Would a 240mm Rad cool a 2600k and a 6970?

[3dchipset]

I'm curious. Would a 240mm RAD RX240 XSPC cool sufficiently a 2600k with a 6970 or possibly a 6990?

[InCredible]

a lot of variables to say how well, at stock most certainly..over clocked will depend on how far and ambient along with what kind of fans are being used etc.
I think it would be a stretch unless you just want them cooled at stock speeds..

[Conumdrum]

Nope, not one teeny chance.

Something I put together.......

http://www.overclockers.com/guide-deltat-water-cooling/

[3dchipset]

How about just cooling the GPU? I think I can get away with just cooling the CPU with an air heatsink. Then would a 240mm cool a 6970 or 6990 reasonably then?

[TJ TRICHEESE]

a 240 on the gpu would be far better than stock.

[CedricFP]

A single 240 would cool them, but you're not going to see great results.

I used to have an E6600 + 8800GTX on a single MCR220 w/ med speed yates. It cooled the components, but after adding another double, I saw temps drop.
Still better than air, though.

[Conumdrum]

A single 240 would cool them, but you're not going to see great results.

I used to have an E6600 + 8800GTX on a single MCR220 w/ med speed yates. It cooled the components, but after adding another double, I saw temps drop.
Still better than air, though.

E6600 and that GPU is like a 1915 car HP vd todays cars. NO comparison. His rig, the water temp would never stabilize, it would heat up to the point of shutting the CPU down, unless he had some 5000RPM 50mm Deltas on it.

[3dchipset]

Thanks for the advice guys, I appreciate the feedback.

I'll end up using a triple rad, the XPSC RX360

[CedricFP]

E6600 and that GPU is like a 1915 car HP vd todays cars. NO comparison. His rig, the water temp would never stabilize, it would heat up to the point of shutting the CPU down, unless he had some 5000RPM 50mm Deltas on it.

E6600 65w TDP, overclocked to 3.6GHz @ 1.45v.
8800GTX vmodded courtesy of Vince. Can't remember the exact clocks. Stock 8800GTX is around 155w for GPU ALONE. OC probably sees it closer to 200-250 on the GPU. Not sure of total combined incl. RAM, circuitry etc.

I don't know, but a 2600K at stock + 6970 at stock, I don't see temps climbing until shutdown on a 2x120 w/ med speed yates (so around 1400rpm fans).

6990 and the game changes, and of course if he's aiming for 50x100 @ 1.5v. But nowhere is that specified.

[InCredible]

like said i think itll work at stock, i used a mcr320 on a i7 920 at 4.5ghz with ht with 1.5v and a 4870x2 vmodded running at 850core no issues at all and my house gets hot in the summer(florida) id imagine these two quite a bit lower on heat output at stock and would suffice with a dual rad at stock settings with a decent ambient.

[3dchipset]

Yeah, the CPU would be stock as this would be on an itx board H67. The GPU would be the only thing pushed moderately. I was looking at a 6990 on an itx board, but it looks like I might go with a MicroATX board with a possible 6970 Crossfire with the 2600k at 4Ghz or possibly 4.5Ghz. I was thinking a XPSC 360 rad. I had a quad rad with a GTX 580 pushed to 900mhz core and a i7 980x pushed to 4.4GHz and temps were a joke on a quad. So I was assuming a 360 rad would handle a sandy bridge to hit 4Ghz and the majority of the heat coming from a 6990 or 6970 Crossfire...

[jayhall0315]

Yes, I have the exact setup your asking about on one of the clients in the lab. At stock (lab ambient is about 27 C), I have 39 C on the 2600 K and 49 C on the 6970 core. Overclocked to 4.5 GHz on the CPU and 950 to 980 on the GPU core, I get 44 C (59 C load with Matlab and 72 C with Prime 95) for the 2600 K and average 60 to 62 C on the 6970 for gpu computation (more rigorous than Crysis or other gaming needs). The rad is a Thermochill PA 120.2 with two S-Flex series F fans on it running at about 1300 rpms.

[gmat]

Idea is to get more cooling surface area than what you can get on stock air cooling, as a rule of thumb use 1x120mm for CPU and 2x90mm for GPU, so if you want both that's a triple rad to have some comfort. A double will work but will need fresh air, and fast (=noisy) fans, not too pleasing as a result. GPU produces the most heat indeed, and needs a double radiator for itself (at least) if you want a lower noise level than stock.

[CedricFP]

Yes, I have the exact setup your asking about on one of the clients in the lab. At stock (lab ambient is about 27 C), I have 39 C on the 2600 K and 49 C on the 6970 core. Overclocked to 4.5 GHz on the CPU and 950 to 980 on the GPU core, I get 44 C (59 C load with Matlab and 72 C with Prime 95) for the 2600 K and average 60 to 62 C on the 6970 for CUDA computation (more rigorous than Crysis or other gaming needs). The rad is a Thermochill PA 120.2 with two S-Flex series F fans on it running at about 1300 rpms.

