[SOLVED] A Kraken G12 Anomaly

[memmeeyee]

Recently, I have bought the EVGA CLC 120mm AIO Cooler in hopes of cooling my 1080Ti so there would be more drops in the cards Mhz as well as being quiet. Even though it EVGA CLC 120mm Aio is not listed as compatible with the Kraken G12 mounting kit, It was the only one up on amazon and I coudln't wait any longer for the Corsair H55 to go back in stock, so I just went with it. I got the AIO, installed it on my card while also putting aluminum and copper heatsinks on the VRM/VRAM. I booted up and temps were sort of normal, could idle anywhere between 28-35c when it was downclocked to 600mhz. I booted up Heaven 4.0 to compare my results with another person who had the same setup (in reference to GPU, Hose orientation, and radiator placement, and airflow) but with the H55, a 120mm aio just like the evga clc 120. When the test was done I was completely unsatisfied and disappointed. He somehow got a max temp of 48c to his card when the benchmark was complete, while I got a max of 66c, people may be happy with a temp like that but im not, i should be getting the same temp and same performance as him with an 120mm AIO. But thats not it, he even outperformed me by 100 points at the benchmark at a higher clock.
Do you know the entire system he was using and exact card?His exact card was the EVGA GeForce GTX 1080 Ti Founders Edition 11GB GDDR5X 23.21.13.8831 (The exact same card as mine) and was using the i7 8700k at 3.7GHz while I have a i7 7700k at 4.2GHz, but still, that does not explain a 20c difference in the gpu and performance. His platform was Windows NT 6.2 (build 9200) 64 bit.
Did he use the standard test or change some settings?He did not change any setting, his settings are the exact same as mineRender: Direct3D11Mode: 1600x900 8xAA WindowedPreset: Extreme
What drivers was he on? Ive seen scores change just from newer drivers.If he is able to install something as complicated as a kraken g12 and an AIO on a 1080TI, he obviously keeps his drivers to date. His review was posted on September 2018, I looked up nvidia game ready drivers from that timeslot and found a Geforce Game Ready Driver that released Sept, 10th 2018, which was the latest one before Sept 16th, so it is very safe to assume he was using that Driver.
https://www.nvidia.com/Download/driverResults.aspx/137727/en-us
What room temps/case temps did he have?My ambient temp is 23c in my room, Its hard to believe that his can make a 20c difference even if his ambient is 10c less then me.

https://imgur.com/a/C0lDKqV

View: https://imgur.com/a/C0lDKqV

https://imgur.com/a/ttTVzHz

View: https://imgur.com/a/ttTVzHz

https://imgur.com/a/UsJf5AR

View: https://imgur.com/a/UsJf5AR

https://imgur.com/a/YrBRZJg

View: https://imgur.com/a/YrBRZJg

1.This is the benchmark with NVIDIA driver 445.87

1. The "latest" driver 446.143.
2. The amazon reviewers score

4.my score with Sept 10th, 2018 driver 399.24
As you can see, this still does not solve my problem, even though i got a higher score than the reviewer by 41 points. My 1080ti still got up to 66C!!! And his is still 48c, I know i installed everything correctly, from applying thermal paste to screwing in the radiator. I used Noctua's NT-H1 Thermal paste and applied it correctly.
But I am not using only ONE persons temps and comparing it to mine, here is a couple excerpts from Amazon reviews from people using the H55 and 1080ti together...
"My 1080 TI now sits at idle of 19C in a 72F Ambient Temp room. when I game it never goes over 54C, seriously!!! I even added 125Mhz to the clock to bump it up to 2025Mhz, still it runs 54C, and my FPS stays much more consistent while gaming"
"Installed this on a PNY GTX 1080ti (Blower edition) paired with a Corsair H55 in a push/pull config. Running it under full load and it so far hasn't even gotten beyond 45c. "
"So I was overclocking my Zotac 1080 ti Amp Edtion and was watching the temps hitting 80+ C on the GPU...Since the installation I haven't seen the GPU get any hotter than 47 degrees no matter what I do!"
"it went from the hight 80s C down to 48 C, almost 40 C degree difference. wow i was super happy. i used a corsair h55 with the g12 and they pair up good. "
"MSI 1080 Ti Aero here. Silent with no load but becomes a Husquarna leaf blower during gaming. Had to summon the Kraken. In the pictures you will observe heatsinks, GPU PWM adapter cable and Corsair H55.Before: 29C idle, gaming delved into 80s (Anno 1800 in 4K), high 60s and mid-70s (GW2 in 4K) - all settings maxed.After: 24C idle, gaming 50C and 40C max respectively for the above.
And this is on stock H55 fan (low RPM non-PWM). "

