Please Explain These 2 Scenarios

[Shadowman]

*PLEASE NOTE THE CPU, MOTHERBOARD, CPU WATER BLOCK  & PC CASE stayed the same in 2018 and 2019 setups.  Only things that changed, removed or added are listed below. Please carefully see both setups.*

 

 

2018 SETUP CPU ONLY WATER COOLED
 

i9 7900x delidded  Overclocked to 4.7Ghz at 1.18V (Using Der8auer direct die frame NO IHS)

Black Ice Gtx 360 radiator (6 fans in push/pull fan configuration)

EK Xtop Dual D5 2 pumps in series (Flow rate at 2 gpm)

*Soft tubing 1/2 id & 3/4 od 

 

Room temp at 31c and Water temp 40c running Aida64 for about 10 minutes results below

 

 

2019 SETUP CPU PLUS GPU WATER COOLED

i9 7900x same cpu as above delidded same overclock and voltage, but put back the original IHS using liquid metal 

Nvida RTX 2080 Ti (Using Watercool Heatkiller gpu water block)

Black ice Gtx 360 radiator (3x fans in push configuration)

Black ice Gtx 480 radiator (4x fans in pull configuration)

Bitspower reservoir and pump combo, single pump (Flow rate 1.1 gpm)

*Hard line tubing 12mm id & 16mm od 

 

Room temp at 27c and Water temp at 35c running Aida64 for about 11 minutes

 

 

 

 

 

Sorry that was long, know the 2019 setup I did as you can see from above has much higher cpu temps by 10c (even the room temp in 2019 was lower by 4c then in 2018)  the cpu overclock and cpu voltage stayed the same. I know a lot of things changed from 2018 to 2019 setup but I was expecting maybe same or even lower cpu temps I mean adding a gpu with a 480 radiator didn’t help.
 

Am I right in thinking things like going from a soft tubing to hard tubing (with a lot of 90 degree bends),  from a dual d5 pumps to a single d5 pump dropping the flow rate from 2 gpm to 1.1 gpm having a 360 radiator only but in push/pull configuration and finally going from direct die cooling to putting back the original IHS on all played a major factor?  that if I had actually kept everything as same as the 2018 setup and ONLY added my GPU and 480 radiator into the water cooling loop I would have definitely seen better temps?

 

 

 

[AngryBeaver]

3 hours ago, Shadowman said:

<snip>

Temps being better would depend on several things like fan speed etc... you could just go for the same level of performance with lower rpms on the fans.

 

Now the biggest reason for your CPU is the IHS. I mean 10c higher is a decent bit... so it probably comes down to the application of the liquid metal and the paste on top of the IHS. Normally bare die is a little cooler, but 10C cooler is a little much.

 

I mean your did go from a deltaT of 9c to 8c on the setups so if the cpu had been left bare it would have dropped by the 1c at any given load. 

 

The coolant flow shouldn't really make a huge difference. You just need enough to create turbulence and also move the water away from the blocks fast enough the heat doesn't pool. That being said if you average the amount of time the water spends in the rads and blocks at 1.1 and 2.0 gpm it would come out to be the same amount of time.

 

TLDR: The only real contributor is going from direct die to an IHS and the gap seems larger than it should be. I think the LM application might have been lacking.

Edited December 8, 2019 by Spotty
snip quote

[cole0622]

I think a better flow rate would help as the slower it moves the more it sits there getting hotter. where the water is moving faster no it doesn't heat up as much its easier to dissipate the heat .2 gph or higher would probably be good. also I agree direct die would be better

[Shadowman]

11 minutes ago, AngryBeaver said:

Temps being better would depend on several things like fan speed etc... you could just go for the same level of performance with lower rpms on the fans.

 

Now the biggest reason for your CPU is the IHS. I mean 10c higher is a decent bit... so it probably comes down to the application of the liquid metal and the paste on top of the IHS. Normally bare die is a little cooler, but 10C cooler is a little much.

 

I mean your did go from a deltaT of 9c to 8c on the setups so if the cpu had been left bare it would have dropped by the 1c at any given load. 

 

The coolant flow shouldn't really make a huge difference. You just need enough to create turbulence and also move the water away from the blocks fast enough the heat doesn't pool. That being said if you average the amount of time the water spends in the rads and blocks at 1.1 and 2.0 gpm it would come out to be the same amount of time.

 

TLDR: The only real contributor is going from direct die to an IHS and the gap seems larger than it should be. I think the LM application might have been lacking.

Thanks, really appreciate your input. Sorry I forgot to mention the fans were actually running at 90% on both 2018 & 2019 setups  all the fans were the same from both setups Noctua NF-A12x25  

 

[Mick Naughty]

Having no control, I assume it’s simple testing method as well and not repeating the build. 

[AngryBeaver]

10 hours ago, Shadowman said:

Thanks, really appreciate your input. Sorry I forgot to mention the fans were actually running at 90% on both 2018 & 2019 setups  all the fans were the same from both setups Noctua NF-A12x25  

 

Remember thought you wert from push/pull to just push. There is a decent difference in airflow moving through the radiator between the two. Now you did offset this with more radiator area, but remember cooling capacity is surface area + fan speed(air movement).

[MaratM]

What is the case, if you place a 480rad with a limited airflow it will not do any good

[Shadowman]

29 minutes ago, MaratM said:

What is the case, if you place a 480rad with a limited airflow it will not do any good

It's the CaseLabs SM8, the top has very good ventilation where the 480 is mounted with the fans.