Finally gathered the courage to install my first GPU waterblock on my MSI RTX 2080ti Gaming X Trio, and now my performance has gone down?

[mlangdon]

TL;DR - Went from stock cooler to EKWB. OC went from +145 core and +795 mem. To +90 core and +760 mem. A decent difference in FPS and almost 400 pts in 3dMark TimeSpy.

 

First off, thank you to anyone who attempts to help with this. I just installed my first custom loop after using AiOs forever. The two products I used to complete this loop are these from EKWB.

 

https://www.ekwb.com/shop/ek-kit-p360

 

https://www.ekwb.com/shop/ek-quantum-vector-trio-rtx-2080-ti-d-rgb-nickel-plexi

 

https://us.msi.com/Graphics-card/GeForce-RTX-2080-Ti-GAMING-X-TRIO

 

So I set up my loop following their instructions provided very closely. After installing thermal pads and paste to my card I attached the block, then removed it to ensure proper contact between the two. After finding it to match descriptions of proper contact, and no physical damage to be seen, I re applied the waterblock with the provided screws and washers (I did follow the instructions to use the smallest screw in the bag of screws provided).

 

After leak testing I finally booted up 3dMark to get some details about my performance. The first big change was that the benchmark would not run until I changed my core voltage to +0, and my power limit back to only 100. These sliders had been maxed while the card was air cooled. Next the benchmark wouldn't run with my current OC settings of +145 to core and +795 to mem, and after some testing I arrived at the highest values that would work which ended up being +90 to core and +760 to mem.

 

I hands down love the cooling, my card hit 50C max during my repeated test attempts, I understand 3dMark may not be the most heat intensive application, but my card was low to mid 60s before. I just am baffled as to why I've lost performance when I keep hearing about how watercooling can help your card step up a few notches. I use a Samsung CRG9 as my monitor so every little bit of extra performance helps games look better in 32:9.

 

Thank you again for the help. I knew I would have some road bumps in my first custom loop endeavor, I just didn't expect this to be it.

[NotSoEpicMods]

Try a remount, I know it sucks but it helped when I had issues with my 2080ti(had decently normal temps much lower than air) and I'm actually planning on a remount again this weekend as my temps are 58C currently. When I had my issues it turned out that the EK provided thermal paste was really thin and watery and somehow didn't cover the entire gpu chip. I went to a slightly thicker paste and it worked a lot better.

[mlangdon]

So I've actually remounted this like 3-4 times now. Figured that sounded a bit neurotic to mention before, but there it is. I used the thermo grizzly paste they provided in the CPU kit for the GPU core, and I used the one dot method. The GPU block came with some EK branded paste, but I figured the named brand would be better. At least it's shown it's strength when it comes to the GPU and CPU temps.

[Juular]

Try running MSI Afterburner auto OC after resetting everything to defaults, maybe you're doing smth wrong. Also, what PSU do you have ?

[amdorintel]

be 8888% sure u use enuff thermsl paste and its legit thermal paste bought from a known good store like Bestbuy, Memoryexpress, microcenter and not a random dude in a back alley hence fleabay aka ebay

[Mick Naughty]

If you could notice the FPS difference, then the testing wasn’t done right. 

Those few steps would only yield a few FPS. 

Didn’t see any clock speeds mentioned either. 

So you had an aio on it before or was the card not hitting 100% to only be on the 60’s?

[mlangdon]

7 hours ago, Juular said:

Try running MSI Afterburner auto OC after resetting everything to defaults, maybe you're doing smth wrong. Also, what PSU do you have ?

Afterburner gives an Auto OC now of +60 to the core. My PSU is a Seasonic Prime Ultra 850W Titanium.

7 hours ago, amdorintel said:

be 8888% sure u use enuff thermsl paste and its legit thermal paste bought from a known good store like Bestbuy, Memoryexpress, microcenter and not a random dude in a back alley hence fleabay aka ebay

Yup, EKWB supplied the thermo grizzly I've been using. I also have noctua paste from my local microcenter that I haven't used yet. Each time I've remounted the paste spread looks good on both the chip and the waterblock. I've seen some people swear by the X method on GPUs so maybe I should try that?

