H20 cooling question HD6990 and 2700K

[Cisco Kid]

Okay looking for some quick feedback, planning to order ,my gear tonight

I am installing one of the 3 combos below into a Coolermaster 690 Advanced, I am running a XFX 850 Black Edition PSU. I want to cool my 2700K and Sapphire HD6990 4GB. I am running a GB P67-UD7 board. Primary concern, good temps and low noise.

Option 1
- Swiftech H2O-220-Edge-HD-WT
(heat exchanger rad w integrated Res, MCP35X PWN pump, Apogee HD cpu block)
- http://www.swiftech.com/h2o-x20-edge-hd.aspx#tab2
- Swiftech MCR 220-quiet Power rad w res
- XSPZ Razor 6990 Block (have already)
- requires some fittings and minimal tubing

Option 2 - Dual pumps
- 2 x Swiftech MCR220-DRIVE-Rev 2 (integrated rad, res, MCP35X pump PWN)
- XSPZ Razor 6990 Block (have already)
-requires fittings, tubing, liquid

Option 3
-Swiftech MCR220-DRIVE-Rev 2 (integrated rad, res, MCP35X pump PWN)
- http://www.swiftech.com/mcr-x20-drive-rev2.aspx
-Swiftech MCR 220-Quiet Power rad w res
-XSPZ Razor 6990 Block (have already)
-Requires Apogee HD block, tubing, fittings, liquid



My question is option 1 or 2 enough to cool my components, I will be happy with a 2700K overclock in the 4.5 range. The 6990 will majority of time be run on switch 1 setting.

I like Option 3 for the fact, lower noise due to 2 pumps and ability to run entire system at lower rpm, redundancy of having a second pump.

Based on what I want to do what would you choose. Fans to be used, Lower rad will have GT 14's , Upper Rad Cougars

[Karnor00]

I'm a little puzzled as to the differences between the three options. It's not clear to me whether the links are simply to the item listed above, or whether these are an additional item.

Assuming they are just links, as far as I can tell each option has 2x120.2 radiators. Options 1 and 3 have a single pump, with option 2 having a dual pump (I'm assuming that your comment at the bottom was intended to refer to option 2).

In terms of heat dissipation, the additional pump isn't really going to make any difference so given that all 3 options have the same amount & type of radiators, they should all perform about the same. So the next question is whether there is enough performance to meet your targets of good temperatures and low noise.

If my calculations are correct, an overclocked 2700k will generate about 200W of heat and a 6990 will generate about 500W of heat. So you are looking at dissipating 700W of power over a 120.4 radiator. Which is equivalent to 525W over a 120.3 radiator, with fans running at about 1400RPM (GT-14s)

Looking at the Swiftech MRC radiator on http://skinneelabs.com/triple-radiator-comparison-v2/4/ a 1400RPM fan will dissipate 472W with a 10C delta and 708W with a 15C delta. Suggesting you'd be at about a 11C delta. Which would be okay (tho not great) for the GPU, but a little on the warm side for the CPU.

Noise is pretty subjective. 1400RPM fans wouldn't be silent, but shouldn't be overly noisy either.

[Cisco Kid]

Originally Posted by Karnor00

I'm a little puzzled as to the differences between the three options. It's not clear to me whether the links are simply to the item listed above, or whether these are an additional item.

Assuming they are just links, as far as I can tell each option has 2x120.2 radiators. Options 1 and 3 have a single pump, with option 2 having a dual pump (I'm assuming that your comment at the bottom was intended to refer to option 2).

In terms of heat dissipation, the additional pump isn't really going to make any difference so given that all 3 options have the same amount & type of radiators, they should all perform about the same. So the next question is whether there is enough performance to meet your targets of good temperatures and low noise.

If my calculations are correct, an overclocked 2700k will generate about 200W of heat and a 6990 will generate about 500W of heat. So you are looking at dissipating 700W of power over a 120.4 radiator. Which is equivalent to 525W over a 120.3 radiator, with fans running at about 1400RPM (GT-14s)

Looking at the Swiftech MRC radiator on http://skinneelabs.com/triple-radiator-comparison-v2/4/ a 1400RPM fan will dissipate 472W with a 10C delta and 708W with a 15C delta. Suggesting you'd be at about a 11C delta. Which would be okay (tho not great) for the GPU, but a little on the warm side for the CPU.

Noise is pretty subjective. 1400RPM fans wouldn't be silent, but shouldn't be overly noisy either.

Yes each option has 120.4 rad, the links where meant to reference the choice of the all in one rads I am considering. I would say the 6990 puts out max of about 400-450w but nothing wrong with being a bit on the high side to be safe....

