EVGA 2080Ti Black Gaming with Kraken G10, G12 and x52 personal experience - Overclock.net - An Overclocking Community

Forum Jump: 

EVGA 2080Ti Black Gaming with Kraken G10, G12 and x52 personal experience

 
Thread Tools
post #1 of 8 (permalink) Old 11-26-2019, 06:36 AM - Thread Starter
New to Overclock.net
 
Join Date: Nov 2019
Posts: 10
Rep: 0
EVGA 2080Ti Black Gaming with Kraken G10, G12 and x52 personal experience

After reading and watching various stuff online, I thought I should share my own experience.

How I ended up with them

Some background: First, there was no plan when I bought the 2080Ti (in September), so in retrospect, I probably wouldn't buy the non-A model again; then, water cooling was never on my mind when choosing the case (MSI Mag Vampiric 011c), which is another thing I would've done differently.

As you know, the stock cooler is pretty loud when the GPU's running at full load (about 2800 rpm at 70% speed) and the 112% power limit is quite annoying, so I flashed in the Palit BIOS to get that extra 12% power, but the fan speed range is borked. Before you ask, yes, I considered the shunt mod, but it's too adventurous to me because EVGA's generous warranty is a big reason why I chose them.

Google took me to Reddit, here, [H]ard, Youtube and other places in my search for an appropriate solution and it ended up with AIO cooler. Even though my case limits my choices a lot if I want to use custom loop, but I actually came up with a configuration that would work. I didn't go that route because:

1) It's kind of expensive although my config only costs ~US$ 400 converted;
2) Custom loop is too much work as I don't enjoy tinkering with my PC all the time;
3) It's more prone to fail since it's a lot more complicated, especially risk of leakage.

Of course, there are aftermarket air coolers, but I didn't find any particularly good one.

So, why Krakens? Because people already tried them and NZXT has a proven software to control the cooler, which, I think, sort of justifys the price tag. There's also a vendor that makes full aluminum AIO adapters for GPUs at a pretty reasonable price, but they don't have my model measured, so I would have to send them my card. If my case didn't limit my choices, I would've gotten an x72. In the end, I bought a second hand x52 for less than half of the retail price.

Installation

You may be wondering how G10 fits. Well, I mounted G12 with G10's backplate while keeping the original backplate as is on the card.

Why this way? Because I want to spread the pressure on the whole PCB as much as possible. It's probably 10-layer, but those thru-holes don't exactly give me confidence due to G12's AMD brackets not fitting in as shown in this thread.

My solution is to file off top threads by ~1.5 mm:
Click image for larger version

Name:	IMG_20191125_190602.jpg
Views:	6
Size:	4.30 MB
ID:	308134

and bottom plates' edges by ~2 mm:
Click image for larger version

Name:	IMG_20191125_183048.jpg
Views:	3
Size:	4.55 MB
ID:	308136

Special thanks to my father for the metalworking and finding the M3 screws and nuts below.

The screws fit in the G10 backplate like this:
Click image for larger version

Name:	IMG_20191125_190534.jpg
Views:	1
Size:	3.29 MB
ID:	308138

Yes, the text is upside down, but it's blocked off after installing in the case, so I didn't bother rotating it. The ideal length of the screws would be 15 mm.

2 other things worth mentioning:

1) I used anti-vibration mounts on G12 for the fan, but it turned out there's not enough space at the molex connectors, so I ended up using only 3 rubber sticks.
2) G12 wasn't designed for inductors on both sides, so I used a PCI fan bracket to mount another 9 mm fan on the left.

Temperatures and Benchmark

Precision X1 suggested +139 after scan, not as high as original BIOS's +145. I launched Furmark's stress test when I started typing this post. This is 2 hours later:
Click image for larger version

Name:	2019-11-26 21_06_59-Window.png
Views:	7
Size:	1.78 MB
ID:	308144
Click image for larger version

Name:	2019-11-26 21_10_21-Window.png
Views:	1
Size:	24.5 KB
ID:	308146

Port Royal auto overclock: https://www.3dmark.com/pr/177366 and manual overclock: https://www.3dmark.com/pr/182015

Also, the thermal paste I used initially has a rated conductivity of 6.2 W/(mK) which I find to be lacking since GPU and liquid temperature can differ up to 10 C, maybe more. The paste I'm using now is rated at 14.3 W/(mK) that kept the temperature difference within 8 C.

