generally provides between 1-10% of the total cooling effect on the CPU, depending on how effective the cooling device on the CPU is.

The upshot of all of the above is that if your water temperatures are cooler by 5C, then your devices will be pretty close to 5C cooler as well. In correlation, if your devices are all now 5C cooler than before, and nothing else changes (heat workload from the devices that is) then the water must now be ~5C cooler than before.

[snowwie]

yes, I knew I could count on the most patient of the wc gurus.

[exhausted mule]

i still dont see how you could not attribute an addional 120mm of rad to 5c.

assume for a moment the all mighty god of the gods pa3120 is equall in flowrate reduction and heat-dissipation at a given airflow rate to a gtx 360. could an aditional 120 be equal to that 5c?

did you change anything between the two tests other than the rads massa?