Oh one last note, I also tried to move my pc right infront of an ac vent so the air from the vent is going straight through the rear, and reconfigured the radiator fans so they're running push/pull as intake.

I idle at 16c-19c with the ac vent blowing directly into it

But the under load temps terrified me, through only half the benchmark of Heaven 4.0 it climbed insanely fast and reached 58c! Still 10c higher than the average review, with an cool AC air blowing directly into it!

I really want to hear what anyone has to say about this, please do so. But the only explanation I could see for this other than it being an anomaly is that the CL11 120 has worse heat transfer for gpu dyes.

 

[bmockeg]

Recently, I have bought the EVGA CLC 120mm AIO Cooler in hopes of cooling my 1080Ti so there would be more drops in the cards Mhz as well as being quiet. Even though it EVGA CLC 120mm Aio is not listed as compatible with the Kraken G12 mounting kit, It was the only one up on amazon and I coudln't wait any longer for the Corsair H55 to go back in stock, so I just went with it. I got the AIO, installed it on my card while also putting aluminum and copper heatsinks on the VRM/VRAM. I booted up and temps were sort of normal, could idle anywhere between 28-35c when it was downclocked to 600mhz. I booted up Heaven 4.0 to compare my results with another person who had the same setup (in reference to GPU, Hose orientation, and radiator placement, and airflow) but with the H55, a 120mm aio just like the evga clc 120. When the test was done I was completely unsatisfied and disappointed. He somehow got a max temp of 48c to his card when the benchmark was complete, while I got a max of 66c, people may be happy with a temp like that but im not, i should be getting the same temp and same performance as him with an 120mm AIO. But thats not it, he even outperformed me by 100 points at the benchmark at a higher clock.
Do you know the entire system he was using and exact card?His exact card was the EVGA GeForce GTX 1080 Ti Founders Edition 11GB GDDR5X 23.21.13.8831 (The exact same card as mine) and was using the i7 8700k at 3.7GHz while I have a i7 7700k at 4.2GHz, but still, that does not explain a 20c difference in the gpu and performance. His platform was Windows NT 6.2 (build 9200) 64 bit.
Did he use the standard test or change some settings?He did not change any setting, his settings are the exact same as mineRender: Direct3D11Mode: 1600x900 8xAA WindowedPreset: Extreme
What drivers was he on? Ive seen scores change just from newer drivers.If he is able to install something as complicated as a kraken g12 and an AIO on a 1080TI, he obviously keeps his drivers to date. His review was posted on September 2018, I looked up nvidia game ready drivers from that timeslot and found a Geforce Game Ready Driver that released Sept, 10th 2018, which was the latest one before Sept 16th, so it is very safe to assume he was using that Driver.
https://www.nvidia.com/Download/driverResults.aspx/137727/en-us
What room temps/case temps did he have?My ambient temp is 23c in my room, Its hard to believe that his can make a 20c difference even if his ambient is 10c less then me.