Do you mean you have this on a single 2x120 radiator? Because if so:

Nope, not one teeny chance[...]unless he had some 5000RPM 50mm Deltas on it.

[jkr266]

I have a I7-870@4.1, and a 5870@1050/1300 being cooled by a MCW60, a apogee xt, and old by todays standards ddc pump and a PA120.2. My temps aren't great but they are a good bit better than what the system was at stock speeds and cooling.

[aerial]

Temps will not be great for cpu (for gpu it doesn't really matter), but it is possible. Optimal solution for gpu+cpu is 360 radiator, so this isn't really that far from optimal. More than 360 makes sense for more than one gpu.

[subtec]

Nope, not one teeny chance.

Something I put together.......

http://www.overclockers.com/guide-deltat-water-cooling/

I disagree with you. For one thing, an i7 @ 4GHz outputs 250w? Really? Xbitlabs measured 190w at 4.7GHz on a 2600k, and if we interpolate their results, just ~115w [edit: sorry, wrong chart - more like ~130w] at 4.0GHz.

So I don't know where you got your 250w number from. Are you measuring from the wall? That's probably your problem: you're not taking into account that 15-20% of the power used is lost through your PSU, or as much as 50w. And then there's the rest of the stuff drawing power on the computer, which can easily add another 50-75w.

So let's do this right.

I'm going to high-ball OP's setup, in terms of power consumption/heat output.

i7 2600k @ 4.7GHz: 190w (from xbit, 100% load)
HD6990 @ stock: 352w (from here; fully loaded, you'd only get this under e.g. furmark)

Total: 542w

Radiator performance numbers courtesy of skinnee; since that's for the 360 version, I'll be dividing by 2/3rds to get approximate numbers for the 240.

Looking at the red line (1400rpm yates) on skinnee's graph, we see that 542w is dissipated at a deltaT of about 11.5C. Dividing by 2/3rds gets us 17.25c for the 240 rad.

So 17.25c is your deltaT under fully loaded stress test conditions with 1400rpm fans.

Is this bad? Well, it's not great. Skinnee, in his testing, refers to a 15c delta as "low performance, an overloaded but capable loop." So a couple degrees beyond that is still in the same ballpark IMO. And remember this is worst-case stress testing - 100% on both OC'd CPU and GPU, something you're never going to see under normal usage. I'd expect a significantly lower delta under more typical conditions.

Consider also that the Corsair/Asetek/Coolit etc. sealed liquid coolers are designed to run a delta of over 20c, and people are pretty happy with those (you can see Asetek's C/W numbers on their site).

"Not one teeny chance"? Hardly.


Edit: I suppose there's some question whether a 17c delta will hold an i7 at 4.7GHz, and for that I would just point to the results people are having with the sealed liquid coolers. It's a moot point though, as I just saw that OP has no intention of OC'ing the CPU. OP, you'll be fine.

[jayhall0315]

I agree with your back of the hand calculation Subtec. A 240mm rad works just fine as long as you are happy with slightly high numbers. As long as you dont regularly ride that 2600K above 62 C or so, you will be fine for several years (on average). Far longer than the average upgrade cycle for guys like us. No disrespect Conundrum but disagree with you on several fronts but too busy this morning to post all the reasons.

[Conumdrum]

I do agree my TDP on the CPU is high. Times have changed since it was published. I do say it's what I used, and you should work it out yourself. It's over a year old and was written long before the publish BTW.

I also agree that under normal usage it will be less, and Gabes statement is correct about a single loop vs a double. Unless your a cruncher where both will be maxxed out. Or you want a very quiet system. Or it's a server in a closet. Or it's a HTPC. Too many variables.

The DT article points people in the right direction and teaches them how to use Skinnees charts and what the general guidelines for DT temps should be on a CPU vs GPU etc Not an end-all-do-it-this-way article.

Yea, maybe a 120x2 will work, but not optimal by any accounts. Up to the OP I guess.

[zanzabar]

I disagree with you. For one thing, an i7 @ 4GHz outputs 250w? Really? Xbitlabs measured 190w at 4.7GHz on a 2600k, and if we interpolate their results, just ~115w [edit: sorry, wrong chart - more like ~130w] at 4.0GHz.

So I don't know where you got your 250w number from. Are you measuring from the wall? That's probably your problem: you're not taking into account that 15-20% of the power used is lost through your PSU, or as much as 50w. And then there's the rest of the stuff drawing power on the computer, which can easily add another 50-75w.

So let's do this right.