I guess I could add a 120 single rad into the loop for further cooling. Appreciate your feedback.

[musicfan]

Nice job Karnor00!

And, if you look at other data on the MCR rads by Swiftech, here is Martinm210's latest testing. The 120.3 MCR removes 180 Watts at 1400 rpm with a delta T of 10°C in his lab.

In Skinnee's testing the flow was 1.5 gpm, pretty hard to achieve without a loud pump or a couple of pumps. Martin's allowed his flow to fluctuate as in real life. The Swiftech rads are slightly more restrictive than a typical rad and benefit from higher pumping action (at a risk of a noisy pump). Martin used insulation and Skinnee did not.

The bottom line is if you purchase something where you are limited by heat dissipation, you will just cut back your overclocks until all is well. The Ivy Bridge is typically cool at TDP (77 Watts) but really cranks out the heat with OC, particularly around 4.5 GHz.

You could go with 2 pumps but if you could fit a 120.3 instead of 120.2 or switch to a different rad, you would be able to dissipate more heat.

Good luck. Please ask any questions you wish.

[Cisco Kid]

Originally Posted by musicfan

Nice job Karnor00!

And, if you look at other data on the MCR rads by Swiftech, here is Martinm210's latest testing. The 120.3 MCR removes 180 Watts at 1400 rpm with a delta T of 10°C in his lab.

In Skinnee's testing the flow was 1.5 gpm, pretty hard to achieve without a loud pump or a couple of pumps. Martin's allowed his flow to fluctuate as in real life. The Swiftech rads are slightly more restrictive than a typical rad and benefit from higher pumping action (at a risk of a noisy pump). Martin used insulation and Skinnee did not.

The bottom line is if you purchase something where you are limited by heat dissipation, you will just cut back your overclocks until all is well. The Ivy Bridge is typically cool at TDP (77 Watts) but really cranks out the heat with OC, particularly around 4.5 GHz.

You could go with 2 pumps but if you could fit a 120.3 instead of 120.2 or switch to a different rad, you would be able to dissipate more heat.

Good luck. Please ask any questions you wish.

I appreciate the feedback on the rad shootout link. Wished I had seen this prior, but all is good. Think what I have pieced together to start is fairly decent. I am using a 2700K Sandy Bridge chip, heat will probably be similar to the IB at higher clocks. I will be happy with 4-4.3 when needed. I am sure the chip will do more but my primary concern is keeping the 6990 at decent temps, most times it will be performing on default switch 1.

I guess I have 3 options once I have this all set up and going.

1) Add an additional Swiftech 120 QP Res rad
OR
2) Add another Swiftech MCP35X pump
OR
3) Add both the above

I am planning to place the rad with the integrated pump on the bottom of my 690 Advanced case with 2 GT14's or 15 on the bottom pushing air through the rad. At the front of the case will be a 140mm or 120 drawing cool air in, at the rear of the case will be a 120 exhausting. The second rad will be placed at top of the case with 2 fans on top with the case cover over them.

Does the above make sense? In a 690 advanced there are really 2 positions for each rad, at the top and on the bottom of the case in front of the bottom mounted psu. I want to retail at least 3 drive bays for 2 SSD's in raid 0 in one bay and 2 640 WD Blacks in raid 0 in the other 2 bays. I need an optical drive & card reader access as well.

I would like some feedback on quality fittings based on what I am wanting to do or suggestions how I may improve what I am doing and if 1 of the 3 options mentioned above are worth adding right off the bat.

Thanks

[musicfan]

Thanks for correcting me on the CPU model. I just built an IB and it's on my mind.

I can't get too dogmatic on your options because there is more than one approach. If using distilled, it's usually cheap to drain and add more gear once you know exactly what you want. It would be fine to start at stock and run load tests to see whether our estimates are accurate. There are many variables (ambient temps for example) and you may be satisfied with the stock CPU speed.

Perhaps get the GT AP15's would be better than the AP14's since you will likely have them under-volted most of the time with a controller to an acceptable level. We know they do well around 1000 rpm when you want quiet. Then when you want the extra ventilation, you have it. If I had to choose, I would go fans first then add more rad next since you have one 35x already and the pump has much headroom.

Your plan makes sense. But you have hot hardware. Some things like an optical drive can work well on a usb port. And Silverstone makes an SSD tray for a drive bay that holds two SSD on one tray to save space if you need that (saw it at Amazon or Newegg).

Quality fittings? Can you be more specific please? Barbs? Compressions? Rotary? What size tubing? The Parts-Buying Guide sticky (also linked in my sig) has a section on fittings with a video and some links if you want to shop a little. Fittings are fun and both barbs and compressions work well. But if one is not careful, it's easy to spend a bundle on fittings.