https://imgur.com/a/C0lDKqV

View: https://imgur.com/a/C0lDKqV

https://imgur.com/a/ttTVzHz

View: https://imgur.com/a/ttTVzHz

https://imgur.com/a/UsJf5AR

View: https://imgur.com/a/UsJf5AR

https://imgur.com/a/YrBRZJg

View: https://imgur.com/a/YrBRZJg

1.This is the benchmark with NVIDIA driver 445.87

1. The "latest" driver 446.143.
2. The amazon reviewers score

4.my score with Sept 10th, 2018 driver 399.24
As you can see, this still does not solve my problem, even though i got a higher score than the reviewer by 41 points. My 1080ti still got up to 66C!!! And his is still 48c, I know i installed everything correctly, from applying thermal paste to screwing in the radiator. I used Noctua's NT-H1 Thermal paste and applied it correctly.
But I am not using only ONE persons temps and comparing it to mine, here is a couple excerpts from Amazon reviews from people using the H55 and 1080ti together...
"My 1080 TI now sits at idle of 19C in a 72F Ambient Temp room. when I game it never goes over 54C, seriously!!! I even added 125Mhz to the clock to bump it up to 2025Mhz, still it runs 54C, and my FPS stays much more consistent while gaming"
"Installed this on a PNY GTX 1080ti (Blower edition) paired with a Corsair H55 in a push/pull config. Running it under full load and it so far hasn't even gotten beyond 45c. "
"So I was overclocking my Zotac 1080 ti Amp Edtion and was watching the temps hitting 80+ C on the GPU...Since the installation I haven't seen the GPU get any hotter than 47 degrees no matter what I do!"
"it went from the hight 80s C down to 48 C, almost 40 C degree difference. wow i was super happy. i used a corsair h55 with the g12 and they pair up good. "
"MSI 1080 Ti Aero here. Silent with no load but becomes a Husquarna leaf blower during gaming. Had to summon the Kraken. In the pictures you will observe heatsinks, GPU PWM adapter cable and Corsair H55.Before: 29C idle, gaming delved into 80s (Anno 1800 in 4K), high 60s and mid-70s (GW2 in 4K) - all settings maxed.After: 24C idle, gaming 50C and 40C max respectively for the above.
And this is on stock H55 fan (low RPM non-PWM). "

Oh one last note, I also tried to move my pc right infront of an ac vent so the air from the vent is going straight through the rear, and reconfigured the radiator fans so they're running push/pull as intake.

I idle at 16c-19c with the ac vent blowing directly into it

But the under load temps terrified me, through only half the benchmark of Heaven 4.0 it climbed insanely fast and reached 58c! Still 10c higher than the average review, with an cool AC air blowing directly into it!

I really want to hear what anyone has to say about this, please do so. But the only explanation I could see for this other than it being an anomaly is that the CL11 120 has worse heat transfer for gpu dyes.

Well, to start, EVGA has some of the best AIO coolers on the market. The issue is that their fans need to spin fast to get these results. What are you setting your fan to during load? Is it auto or are you at 100%?
2nd issue is ambient temp means little if you have your radiator set as exhaust. Currently your “cool” air is the heat from everything else in your case, which can easily make a 10c difference.

 

[memmeeyee]

Well, to start, EVGA has some of the best AIO coolers on the market. The issue is that their fans need to spin fast to get these results. What are you setting your fan to during load? Is it auto or are you at 100%?
2nd issue is ambient temp means little if you have your radiator set as exhaust. Currently your “cool” air is the heat from everything else in your case, which can easily make a 10c difference.

Minimum fan duty is at 60%, I even put both of them at max and still barely made any of a difference. I didnt set it as an exhaust, i set both push/pull fans as intake so they are pulling the ac air into the case.

 

[bmockeg]

Minimum fan duty is at 60%, I even put both of them at max and still barely made any of a difference. I didnt set it as an exhaust, i set both push/pull fans as intake so they are pulling the ac air into the case.

Sounds like a poor install then. Either there is too much or too little paste or there is not a good connection from the GPU die and the cold plate.

 

[memmeeyee]

Sounds like a poor install then. Either there is too much or too little paste or there is not a good connection from the GPU die and the cold plate.