Budget 2080Ti
(16 items)
CPU
Ryzen 7 3700X
Motherboard
MSI MPG X570 Gaming Plus (MS-7C37) 2.0
GPU
EVGA 2080Ti Black Gaming (11G-P4-2281-KR)
RAM
G.Skill Trident Z Neo F4-3600C16D-32GTZNC
Power Supply
EVGA Supernova 750 P2
Cooling
Noctua NH-U12S
Cooling
NZXT Kraken X52
Case
MSI MAG Vampiric 011C
Monitor
Dell Ultrasharp U2417H
Keyboard
Realforce TKL SA / R2TLSA-US4-IV
Mouse
Logitech G Pro Wireless
Mousepad
Logitech G PowerPlay
Audio
AudioQuest DragonFly Red
Audio
Etymotic ER4XR
Audio
Sanwa Supply MM-MC24
Audio
XOX MX-3
▲ hide details ▲

Last edited by spot; 12-02-2019 at 04:13 AM.
spot is offline  
Sponsored Links
Advertisement
 
post #2 of 8 (permalink) Old 12-01-2019, 08:54 AM
New to Overclock.net
 
yoadknux's Avatar
 
Join Date: Jun 2016
Posts: 55
Rep: 0
Thanks for your post. It seems you put up a lot of effort to getting things to be "just right" so it was an interesting read. Just some points that I want to bring up:
a) People are gonna look at the "full view" picture and think "holy crap this guy doesn't care about cables at all", if you're posting something like this I'd advise cleaning the cables a little.
b) FurMark is really NOT The way to test GPU temperature, because the card doesn't run at its turbo's speed. From my experience, the game that heated my card the most was Plague Tale on 4k with supersampling, the stress test that heated my card the most was FireStrike Ultra and the long-term (2h+ loop) benchmark that heated my card the most was Superposition on 8k. I would run one of them for better indication about the temperatures.
c) I have Kraken X41 with 2x Noctua NF-A14 (~1100 RPM load), with Gelid GC Extreme paste, and on my 1080ti (same TDP as 2080ti) temps rarely hit 50c, I played a few hours of RDR2 and max was 44c, so I think your temps seem a bit off (especially considering you have a 240mm rad, not 140mm like me). I see your card has some sort of front plate, maybe it affects the contact? JayzTwoCents made a video about it once. I think your temps can and should be lower than that with this radiator.
yoadknux is offline  
post #3 of 8 (permalink) Old 12-01-2019, 03:11 PM - Thread Starter
New to Overclock.net
 
Join Date: Nov 2019
Posts: 10
Rep: 0
Quote: Originally Posted by yoadknux View Post
a) People are gonna look at the "full view" picture and think "holy crap this guy doesn't care about cables at all", if you're posting something like this I'd advise cleaning the cables a little.
D: I'm exactly that kind of guy, but I do understand your point, so photo deleted. And thanks for telling me what you really think.

Quote: Originally Posted by yoadknux View Post
b) FurMark is really NOT The way to test GPU temperature, because the card doesn't run at its turbo's speed. From my experience, the game that heated my card the most was Plague Tale on 4k with supersampling, the stress test that heated my card the most was FireStrike Ultra and the long-term (2h+ loop) benchmark that heated my card the most was Superposition on 8k. I would run one of them for better indication about the temperatures.
Well, I don't understand how the card really works, so I just looked at FurMark's TDP value and power limited indication. Will run FireStrike Ultra for 2 hours when I have the time and report back.

Quote: Originally Posted by yoadknux View Post
c) I have Kraken X41 with 2x Noctua NF-A14 (~1100 RPM load), with Gelid GC Extreme paste, and on my 1080ti (same TDP as 2080ti) temps rarely hit 50c, I played a few hours of RDR2 and max was 44c, so I think your temps seem a bit off (especially considering you have a 240mm rad, not 140mm like me). I see your card has some sort of front plate, maybe it affects the contact? JayzTwoCents made a video about it once. I think your temps can and should be lower than that with this radiator.
Umm...contact should be fine as I tried moving the pump a little when fastening the screws and it felt stable, no wedging. Also, I can clearly see there's clearance between the pump and the bracket, so the pressure is applied on the die, not the bracket. I suspected the temperature at 310W is a bit high, but it should not have anything to do with contact.