[Cisco Kid]

Originally Posted by musicfan

Thanks for correcting me on the CPU model. I just built an IB and it's on my mind.

I can't get too dogmatic on your options because there is more than one approach. If using distilled, it's usually cheap to drain and add more gear once you know exactly what you want. It would be fine to start at stock and run load tests to see whether our estimates are accurate. There are many variables (ambient temps for example) and you may be satisfied with the stock CPU speed.

Perhaps get the GT AP15's would be better than the AP14's since you will likely have them under-volted most of the time with a controller to an acceptable level. We know they do well around 1000 rpm when you want quiet. Then when you want the extra ventilation, you have it. If I had to choose, I would go fans first then add more rad next since you have one 35x already and the pump has much headroom.

Your plan makes sense. But you have hot hardware. Some things like an optical drive can work well on a usb port. And Silverstone makes an SSD tray for a drive bay that holds two SSD on one tray to save space if you need that (saw it at Amazon or Newegg).

Quality fittings? Can you be more specific please? Barbs? Compressions? Rotary? What size tubing? The Parts-Buying Guide sticky (also linked in my sig) has a section on fittings with a video and some links if you want to shop a little. Fittings are fun and both barbs and compressions work well. But if one is not careful, it's easy to spend a bundle on fittings.

Thanks for your suggestions and feedback. I have a double ssd drive bay already and would like to retain my 2 x WD640 drives if possible but I can forgo them and add a third 180GB SSD to system on top of the 2 x 120 in raid 0. My main OS drive is a 180GB RevoDrive.

Is it best to stick with stainless as far as material goes vs coloured or annodized fittings? All I have ordered so far is 2 x Dangerden fat Boy G 1/4 Fittings - 1/2" Barb. I am unsure what is the recommended tubing size I should go with, will have to read some of the stickies. I do know that I will probably need a couple of 45 degree fittings for my bottom mounted rad and I am guessing a couple of 90 degree fittings will also be needed for my top rad. Are barbs and compression the way to go?

This is a bit of a project for me, had my Asus CHIV not crapped out and require an RMA I would still be running it w 1090T and 6950. The rma gave me a reason to go with water as I found the other components on a real deal indivdually.

[MaadDaawg]

If I were you I'd take a look at Switech's HD CPU block coupled with the pump/rad combo that has multiple outlets that allow you to cool everything in parallel. the HD has 1 inlet and 3 outlets, one of which could go to your vid card, then the vid card would feed back to the rad as would the CPU. From what Hvac people tell me, parallel cooling is more efficient than serial cooling which is what most of us do.

I have a 3 loop system which works great and looks cool, but if I were to do it again I think I'd look a the above system first.

Your CPU should reach your OC goal with a good HSF, so your main focus is the card, which would operate fine on air, so....

Whichever way you go, GL

[Cisco Kid]

Originally Posted by MaadDaawg

If I were you I'd take a look at Switech's HD CPU block coupled with the pump/rad combo that has multiple outlets that allow you to cool everything in parallel. the HD has 1 inlet and 3 outlets, one of which could go to your vid card, then the vid card would feed back to the rad as would the CPU. From what Hvac people tell me, parallel cooling is more efficient than serial cooling which is what most of us do.

I have a 3 loop system which works great and looks cool, but if I were to do it again I think I'd look a the above system first.

Your CPU should reach your OC goal with a good HSF, so your main focus is the card, which would operate fine on air, so....

Whichever way you go, GL

I ordered the Apogee HD CPU block w my other parts however for the combo (MCR220-DRIVE rad/res/pump combo) I went with Rev 2 which has 1 outlet and one inlet, Rev 3 was modified adding 2 more inlets to the plenum, and I can not go bigger than dual 120 rad so I figured I would run a total of 120.4.

Serial is what I plan to do based on > rad/res/pump unit to cpu, out cpu, into gpu, out gpu into rad 2, out rad 2 back to rad 1. If my flow rate is to low I can always cut back the overclock and if |I want more there is option of better fans and last adding another identical mcp35x pump.
http://www.swiftech.com/apogeehd.aspx

Make sense, I don't want the video card on air. Now if I was to exchange my MCR220-Drive for a rev 3, how would one incorpoorate the second rad into the loop with a paralell/serial. I can't see how a 120.2 rad alone using the paralell method would properly cool 6990 and 2700K alone

[apw63]

Originally Posted by MaadDaawg

If I were you I'd take a look at Switech's HD CPU block coupled with the pump/rad combo that has multiple outlets that allow you to cool everything in parallel. the HD has 1 inlet and 3 outlets, one of which could go to your vid card, then the vid card would feed back to the rad as would the CPU. From what Hvac people tell me, parallel cooling is more efficient than serial cooling which is what most of us do.