46 minutes ago, Mick Naughty said:

If you could notice the FPS difference, then the testing wasn’t done right. 

Those few steps would only yield a few FPS. 

Didn’t see any clock speeds mentioned either. 

So you had an aio on it before or was the card not hitting 100% to only be on the 60’s?

You are correct, it is several frames higher not a tremendous amount. This amount is the difference between always being above 60 and dipping below at moments with a screen of my size though. As for noticing I was just sitting there staring at the fps counter during the tests. 

 

Sorry about the clock speeds, I was just giving the modifiers from Afterburner. With the new OC the core is 1845mhz and memory is 7760mhz.

 

I used air cooling for my gpu before, sorry my main post was not clear in distinguishing between prior cpu and gpu cooling. So the 60s was the triple fans running at 100%

[Mick Naughty]

So what did it boost to before with and without an overclock and what is it doing now with and without?

Was there anything being ran as a comparison aside from 3d Mark? 

I assume you’re running 4K bringing up 60fps. 

[SpaceGhostC2C]

8 hours ago, mlangdon said:

I used the thermo grizzly paste they provided in the CPU kit for the GPU core, and I used the one dot method.

I haven't done it myself, but my understanding is that the dot method is recommended when applying over a heatspreader (like CPUs' IHS), but for direct die contact it's better to manually spread the thermal paste to ensure full coverage of the die.

I don't know if it matters much, you're better served by someone with more GPU watercooling experience.

[Zberg]

I dont know what to make of that especially with that many remounts.  The card itself at stock as an aircooled card is one of the better ones (that and strix have big beefy 2.5 slot, 3 fan coolers and tend to be the top performers) so I would assume there would be less of a difference between the better air cooled ones (vs something like the evga black or gigabyte windforce with their less impressive coolers/heatsinks) and water, but I wouldn't expect it to go in the opposite direction though, weird.

[mlangdon]

42 minutes ago, Mick Naughty said:

So what did it boost to before with and without an overclock and what is it doing now with and without?

Was there anything being ran as a comparison aside from 3d Mark? 

I assume you’re running 4K bringing up 60fps. 

Before it would OC to 1895 core and 7795 memory. Both waterblock and air boost to 1750 core and 7000 memory from factory.

 

I've also been running Heaven as well, fps difference is noticeable there too. Pretty much same difference as 3dMark.

 

I'm close to 4k. The CRG9 has a resolution of 5140x1440 so you're driving quite a lot of pixels.

 

[mlangdon]

42 minutes ago, SpaceGhostC2C said:

I haven't done it myself, but my understanding is that the dot method is recommended when applying over a heatspreader (like CPUs' IHS), but for direct die contact it's better to manually spread the thermal paste to ensure full coverage of the die.

I don't know if it matters much, you're better served by someone with more GPU watercooling experience.

 

I will give that a try this weekend. I'm pretty much down for anything at this point.

[Mick Naughty]

Well if it only boosts to what it’s default clock is, there isn’t much left. 

Using gpu z would say what’s holding it back. 

If you were getting the exact overlock you put in and can’t get it how, then there’s contact issue somewhere on the card or the drivers need to be reinstalled. 

Or it was never stable to begin with. 

[DobertRownySr]

Why would you waste time and money to watercool a 2080ti triple fan card? They barely even reach 70C under load......

Like I have the Zotac Amp, and mine has 2050-2150 mhz on the core and 8220 Memory clock with the stock fan profile which reaches 65% at 70C and my GPU never even gets to 72C.

[Kilrah]

If the GPU temp is lower it's unlikely the GPU thermal paste application is the problem. 

Would concentrate on VRM/memory cooling. For these you have no temp readings so you're going blind to fix it.