another fool with a new RTX 2080 Ti

[batboy]

Well, I picked up the EVGA RTX 2080 Ti Black Edition the other day, but before I opened the box, I decided to return it because it has a lower power limit (112%) then most of the other cards (129%). The one I have now is the next step up: EVGA RTX 2080 Ti XC. I had to pay a bit more, but since I plan to put it underwater, I wanted the option of higher power limits when I benchmark. I also bought an EK Vector water block for the card. I plan on getting some baseline benchmarks while it's still on stock air cooling and then compare to what I get watercooled. I'll post some photos as I install the block and report how it does in benchmarks. This should be fun.

Is the MSI Afterburner still the preferred way to OC these Nvidia's? Maybe there are others apps that are better nowadays? I've been away from the flock for a long time.

 

[Brando]

I bounced between precision and afterburner multiple times but settled on afterburner. it's just better in every way. precision seems buggy to me. it doesn't load or save profiles sometimes and the osd seems to be temperamental. afterburner is more intuitive and just werks.

PS: looking forward to seeing the extra performance from being able to raise voltage. I need to see if the extra 15% cost comes out to 15% extra performance before my step up window is closed.

 

[batboy]

A teaser pic of the new card. I'll install it tomorrow morning.

 

[Woomack]

Well, I picked up the EVGA RTX 2080 Ti Black Edition the other day, but before I opened the box, I decided to return it because it has a lower power limit (112%) then most of the other cards (129%). The one I have now is the next step up: EVGA RTX 2080 Ti XC. I had to pay a bit more, but since I plan to put it underwater, I wanted the option of higher power limits when I benchmark. I also bought an EK Vector water block for the card. I plan on getting some baseline benchmarks while it's still on stock air cooling and then compare to what I get watercooled. I'll post some photos as I install the block and report how it does in benchmarks. This should be fun.

Is the MSI Afterburner still the preferred way to OC these Nvidia's? Maybe there are others apps that are better nowadays? I've been away from the flock for a long time.

I would pick higher power limit card too but afaik all have the same max voltage limit so in most cases what matters more is luck to good GPU than anything else.
In PrecisionX you can typically use K mode which you can't really set in Afterburner (well, can set constant voltage, etc. but it doesn't work the same way). In real it doesn't change much but in some benchmarks results with K mode are a bit higher just because card is not going into power saving mode between tests. I would say check both and pick the one which works better for you. I had no problems with PrecisionX on RTX2060 or GTX1080Ti.
Not going to buy any higher Nvidia right now. Will probably wait till new AMD hit the stores as I have much more fun with Vega 56 than I had with any Nvidia in past 3-4 years.

I guess we will see some results soon

 

[batboy]

Thanks for reminding me about Precision X, That's EVGA's software too. Might try that first then.

Yeah, whether it was worth the premium price for a card that has a slightly higher power limit is debatable. The typical gamer would never notice the difference. But the competitive benchmark guys need mo power (even us part time benchers).

I have a XFX Radeon RX 580 on the desk and was comparing the two cards. Both are a two slot design and have two fans. The 2080 Ti is almost a 1/2 inch longer. The RX 580 has a single 8-pin connector compared to two 8-pin connectors to power the hungry vid card. The 2080 Ti is very heavy, but that's mostly heat sinks weight which will be removed to install the water block.

 

[Brando]

Once you have your max oc dialed in under water I'd appreciate it if you could run a superposition benchmark in 4k and post a pic here. Also curious if the power limit lets it run higher in 4k consistently in game. The best I can get in 4k ultra (no rtx) in BFV while actually playing is around 2050mhz with good cooling. TBH is bounces between 2000-2050 sometime a tiny bit higher or lower for a second here and there. Looking forward to seeing what water can do with higher power limit and higher voltage.

 

[Woomack]

Most these cards can make 2000-2100 with average of 2050. What is interesting, the same on RTX2060, 2070 and 2080. Also, voltage is the same on all of them. Depends on GPU, boost frequency is higher or lower and it's adjusted additionally by temps and power limit. Water cooling should help to keep higher boost stable. So of course we are waiting on results but I wouldn't expect anything much different. Up to 50MHz is a typical delta, regardless if card is from cheaper or the most expensive series. It's also one of the reasons why this EVGA is interesting option.

 

[batboy]

I've never ran Superposition in 4K before. But, I did today just for you. Here's a conservative OC with stock air cooling:



I'm having fun playing with this card which is still in the stock air cooling mode. I guess folks are anxious for me to start installing the water block. Ok, let me remove this bad boy vid card and start working. I have to make a few changes to my loop to add the GPU. So, this might take a while.

 

[Brando]

That's pretty good. I had to get down to 50c to get that. I wonder if having a 10 core adds to the score?

 

[batboy]

Here's what a naked 2080 Ti looks like (middle). On the right are the heat sinks and fans, and on the left is the back plate. I bought an EK back plate that will help cool the card and looks good.