I have a 3 loop system which works great and looks cool, but if I were to do it again I think I'd look a the above system first.

Your CPU should reach your OC goal with a good HSF, so your main focus is the card, which would operate fine on air, so....

Whichever way you go, GL

Parallel cooling in theory is a better method of cooling, if all the components being cooled are identical. Coolant will always take the path of least resistance. IF all components are not identical you have to balance the system. It would be a pain in the *** to balance a PC cooling system. Without balancing the system the component with the least amount of resistance gets the most cooling. Hope you can see the problem with parallel cooling as it applies to PC cooling system. A good example of parallel cooling is GPUs in a SLI or Cross-fire configuration. That is if they are on a separate loop. If the GPUs are not on single dedicated loop it’s only a modified series configuration.

Kid you are getting some great advice and are going to have a super system once built.

Andy

[musicfan]

Originally Posted by Cisco Kid

Thanks for your suggestions and feedback. I have a double ssd drive bay already and would like to retain my 2 x WD640 drives if possible but I can forgo them and add a third 180GB SSD to system on top of the 2 x 120 in raid 0. My main OS drive is a 180GB RevoDrive.

Please forgive me. I am getting confused so am going to number and use a long post. Please check to see if this is what you are saying. Thanks.

1. Cooling: Karnor00 did the math for the delta T around 11°C @1400 rpm. My opinion is your best option is to get the AP15 rather than the AP14. At 1800 rpm your C°/W (thermal resistance) that Karnor00 links to with the Swiftech MCR drops to 0.0172032 C°/W. At 700 Watts over 120.4 that would 175 Watts per 120.1 unit...or about 525 Watts * 0.0172032 C°/W = 9. So you can drop delta T from 11 to 9 with slightly faster fans. If you also add another rad (maybe a Swiftech XP), the 120.5 then would be 140 Watts per 120.1 unit...or about 420 Watts per 120.3. Then 420 Watts * .0172032 C°/W = 7.23, which is a pretty good delta T. Bottom line is you can likely handle the cooling with AP15's and one more rad (preferably a Swiftech XP or one that likes 1800 rpm fans).

2. What if you went dual loops? This would be a situation where you could also improve cooling by going with two or dual loops. This would mean duplicating a loop so one loop is for the CPU and one for the GPU. That means 2 pumps, 2 rads, and a dual res or 2 fill lines or 2 reservoirs. If you did that you would have 500 Watts on one loop and 200 Watts on the other. On your original 120 x 2 that would 250 Watts per 120.1 in your VGA loop or 750 W per 120.3 to use Skinnee's table. The delta T in the VGA side would be 750 W * .0172032 C°/W = 13. That's OK for a VGA loop. On the CPU side would be 200 W per 120.2 or 100W per 120.1. At 300 Watts to use Skinnee's table we would get a delta T = 5.2°C, which is very good. So two loops is an option if you can afford it and fit it. Here is more on dual-loops from Gabe Rouchon of Swiftech. And here is a quote from the start of his analysis:

Gabe Rouchon: Under typical computer use, the above test data [dual loops] suggests as a general rule that users would not benefit from setting up dedicated loops for CPU and GPU. Serializing pumps in the same loop also adds a redundancy factor that dedicated loops cannot provide. With superior reliability and lower temperatures at both CPU and GPU levels, single loops appear to win hands down. [But,] Under extreme computer use, this setup [dual loop] recorded a notable advantage at the CPU temperature level for the dual loop, counterbalanced by the opposite effect at the GPU level.

3. What MaadDaawg discussed is seen in the attached photo from Swiftech of an Apogee HD. If you were going to also cool a chipset, VGA card, and RAM; the idea would be more relevant. However since you are only cooling CPU and VGA card, I don't think using the extra ports from the Apogee HD will make a difference. Read the Swiftech information here.

Originally Posted by Cisco Kid

Is it best to stick with stainless as far as material goes vs coloured or annodized fittings? All I have ordered so far is 2 x Dangerden fat Boy G 1/4 Fittings - 1/2" Barb. I am unsure what is the recommended tubing size I should go with, will have to read some of the stickies. I do know that I will probably need a couple of 45 degree fittings for my bottom mounted rad and I am guessing a couple of 90 degree fittings will also be needed for my top rad. Are barbs and compression the way to go?