I'm going to high-ball OP's setup, in terms of power consumption/heat output.

i7 2600k @ 4.7GHz: 190w (from xbit, 100% load)
HD6990 @ stock: 352w (from here; fully loaded, you'd only get this under e.g. furmark)

Total: 542w

Radiator performance numbers courtesy of skinnee; since that's for the 360 version, I'll be dividing by 2/3rds to get approximate numbers for the 240.

Looking at the red line (1400rpm yates) on skinnee's graph, we see that 542w is dissipated at a deltaT of about 11.5C. Dividing by 2/3rds gets us 17.25c for the 240 rad.

So 17.25c is your deltaT under fully loaded stress test conditions with 1400rpm fans.

Is this bad? Well, it's not great. Skinnee, in his testing, refers to a 15c delta as "low performance, an overloaded but capable loop." So a couple degrees beyond that is still in the same ballpark IMO. And remember this is worst-case stress testing - 100% on both OC'd CPU and GPU, something you're never going to see under normal usage. I'd expect a significantly lower delta under more typical conditions.

Consider also that the Corsair/Asetek/Coolit etc. sealed liquid coolers are designed to run a delta of over 20c, and people are pretty happy with those (you can see Asetek's C/W numbers on their site).

"Not one teeny chance"? Hardly.


Edit: I suppose there's some question whether a 17c delta will hold an i7 at 4.7GHz, and for that I would just point to the results people are having with the sealed liquid coolers. It's a moot point though, as I just saw that OP has no intention of OC'ing the CPU. OP, you'll be fine.

your link has it at 254W for the 6970, im not sure where u go your 352W unless u found a number with from the wall that would include more than the card, amd rates their TDP as the max that the card will do stock (sometimes its off +-5-10W but thats normal.) if u look at NV they rate the gpu under max game load, so u need to test for wattage there but amd is clean, its like the cpus were amd is clean and intel rates at max normal load.

i have 1200rpm fans and i run a pa120.2 for my liquid, when i was on an 720be and a 4890 it ran fine but i could not overclock both and only little overclocking, when i changed to a 1090 i had the same problem (thats about 300W overclocked into the liquid, i use gpu only so its about 125W measured from the voltera pwm, the card is about 230-240W.) this year i added a single fan black ice gtx with ~1200rpm fan, it now overclocks both the cpu and gpu and keeps good temps (around 50c) with the gpu pulling 160W (gpu only) and about 200W from the cpu (estimated) were with the 720be and 4890 gpu only on the pa120.2 i was at 60-65c with the cpu at 3.8ghz (about 150-160W on the cpu.) so a 6970 and a 2600 will not work well with ether overclocked on a dual fan and if u cannot get over 4.4ghz with a 2600k it kinda defeats the purpose so u need 3 fans atleast if not 4 (if you want to bench it.)

[subtec]

your link has it at 254W for the 6970, im not sure where u go your 352W

I was using the 6990 power draw, not the 6970. OP is considering the 6990, which is why I went with that card (and to give a high ball estimate).

[zanzabar]

I was using the 6990 power draw, not the 6970. OP is considering the 6990, which is why I went with that card (and to give a high ball estimate).

oh nice call then, sorry i missed it.

2 gpus and a cpu will not work with a 240 rad, so i guess we are in perfect agreement.

[subtec]

oh nice call then, sorry i missed it.

2 gpus and a cpu will not work with a 240 rad, so i guess we are in perfect agreement.

Actually, it appears we are not. I mean, I ran the numbers (re-read my post if you need to) and going by those, it should work. Maybe not a lot better than air cooling, and maybe without much extra headroom for overclocking, but otherwise, the numbers suggest it should work.

Have you seen the PNY GTX580 GPU/CPU sealed liquid cooler? Is it so hard to imagine that if a single 120mm rad is capable of dissipating the heat from a GTX580 and a modern CPU, a dual rad could manage a CPU + two GPUs?

[glide 1]

This is my system - very similar to what the op asked about.

http://www.xtremesystems.org/forums/...allery/page192 - post #4799

single loop
cpu and gpu blocks
daul 120 rad
2x1800rpm fans

Just tried running furmark (1680-1050) and prime95 at the same time. 2600k is at 4.5ghz 1.20v and 6950 is at 840 stock voltage on the core.

this is what i get:

cpu 52-56deg
gpu 60-64deg
ambient is 26deg

idle temps are:

cpu 32-38deg
gpu 40deg
ambient 26deg

so far the system is ok, but i get a feeling once i go for higher clocks, temps will get out of hand.

[3dchipset]

I thank everyone for contributing to the thread. Great information. I settled on a 3x120 RAD setup. I really digged Glide 1's setup as it's very similiar to what I wanted to do. His temps are perfectly in line for what I'm achieving. :-)