Trust me, it is installed properly, reinstalled the pump three times, screwed in the screws until there was no torque. all while apply a perfect amount of nt-H1 thermal paste, either way, it shouldn’t be getting anywhere near those temps with and 120mm AIO, it shouldn’t be getting over 54c after hours of gaming at full load, as of the h55.

 

[bmockeg]

Trust me, it is installed properly, reinstalled the pump three times, all while apply a perfect amount of nt-H1 thermal paste, either way, it shouldn’t be getting anywhere near those temps with and 120mm AIO, it shouldn’t be getting over 54c after hours of gaming at full load, as of the h55.

I used a 280mm and it stayed at 35-40c under load. Not a fan of using 120 or 140mm on a top end GPU anyway, even as it is better than stock. Additionally, even at your temps it’s going to run at max speed without issue.

 

[memmeeyee]

I used a 280mm and it stayed at 35-40c under load. Not a fan of using 120 or 140mm on a top end GPU anyway, even as it is better than stock. Additionally, even at your temps it’s going to run at max speed without issue.

Like max speed at performance, it won’t, it thermal clocks, a 280mm AIO on a GPU is very overkill as well as pretty expensive but whatever floats your boat, I’m either gonna sell the evga clc or put in on my cpu when I get the h55, but I should be getting 48c at heaven 4.0 like basically anyone else with an 120mm no problem, not 66c.

 

[Phaaze88]

a 280mm AIO on a GPU is very overkill

A typical 280mm AIO of 25-30mm thick is good for ~300w of heat, and for a gpu that can easily pull 300w, a 280mm isn't really overkill.
Granted, all the waste heat isn't coming from the gpu die directly.

I've already ordered at Kraken G12 along with a Fractal Design Celsius S36 to pair with it, and will see before too long how that turns out.
Doing it to sate my curiosity and as a learning experience - I'm not having thermal issues at all; the darn thing never saw over 65C.

 

[memmeeyee]

A typical 280mm AIO of 25-30mm thick is good for ~300w of heat, and for a gpu that can easily pull 300w, a 280mm isn't really overkill.
Granted, all the waste heat isn't coming from the gpu die directly.

I've already ordered at Kraken G12 along with a Fractal Design Celsius S36 to pair with it, and will see before too long how that turns out.
Doing it to sate my curiosity and as a learning experience - I'm not having thermal issues at all; the darn thing never saw over 65C.

I mean you saw the temps people got in my main posts, they literally don’t get over 49c, no thermal clocking will occur. But that doesn’t explain how I’m getting 20c more running push/pull config with a 120mm AIO

 

[Phaaze88]

Oh yeah, I think I mentioned this to you before, but the reported gpu thermals are an average of 3: gpu core, memory, and VRM.
If you really do have the AIO mounted correctly, then I'd say you have an anomaly around:
-memory
-VRM
-airflow setup in general
The gpu core may be cool, but one or perhaps both of the other two aren't. It's the only other thing I think could be going on.

 

[memmeeyee]

Oh yeah, I think I mentioned this to you before, but the reported gpu thermals are an average of 3: gpu core, memory, and VRM.
If you really do have the AIO mounted correctly, then I'd say you have an anomaly around:
-memory
-VRM
-airflow setup in general
The gpu core may be cool, but one or perhaps both of the other two aren't. It's the only other thing I think could be going on.

Memory and vrm have aluminum and copper heatsinks placed on them, and as we know a 92mm fan blowing on it, so that eliminates it from being the anomaly. Aio pump was screwed in until it could barely be screwed in anymore. airflow isnt a problem, as i said even with an ac blowing directly onto the gpu and radiator with push/pull it only makes the max temp on heaven climb fast up to 58c. 10c more than average of reviews, videos, and on forums.

 

[Phaaze88]

I just remembered something else...
@bmockeg @memmeeyee
Are the pads and heatsinks on the VRMs and memory even necessary? While I was looking into how to assemble the G12 kit, it doesn't mention or even show in their own video, of the need to install such heatsinks.
It appears that the single fan and good chassis airflow are enough for those - I recall seeing some other posts saying they weren't necessary.