This is confusing. Most fittings are brass with a shiny plating like nickel (typical Bitspower or Danger Den). I don't think you want wetted steel and when I hear anodized, I think Aluminum so you don't want that wetted. Some say anodized is OK but it can scratch so I won't use an anodized fill-port for example but some will. That is discussed a little more in the Parts-Buying Guide under corrosion - the section is coolant. Please link to the fittings you are considering. Danger Den Fat Boy 1/2" barbs are great and cheap. I have many. If you get barbs instead of compressions fittings, you have a choice of tubing. You can either get 7/16" Internal Diameter & 5/8" Outer Diameter like Masterkleer, which is cheap but will require clamps. 1/2" ID & 3/4" OD tubing like Primochill Primoflex LRT or Tygon 3603 will also fit easily over your barb and will require clamps but be much bigger. Some choose based on case size with big case getting big tubing. I go in between. The 7/16" ID is harder to get on and off barb (need hot water).

Most choose barbs or compressions. I started with all barb and am gradually introducing compressions over the last few tear-downs. I like barbs and they seem more secure but the convenience of getting tubing on and off compressions is very nice - it speeds me up with tear-down and assembly. Both will work. I have a mix right now. You can always lay out your loop and then see what angled fitting you prefer. It is hard to get it perfect the first time so use a little more tubing for gentle bends to avoid kinds.

Originally Posted by Cisco Kid

This is a bit of a project for me, had my Asus CHIV not crapped out and require an RMA I would still be running it w 1090T and 6950. The rma gave me a reason to go with water as I found the other components on a real deal indivdually.

We are happy to work with you. Take your time. Please enjoy the stickies as I tried to make them fun. Ask any questions you like. Check my math too. Perhaps others will comment if the math hasn't bored anyone. We have no pretensions of omniscience. We all here to learn and have fun.

[Cisco Kid]

All the info provided has been awesome and I really appreciate it and am trying to take it all in ,understand it and ask sensible questions.

One thing I have not mentioned is I would like to keep reservoirs within the rads thus the use of the MCR220-DRIVE-Rev 2 (MCP35X pump,res integrated) and MCR 220 Quiet Power rad/res's only for this build. Therefore any additional rad added to the loop will be the MCR 120 QP, I am limited with what, hardware/drives I need to use within my 690 II Advanced.

I would prefer to put this all together and off the bat be done with it, meaning additional modification would be fan related only. I do not want to add more once it is put together so at this moment what I am trying to determine is whether I should splurge & add an additional MCR 120.1 QP rad now, or add another MCP35X pump or do both?

Where I question myself based on the info provided and what I have ordered and what will arrive to me July 23(money aside sinceI will splurge on this build) is whether an additional MCP35X is worthwhile above its redundancy benefit which is a big positive. 2 rads at reduced rpm will move the same GPM vs 1 running higher rpm's and creating more noise and also requiring fans to be run at a higher speed correct

As far as fittings go 2 Dangerden barbs I purchased are below
http://www.jab-tech.com/dangerden-fb...b-pr-4181.html

I think I will go with 7/16" Internal Diameter & 5/8" OD, but I would also like to introduce some compression fittings for convenience.

Again, thanks for baring with me as I try to solve what i want to do.

[musicfan]

Originally Posted by Cisco Kid

All the info provided has been awesome and I really appreciate it and am trying to take it all in ,understand it and ask sensible questions.

You do well. I was going to welcome you here until I saw you joined EOCF before me.

Originally Posted by Cisco Kid

One thing I have not mentioned is I would like to keep reservoirs within the rads thus the use of the MCR220-DRIVE-Rev 2 (MCP35X pump,res integrated) and MCR 220 Quiet Power rad/res's only for this build. Therefore any additional rad added to the loop will be the MCR 120 QP, I am limited with what, hardware/drives I need to use within my 690 II Advanced.

I would prefer to put this all together and off the bat be done with it, meaning additional modification would be fan related only. I do not want to add more once it is put together so at this moment what I am trying to determine is whether I should splurge & add an additional MCR 120.1 QP rad now, or add another MCP35X pump or do both?

Where I question myself based on the info provided and what I have ordered and what will arrive to me July 23(money aside sinceI will splurge on this build) is whether an additional MCP35X is worthwhile above its redundancy benefit which is a big positive. 2 rads at reduced rpm will move the same GPM vs 1 running higher rpm's and creating more noise and also requiring fans to be run at a higher speed correct

That's mostly a budget question to me. You should put the MCR 120.1 QP in now then and run a single loop. I am not sure that adding a second 35x will improve temps. But as you say, it will provide redundancy (I have 2 pumps so can't be a hypocrite) and perhaps enough extra pumping power that your pumps run quieter.

Originally Posted by Cisco Kid

As far as fittings go 2 Dangerden barbs I purchased are below
http://www.jab-tech.com/dangerden-fb...b-pr-4181.html

I think I will go with 7/16" Internal Diameter & 5/8" OD, but I would also like to introduce some compression fittings for convenience.