Gelid GC Extreme's rated thermal conductivity is 8.5 W/(mK), so unless mine falsely labels that, it shouldn't be the reason either.

I'm using x52's original fans, but due to the case's design, the blades are not fully exposed as in about 1/5 is covered by solid metal without mesh, which may or may not matter.

As for the in-game temperature, I've been playing Destiny 2 a lot recently to catch up on the season pass. The game runs at 4k 60fps with maxed settings and according to Argus Monitor's logs, GPU temperature hovers around 40 C, never exceeded 45.

So, was your 50 C temperature measured at 310 W?

---------------- Update ----------------

I ran FireStrike Ultra in windowed loop for 10 minutes and watched the power target in Precision X1. Sadly it only reached ~110% while FurMark kept it at 124%. So I will not be wasting time on FireStrike Ultra. By the way, 3DMark's stress test passed with 98.5%: https://www.3dmark.com/fsst/1312985

Budget 2080Ti
(16 items)
CPU
Ryzen 7 3700X
Motherboard
MSI MPG X570 Gaming Plus (MS-7C37) 2.0
GPU
EVGA 2080Ti Black Gaming (11G-P4-2281-KR)
RAM
G.Skill Trident Z Neo F4-3600C16D-32GTZNC
Power Supply
EVGA Supernova 750 P2
Cooling
Noctua NH-U12S
Cooling
NZXT Kraken X52
Case
MSI MAG Vampiric 011C
Monitor
Dell Ultrasharp U2417H
Keyboard
Realforce TKL SA / R2TLSA-US4-IV
Mouse
Logitech G Pro Wireless
Mousepad
Logitech G PowerPlay
Audio
AudioQuest DragonFly Red
Audio
Etymotic ER4XR
Audio
Sanwa Supply MM-MC24
Audio
XOX MX-3
▲ hide details ▲

Last edited by spot; 12-02-2019 at 02:35 AM.
spot is offline  
Sponsored Links
Advertisement
 
post #4 of 8 (permalink) Old 12-02-2019, 10:33 AM
New to Overclock.net
 
yoadknux's Avatar
 
Join Date: Jun 2016
Posts: 55
Rep: 0
Quote: Originally Posted by spot View Post
D: I'm exactly that kind of guy, but I do understand your point, so photo deleted. And thanks for telling me what you really think.


Well, I don't understand how the card really works, so I just looked at FurMark's TDP value and power limited indication. Will run FireStrike Ultra for 2 hours when I have the time and report back.


Umm...contact should be fine as I tried moving the pump a little when fastening the screws and it felt stable, no wedging. Also, I can clearly see there's clearance between the pump and the bracket, so the pressure is applied on the die, not the bracket. I suspected the temperature at 310W is a bit high, but it should not have anything to do with contact.

Gelid GC Extreme's rated thermal conductivity is 8.5 W/(mK), so unless mine falsely labels that, it shouldn't be the reason either.

I'm using x52's original fans, but due to the case's design, the blades are not fully exposed as in about 1/5 is covered by solid metal without mesh, which may or may not matter.

As for the in-game temperature, I've been playing Destiny 2 a lot recently to catch up on the season pass. The game runs at 4k 60fps with maxed settings and according to Argus Monitor's logs, GPU temperature hovers around 40 C, never exceeded 45.

So, was your 50 C temperature measured at 310 W?

---------------- Update ----------------

I ran FireStrike Ultra in windowed loop for 10 minutes and watched the power target in Precision X1. Sadly it only reached ~110% while FurMark kept it at 124%. So I will not be wasting time on FireStrike Ultra. By the way, 3DMark's stress test passed with 98.5%: https://www.3dmark.com/fsst/1312985
Ok, so your gaming temps are really low, it's a good thing.
I ran Furmark and FireStrike Ultra for 10 minutes each today. First of all I want to say FurMark heats the card more than I thought (46c after 10 minutes) BUT still not as much as FireStrike Ultra for me (47c).