 

[batboy]

Between being overly cautious and tired, I was really fighting with the block install yesterday. Instructions could have been way better. After more research and watching some youtube vids, I finally got it figured out and installed. Then the back plate got stubborn and I got grumpy, so I went to bed. This morning, I took it back apart to double check things. The heat tape just needed compressed a bit more. So, got the water block and back plate on now. Whew! Looks good, sleek, and trim. The only thing that I don't like is the that slide-in, hold-on bracket. It looks goofy, but it's not really visible since it'll be underneath. I promised photos, so here they are (sorry they're kind of crappy). Now I have to drain my WC loop and start plumbing the GPU block into the system. That will probably be later this afternoon. Maybe by tonight I'll be back up and running if the leak test is okay.





EDIT: The water block has a RGB module with a cable that plugs into the RGB header on your motherboard. But, I don't have one of those headers on my TUF X299 Mark 2 board. It's not a big deal, if a water block had been in stock without the RGB, I would have got it instead. I suppose to make it work I'd have to get some complicated and expensive RGB controller?

I wasted time surfing the net when I should have been working on this project, but XSPC makes a $12 RGB controller that works with the EK RGB module.

 

[Brando]

Looks pretty nice. Sometimes I think about going custom water but I'm too much of a cheapskate. Anticipating results.

 

[batboy]

Leak testing is currently ongoing. Lots of air bubbles, so it still needs burped.

 

[EarthDog]

Pretty sure rgb controllers are cheap... 20-40 or so...

 

[batboy]

Yeah, I ordered a XSPC RGB controller for $12. I found at least 3 people saying it worked on the EK modules.

http://www.xs-pc.com/lighting/rgb-sata-controller

Leak test is complete, but still trying to work more air out of the loop.

I couldn't wait any longer. The system is running and after 5 minutes of idle while I checked everything, I took this screen shot of the vid card temp. Looks good, but about the same as air cooled.




Well, that Precision X1 sure looks pretty, but the reason why air and water temps were the same is the program is stuck on 29 degrees. I tried running 3D benchmarks and temp don't budge from 29 degrees. Nice, I get to brag that my GPU load temp is 29 degrees C. The Futuremark log said 78 degrees C. was my GPU temp for most of the benchmark. That's also with the fans set to quiet mode. I don't have a way to measure it, but I'd bet 78 was the coolant temp.

I'm not going to overclock the GPU until I bleed more air out of the system.

 

[habbajabba]

$1200+ card and the software doesn't even know your under water. Games will be epic though on the BIG screen.

 

[batboy]

The GPU must be making insufficient contact on the block or something. I tried Firestrike at default card settings to see what temps looked like. Under load the temp shot up to 78 degrees and stayed there. Now that's really weird. Benchmark results were way low, like half as much as with air cooling. I guess it comes back apart and see what I screwed up.

 

[EarthDog]

It probably throttled to keep the temps in line...

78C was NOT the coolent temp....no way.

 

[batboy]

Yep, not coolant temp when the CPU is averaging 46 C during the benchmark. It all points to improper contact. I have to disassemble it. I just drained the loop again. Sigh. Think I'm getting too old for this.

I took it apart. Sure disassembles quicker than assembles. Only about a third of the thermal paste showed up as a footprint on the block. What did touch was not touching by much. I think I see what's wrong. One strip of heat tape is thicker than the rest. I need to find the instructions and see what I did wrong. At least this is fixable.

EDIT: I'm an idiot, I put one heat tape strip on the wrong place. I corrected the error and reapplied thermal paste. I got it back together again. I was super careful this time and double checked everything. I'm sure it'll be okay this time.

Famous last words.

It's doing the same exact thing again. Phooey. I'm shutting it down and ignoring it until tomorrow.

 

[batboy]

It 's finally working right. Third time's a charm. I took it apart again and could not see anything wrong at first. The thermal paste once again was not making good contact. I did a couple hours of searching google and found several people with the same problem, a couple of the guys said they fixed it by changing the screws. They had used the original 4 spring screws around the GPU, but when they used the small screws provided by EK, it supposedly worked. I too had used the original spring screws. When I checked the instructions, it was a little vague, but it did say to install the water blocks with the provided screws. I was skeptical, but that worked. Wow, hard to believe screws were causing that big of a problem. The instructions need to be improved. Whew, glad I didn't ruin this bad boy.

I'm still bleeding air from the water loop, but I did run Firestrike with no problem and got a good score, a few points higher than using air cooling (with the same settings). Highest GPU and CPU temps were both in the 40s. Oh yeah, that's more like it.

Earthdoggy, you were right about the thermal throttling. Now I know, if you see the temp spike and then flatten out (like a plateau) and stay the same, that's because it's throttling. It's a good thing it did thermal throttle or I'd be guano out of luck with a dead GPU.