Again, thanks for baring with me as I try to solve what i want to do.

$2.50 for a DD fatboy barb is fair since DD charges the same. If you are ordering from Jab-tech, they have Masterkleer or Primochill. I have used both and do not think one is better than the other. MK is a little stiffer and clearer but they both work fine. I am not a fan of Tygon 3603 since the plasticizer leached very fast (clouded by 6 months). Now that I have tried the "big 3", I may stick to MasterKleer. Automotive worm-gear clamps work great and are often cheaper at an auto parts store.

Keep asking questions. That's how we all learn - me too.

[Karnor00]

Originally Posted by Cisco Kid

One thing I have not mentioned is I would like to keep reservoirs within the rads thus the use of the MCR220-DRIVE-Rev 2 (MCP35X pump,res integrated) and MCR 220 Quiet Power rad/res's only for this build. Therefore any additional rad added to the loop will be the MCR 120 QP, I am limited with what, hardware/drives I need to use within my 690 II Advanced.

Is there a reason you are wanting so many reservoirs in your build? Normally a WC loop would have a single reservoir just before the pump - the purpose of it is to ensure that the pump never runs dry.

I can certainly understand why you want the convenience and space saving benefits of having the pump, reservoir and radiator together in one item. And if you go for the two pump option then it makes sense to get two of these.

But if you aren't going for a second pump then the second radiator (and third if you add one) in the system doesn't need its own reservoir which would give you a lot more flexibility as to which radiators to choose.

One other thing to bear in mind with the combination radiators is that they need to be in particular orientations to work properly : http://www.swiftech.com/images/produ...nsx354x929.jpg Most times you should be able to design your loop to suit this orientation, but it would be a good idea to think through how you will set up the loop before committing to buying anything.

Additional Comment:

Originally Posted by musicfan

And, if you look at other data on the MCR rads by Swiftech, here is Martinm210's latest testing. The 120.3 MCR removes 180 Watts at 1400 rpm with a delta T of 10°C in his lab.

Slightly off topic, but hoping to learn more myself. Why are the Martinslab tests so different to the Skinneelabs tests?

For the MCR 320 with fans at 1400rpm, the Skineelabs tests indicate that they can dissipate 472W at a 10C delta. As you noted, Martinslabs tests indicate that they can dissipate 180W.

That's a massive difference in results, and is replicated across all the radiators. Especially when both sets of results indicate that there really isn't that much difference between different types of radiators (typically only up to 20% difference between the best and worst performers).

Based on my knowledge of what people are actually cooling, the Skineelabs tests seem more accurate. Taking my system for example I'm cooling 500W over 120.9 with slow fans which should be (and is) overkill on radiator and temperatures are very low. However the Martinslab results would suggest that I don't have enough radiator + fan which I know isn't the case.

Given how meticulous Martin is with his testing, I can't imagine he would have gotten things so wrong. So I'm guessing I'm missing some hidden variable between these tests would explain the difference?

[musicfan]

Originally Posted by Karnor00

Slightly off topic, but hoping to learn more myself. Why are the Martinslab tests so different to the Skinneelabs tests?....Given how meticulous Martin is with his testing, I can't imagine he would have gotten things so wrong. So I'm guessing I'm missing some hidden variable between these tests would explain the difference?

This spring when I updated the Part-Buying Guide, I tried to explain. Please see the Discussion section of 6.Radiators since I captured a couple of Martin's quotes when he was asked that. You can tell me if I explained it very well.

Basically Skinnee's V1 test bench was more open so heat could escape and he used a fixed 1.5 gpm which favors restrictive components like Swiftech rads.

Martin's V2 bench used aggressive insulation so his heat source did not escape into air like Skinnee's did. He also used a fixed setting on the pump so the flow rate would vary (like most of us). This penalizes the more restrictive components...like Swiftech rads.

There are other differences too but those are some main ones. Martin invented the V1 test bench for rads, which he described in 2008 in his archive section and Skinnee copied it successfully. They both evolved and Martin then invented his V2 bench, which he described at his present site. The V2 bench is far different than the V1 partly in the way I described. You can read more about the test setup at their sites. All the radiator testing complements each other as long as folks understand how the results were derived.

[Karnor00]

Originally Posted by musicfan

This spring when I updated the Part-Buying Guide, I tried to explain. Please see the Discussion section of 6.Radiators since I captured a couple of Martin's quotes when he was asked that. You can tell me if I explained it very well.

I understand your explanation of the differences in the testing methodology. And I can see that broadly the results are very similar in how they rank different radiators at different fan speeds - a very important metric for deciding which one to use.

But in order to know how much radiator is needed, we need to also know the absolute wattage results, not just the ranking.