Here is the screenshot from FurMark. As you can see peak TDP is 150% and power consumption is 380W according to GPU-Z. Those are the "good for heat test" paramters. However, the core clock is low (~1900) and gpu voltage is very very low (0.95V), I highlighted those things in my photo:
https://i.imgur.com/9lwd5ED.png

And here is FireStrike Ultra (after the run, I closed the windows by mistake). The temperature is not shown here, it reached 47c (slightly higher than FurMark) but voltage was higher at 1.05V and core clock was actually at turbo speed (~2076 MHz). Peak power was 375W, which is almost identical to what FurMark does. So the way I see things FireStrike actually uses the card unlike FurMark.
https://i.imgur.com/adfSAV6.png

If you got good temps with FireStrike Ultra, then everything is probably fine. The 60c just seemed a bit high for what I was expecting, but maybe FurMark does some black magic that I'm missing. Could be good if you could show me the voltage and clocks.
yoadknux is offline  
post #5 of 8 (permalink) Old 12-02-2019, 11:48 AM
New to Overclock.net
 
Zfast4y0u's Avatar
 
Join Date: Dec 2017
Posts: 305
Rep: 1 (Unique: 1)
Quote: Originally Posted by yoadknux View Post
Ok, so your gaming temps are really low, it's a good thing.
I ran Furmark and FireStrike Ultra for 10 minutes each today. First of all I want to say FurMark heats the card more than I thought (46c after 10 minutes) BUT still not as much as FireStrike Ultra for me (47c).

Here is the screenshot from FurMark. As you can see peak TDP is 150% and power consumption is 380W according to GPU-Z. Those are the "good for heat test" paramters. However, the core clock is low (~1900) and gpu voltage is very very low (0.95V), I highlighted those things in my photo:
https://i.imgur.com/9lwd5ED.png

And here is FireStrike Ultra (after the run, I closed the windows by mistake). The temperature is not shown here, it reached 47c (slightly higher than FurMark) but voltage was higher at 1.05V and core clock was actually at turbo speed (~2076 MHz). Peak power was 375W, which is almost identical to what FurMark does. So the way I see things FireStrike actually uses the card unlike FurMark.
https://i.imgur.com/adfSAV6.png

If you got good temps with FireStrike Ultra, then everything is probably fine. The 60c just seemed a bit high for what I was expecting, but maybe FurMark does some black magic that I'm missing. Could be good if you could show me the voltage and clocks.
furmark is less tasking on ur gpu then other benchmarks cause nvidia detects it with driver, and dosent let card get overloaded as much as furmark wants. if u want to test card with furmark go with 390.00 driver or older.
Zfast4y0u is offline  
post #6 of 8 (permalink) Old 12-02-2019, 12:37 PM - Thread Starter
New to Overclock.net
 
Join Date: Nov 2019
Posts: 10
Rep: 0
Quote: Originally Posted by yoadknux View Post
As you can see peak TDP is 150% and power consumption is 380W according to GPU-Z. Could be good if you could show me the voltage and clocks.
380 W is quite a bit higher than 310 W, so I think there could be something "wrong" with my system... The clocks were as shown in the screenshot: 1515 Mhz core and 7000 Mhz memory. I didn't take note of the voltage, so another run of FurMark showed 856 mV with the same overclock settings.

For comparison, what's your liquid temperature and core temperature when stablized at 380 W?

Budget 2080Ti
(16 items)
CPU
Ryzen 7 3700X
Motherboard
MSI MPG X570 Gaming Plus (MS-7C37) 2.0
GPU
EVGA 2080Ti Black Gaming (11G-P4-2281-KR)
RAM
G.Skill Trident Z Neo F4-3600C16D-32GTZNC
Power Supply
EVGA Supernova 750 P2
Cooling
Noctua NH-U12S
Cooling
NZXT Kraken X52
Case
MSI MAG Vampiric 011C
Monitor
Dell Ultrasharp U2417H
Keyboard
Realforce TKL SA / R2TLSA-US4-IV
Mouse
Logitech G Pro Wireless
Mousepad
Logitech G PowerPlay
Audio
AudioQuest DragonFly Red
Audio
Etymotic ER4XR
Audio
Sanwa Supply MM-MC24
Audio
XOX MX-3
▲ hide details ▲
spot is offline  
post #7 of 8 (permalink) Old 12-03-2019, 03:20 AM
New to Overclock.net
 
yoadknux's Avatar
 
Join Date: Jun 2016
Posts: 55
Rep: 0
Quote: Originally Posted by spot View Post
380 W is quite a bit higher than 310 W, so I think there could be something "wrong" with my system... The clocks were as shown in the screenshot: 1515 Mhz core and 7000 Mhz memory. I didn't take note of the voltage, so another run of FurMark showed 856 mV with the same overclock settings.