Perhaps the heatsinks you have installed are acting as insulators???

 

[memmeeyee]

Perhaps the heatsinks you have installed are acting as insulators???

thats impossible, many people get heastinks to get their card as cool as possible, They have thermal adhesive tape on the bottom as well, so Its impossible

 

[Phaaze88]

Sorry, I don't know what else to tell you. You've refuted just about all the possibilities:
1)Bad cooler mount
Not possible, it's been done X number of times already. It's correct.

2)High memory or VRM thermals
Not possible, that's why the thermal pads, tape and heatsinks are there.

3)Poor airflow
Not possible, there's the G12's 92mm fan, and an AC blowing into it as well.


-The Corsair H55 and EVGA CLC 120 have the same OEM - Asetek - so the only difference between them would be the fans and how the rads are mounted in the systems.
-If you are running the exact same gpu settings as others in Afterburner and in Nvidia Control Panel... well, that just takes us back to the 3 points you already refuted.

 

[memmeeyee]

Sorry, I don't know what else to tell you. You've refuted just about all the possibilities:
1)Bad cooler mount
Not possible, it's been done X number of times already. It's correct.

2)High memory or VRM thermals
Not possible, that's why the thermal pads, tape and heatsinks are there.

3)Poor airflow
Not possible, there's the G12's 92mm fan, and an AC blowing into it as well.


-The Corsair H55 and EVGA CLC 120 have the same OEM - Asetek - so the only difference between them would be the fans and how the rads are mounted in the systems.
-If you are running the exact same gpu settings as others in Afterburner and in Nvidia Control Panel... well, that just takes us back to the 3 points you already refuted.

If they are able to score higher than me, and get 20c less, at the same settings it doesnt matter what settings the have in afterburner or NVIDIA control panel

 

[memmeeyee]

ok so i contacted corsair regarding this and there is actually quite some differences which could indeed break the anomaly, heres what the cooling product manager said
"
The main difference is that the EVGA CLC is a Asetek Gen5 design while the H55 is an Asetek Gen4 design.

RPM/flow through the H55 may be slightly higher since I think the max pump RPMs are different, which would have an effect on performance. It's also possible that since the liquid routing for the H55 is side-inlet, the pump is fighting gravity a lot less.

Mounting pressure for the Gen5 may also be lower if there's any interference with the plastic bits on the pump cap. H55 is a lot lower profile and has less chances for any interference.

Higher pressure on the die, less air between the copper and die, but also more of a chance for you to bow the card or crack the die if there's ANY debris on it."

 

[Phaaze88]

I see... so that must account for a 20C difference.

 

[bmockeg]

I just remembered something else...
@bmockeg @memmeeyee
Are the pads and heatsinks on the VRMs and memory even necessary? While I was looking into how to assemble the G12 kit, it doesn't mention or even show in their own video, of the need to install such heatsinks.
It appears that the single fan and good chassis airflow are enough for those - I recall seeing some other posts saying they weren't necessary.

Perhaps the heatsinks you have installed are acting as insulators???

1080ti uses gddr 5 and doesn’t need heatsink a for memory. I had my memory clocked at 5950 without heatsinks. Doesn’t hurt to use them obviously. If you are using a 2000 series then the GDDR6 needs heatsinks.

 

[Phaaze88]

1080ti uses gddr 5 and doesn’t need heatsink a for memory. I had my memory clocked at 5950 without heatsinks. Doesn’t hurt to use them obviously. If you are using a 2000 series then the GDDR6 needs heatsinks.

Thanks for clearing that up!
I didn't bother ordering any heatsinks, due to the install video, and some older threads saying the same thing about GDDR5/X.

GDDR6 gets toasty, huh? Heh.

 

[memmeeyee]

1080ti uses gddr 5 and doesn’t need heatsink a for memory. I had my memory clocked at 5950 without heatsinks. Doesn’t hurt to use them obviously. If you are using a 2000 series then the GDDR6 needs heatsinks.

Many reviews and videos i seen, they were using heatsinks for their card