As you quoted from Martin, he himself isn't quite sure of what to make of the difference in wattage numbers:

Yeah, I'm still not quite sure what to think of the actual wattage numbers. They generally seem low, but all the rads are being tested the same, so relative performance should be ok. I guess insulating the lines and adding the enclosure to minimize heat losses makes a bigger difference than I thought. I always knew open air rad testing was a problem if there is any changes in air flow in the area of testing. I am probably also getting a touch of restriction from the test bench ports but I did try one test with and without the top cover and got results that were within a few percent so I don't think there is enough restriction to cause any major differences. I think most of it is in the line insulation work and testing all of the points with heat loads that near a 10°C delta.

My understanding of flow rate is that it makes a relatively small difference in cooling, provided you reach a minimum level. And I think this is confirmed by:
http://skinneelabs.com/assets/images..****s-rpm-16.png
http://skinneelabs.com/assets/images..****s-rpm-08.png
http://skinneelabs.com/assets/images..****s-rpm-04.png

These show that changing the LPM from 1.6 right down to 0.4 has a comparatively minor impact on heat dissipation. And I say comparatively because looking at radiators with 1400 fans there is roughly a 50W difference between 1.6LPM to 0.4LPM out of a total dissipation of c.400W. So like a 12.5% difference.

So differences in flow rate aren't going to account for the difference between Martin's results (180W) and Skinnee's results (472W), Which is a difference of 160%! To add to the confusion, looking at the graphs I linked above (which are from Skinnee's 2011 testing), the results are 1400rpm are around 350-400W. Which is pretty different again from the 2009 results of 472W.

I've heard of a rough rule of thumb of 120.1 of radiator per 125W of heat at reasonable fan speeds. Which would tend to indicate that Skinnee's 2009 testing is the best to use in practice.

Apologies for the somewhat long and rambling post (and further thread derailment).

[Cisco Kid]

Awesome info guys I am reading up on it in spare time. I will be buying a Lamptron FC5V2 5.25" 4 channel controller for my rad fans.

I read the Swiftech MCR 220 QP rads perform best w fan speeds of 1200 rpm and under so I decided on 2 Cougar Turbine 120MM Hyper Spin bearing Fan 60.4/102.6 CFM 1200rpm max for the bottom MCR220 QP rad. For the top rad I have decided on 2 x Cougar Vortex CF-V12HPB 120 mm fans at 70.5 / 119.8 1500 rpm.

I have a question on fan direction for the 2 rads as I will be using w a pair of fans on each. For additional case ventilation I will be using 2 Cougar 140mm Hyper Spin Bearing Fans at 1000rpm 64.5/109.6 CFM 17.9DB. One at front intake bringing cooler air in & 1 on side of case doing the same.

For the bottom mounted MCR 220QP I plan to install the fans on top of the rad in push mode and for the top pump/res/rad I was planning to have the fans on top under the 690 Advanced top case pushing cooler air into/over the rad. I am willing to give the Cougars fans a try for both rads to start and see where my temps end up

[musicfan]

Originally Posted by Karnor00

I understand your explanation of the differences in the testing methodology. And I can see that broadly the results are very similar in how they rank different radiators at different fan speeds - a very important metric for deciding which one to use.

Might be the most important metric since we are cross-comparing data from different labs.

Originally Posted by Karnor00

But in order to know how much radiator is needed, we need to also know the absolute wattage results, not just the ranking.

Agree it would be nice but not sure it's possible. The insulation for example makes it difficult if not impossible to compare absolute Watts. It's hard to say whether Skinnee's data is optimistic or Martin's pessimistic.

Originally Posted by Karnor00

As you quoted from Martin, he himself isn't quite sure of what to make of the difference in wattage numbers.

My understanding of flow rate is that it makes a relatively small difference in cooling, provided you reach a minimum level. And I think this is confirmed by...[skinnee lab images]

These show that changing the LPM from 1.6 right down to 0.4 has a comparatively minor impact on heat dissipation. And I say comparatively because looking at radiators with 1400 fans there is roughly a 50W difference between 1.6LPM to 0.4LPM out of a total dissipation of c.400W. So like a 12.5% difference.

So differences in flow rate aren't going to account for the difference between Martin's results (180W) and Skinnee's results (472W), Which is a difference of 160%! To add to the confusion, looking at the graphs I linked above (which are from Skinnee's 2011 testing), the results are 1400rpm are around 350-400W. Which is pretty different again from the 2009 results of 472W.

I've heard of a rough rule of thumb of 120.1 of radiator per 125W of heat at reasonable fan speeds. Which would tend to indicate that Skinnee's 2009 testing is the best to use in practice.