For comparison, what's your liquid temperature and core temperature when stablized at 380 W?
I wouldn't say there is anything wrong, just maybe not "100% optimized", you are after all getting 40c temps on gaming loads on a flagship card. By the way, those 375-380W numbers appear because my card is overclocked.
Your low voltage in Furmark is the reason I don't like Furmark as a heat test. Typical load voltages for Pascal and Turing cards are 1.05V-1.093V. This does raise some questions about how Furmark can use max power at low voltage, I think maybe it just draws more current from the PSU. As the other poster noted here, there is some built-in protection against Furmark.
Liquid temps are at about 50c under full stress. Theoretically liquid temp shouldn't be higher than gpu temp. It means that either the sensors have a different calibration, or the gpu sensor is located on a cooler spot of the die.
If we assume the X41 and X52 have the same sensors, then we more or less have the same temperatures. You can probably get those temps a bit lower with a different case/fans. The Noctua NF-A14 on my X41 are among the best, and they also have a lot of room to breath on my Cooler Master H500P.
yoadknux is offline  
post #8 of 8 (permalink) Old 12-03-2019, 04:23 AM - Thread Starter
New to Overclock.net
 
Join Date: Nov 2019
Posts: 10
Rep: 0
Quote: Originally Posted by yoadknux View Post
I wouldn't say there is anything wrong, just maybe not "100% optimized", you are after all getting 40c temps on gaming loads on a flagship card. By the way, those 375-380W numbers appear because my card is overclocked.
Your low voltage in Furmark is the reason I don't like Furmark as a heat test. Typical load voltages for Pascal and Turing cards are 1.05V-1.093V. This does raise some questions about how Furmark can use max power at low voltage, I think maybe it just draws more current from the PSU. As the other poster noted here, there is some built-in protection against Furmark.
Liquid temps are at about 50c under full stress. Theoretically liquid temp shouldn't be higher than gpu temp. It means that either the sensors have a different calibration, or the gpu sensor is located on a cooler spot of the die.
If we assume the X41 and X52 have the same sensors, then we more or less have the same temperatures. You can probably get those temps a bit lower with a different case/fans. The Noctua NF-A14 on my X41 are among the best, and they also have a lot of room to breath on my Cooler Master H500P.
That kind of confirmed my suspicion that the temperature you are getting is a bit off since x52 is designed for CPU and by some vague standard, only meant to dissipate 300 W or less power. I'm using the pull configuration on the radiator to avoid wires getting caught in the fan blades and I have to say the stock Aer P120 fans are loud at full speed which is required when venting 310 W, but unless I want to play RTX titles or go above 60 fps in 4k (Destiny 2 can reach 144 fps in certain scenes and mostly stay above 100 after my manual overclocking), the fans are quiet.

Speaking of which, you may have missed my updated link in the original post about the manual overclock (1980 MHz core, 8200 MHz ram) result: https://www.3dmark.com/pr/182015 . That's 8.5% increase compared to auto scan. So if I had your 150% power target, I think it's easy to maintain 2100 MHz core clock... Too bad nVidia implemented "secure boot".

Your case must be quite a bit larger than mine and it looks fancier. I knew x52 would be barely enough before buying it, but there's no room to install x62 or x72. There are also 3 A14s in the front of my case, so it's Cool'n'Quiet(tm). =)

In the end, I'm happy with the result that my card still works, performance improvement is obvious and it didn't cost too much. If I had to replace Aer P120s, Noctua NF A12x25 PWM would be the way, but man, aren't they expensive.

Budget 2080Ti
(16 items)
CPU
Ryzen 7 3700X
Motherboard
MSI MPG X570 Gaming Plus (MS-7C37) 2.0
GPU
EVGA 2080Ti Black Gaming (11G-P4-2281-KR)
RAM
G.Skill Trident Z Neo F4-3600C16D-32GTZNC
Power Supply
EVGA Supernova 750 P2
Cooling
Noctua NH-U12S
Cooling
NZXT Kraken X52
Case
MSI MAG Vampiric 011C
Monitor
Dell Ultrasharp U2417H
Keyboard
Realforce TKL SA / R2TLSA-US4-IV
Mouse
Logitech G Pro Wireless
Mousepad
Logitech G PowerPlay
Audio
AudioQuest DragonFly Red
Audio
Etymotic ER4XR
Audio
Sanwa Supply MM-MC24
Audio
XOX MX-3
▲ hide details ▲
spot is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off