Apologies for the somewhat long and rambling post (and further thread derailment).

No apologies necessary. Your thoughtful posts are relevant to the OP choosing Swiftech rads. I looked at the charts you linked to from Skinnee Labs and picked the MCR-QP since that is most on-topic. Those charts are hard to see so forgive my estimate. I also use a different denominator to describe "improvement". I see the 1400 rpm QP at .4 gpm cool about 310W and at 1.6 gpm about 390W. 80/310 = 26%.

With radiator fans, Skinnee used Gentle Typhoon AP15's and Martin used an average fan, the Titan Kukri. The other Martin quote discusses how the AP15 delivers about 400 more cfm per rpm than other fans tested on a rad. The link is here and the chart is attached.

If we are at 1.6 gpm, then going from 1400 to 1850 rpm means going from the above value of 390W to 430W. 40/310 = 13%. So adding the flow and changing brand (not absolute rpm) of fan is almost a 40% improvement.

You are correct that 40% improvement in Skinnee's data from flow & fans won't explain it all. But when added to the insulation issue, it starts adding up. But all this uncertainty combines to gets us farther from absolute measurements like Watts.

That's how I reconcile the data. In 2012, I remain flexible and seek an acceptable range of performance relative to overclock and noise and most importantly, relative to what the user is doing with the machine. So even if the delta T is too high to OC the CPU, the noise reduction from tiny GPU fans is the best benefit for some gamers. Most rarely use the system at max load, which is different than Folding so synthetic testing is relative. And much of the heat load is never cooled by our blocks because it escapes to case air. So, system air cooling remains important.

I almost went back to formulae but one size won't fit all. But doing the sticky on All-in-Ones bent my mind. There is an army dedicated to squeezing a 120.1 rad for all it's worth, LOL. And there are some who think us insane to recommend fans we can hear at 12 volts.

Collating radiator old V1 and new V2 data from Skinnee Labs and Martin's Lab with data from Bundymania and HESmelaugh humbles me. The method changes are confusing but I think the data is solid considering the small sample size per test. At any rate, it makes more sense to me than what the manufacturers tell us.

Please ask questions and I will share what I know. This is fun to have a discussion of water-cooling testing.

Originally Posted by Cisco Kid

Awesome info guys I am reading up on it in spare time. I will be buying a Lamptron FC5V2 5.25" 4 channel controller for my rad fans.

I read the Swiftech MCR 220 QP rads perform best w fan speeds of 1200 rpm and under so I decided on 2 Cougar Turbine 120MM Hyper Spin bearing Fan 60.4/102.6 CFM 1200rpm max for the bottom MCR220 QP rad. For the top rad I have decided on 2 x Cougar Vortex CF-V12HPB 120 mm fans at 70.5 / 119.8 1500 rpm.

I have a question on fan direction for the 2 rads as I will be using w a pair of fans on each. For additional case ventilation I will be using 2 Cougar 140mm Hyper Spin Bearing Fans at 1000rpm 64.5/109.6 CFM 17.9DB. One at front intake bringing cooler air in & 1 on side of case doing the same.

For the bottom mounted MCR 220QP I plan to install the fans on top of the rad in push mode and for the top pump/res/rad I was planning to have the fans on top under the 690 Advanced top case pushing cooler air into/over the rad. I am willing to give the Cougars fans a try for both rads to start and see where my temps end up

A 30W per channel Lamptron should be a very nice fan controller - perfect for controlling fast fans. You should get whatever fans you like but you can get the AP15's for a good price now at Sidewinder (or almost at good at Amazon) for the reasons discussed above. They will cool better than the Cougars. Even a low fpi low rpm rad will cool better with faster fans than slower ones. Use that nice new controller - even for case fans. There is no reason for you to run your fans straight into 12 volts with a controller like that. It will likely handle all your fans.

Push seems better especially over about 1200 rpm...or maybe all the time (Martin's Lab). In theory, bringing cooler room air rather than warmer case air over rads should improve temps because it is like a free lowering of ambient temperature. But, the entire cooling of your case needs to be considered. Just set it up most logically and if you get the urge, switch and recheck. I have changed my 1200 rpm fans from top to bottom to push to pull and can't detect a difference except one thing. The farther from my ear (as in behind or below the radiator), the less I hear.

Please keep asking questions until you are satisfied. Good luck.

[mimart7]

With the 30 watt Lamptron controller, you can easily have 2 fans per channel.

[Cisco Kid]

Originally Posted by mimart7

With the 30 watt Lamptron controller, you can easily have 2 fans per channel.

Excellent. Monday all my gear arrives. I am currently riding on motorbike trip down in WA with a buddy. Looking forward to starting this build and will have questions for sure!