1080 TI vs 1080 TI SLI vs 2080 TI Personal Review

JMCB

Limp Gawd
Joined
May 26, 2010
Messages
421
Being that I was on the fence for the longest time, and the upgrade bug finally made me itch, I am now the proud owner of a brand new Nvidia Geforce RTX 2080 Ti Founder’s Edition. Previously, I was running two Gigabyte Geforce GTX 1080 Tis in SLI, and this was giving me awesome performance at 2k with Gysnc. I knew this was going to be a slight downgrade in some applications, but with SLI seeing less support, and the resale of 1080 Tis still maintaining their value, this was the perfect time to upgrade. To fully convince myself I decided I would do a graphics card comparison between the 1080 TI, 1080 TI SLI, and 2080 TI. Let’s look at the two cards we’ll be comparing:

Gigabyte Nvidia GeForce GTX 1080 Ti Turbo
VRAM: 11GB GDDR5
Memory Bus: 352-bit
Memory Bandwidth: 484 GB/s
Base Clock: 1480Mhz
Boost Clock: 1620Mhz
TDP: 250W

20181217_234612.jpg


Nvidia GeForce RTX 2080 TI Founder’s Edition
CUDA cores: 4352
Giga Rays/sec: 10
RTX-OPS: 78T (76T)
VRAM: 11GB GDDR6
Memory Bus: 352-bit
Memory Bandwidth: 616 GB/s
Base Clock: 1350Mhz
Boost Clock: 1635Mhz
TDP: 260W


20190225_195201.jpg

Here is the system specs for the system we’ll be testing on.

Test System:
CPU: Intel i7 8700K (stock speed @ 4.7ghz on watercooling EKWB EK-Supremacy EVO CPU)
Motherboard: Asus ROG Strix Z370
RAM: 32 GB DDR4 G.Skill TridentZ RBG 2400mhz
SSD: Samsung 960 EVO 500GB M.2
HD: WD 4TB Blue
Power Supply: Corsair HX1200
OS: Windows 10 Pro
Monitor: Asus PG279Q 2K 165hz
Nvidia Driver - 419.17


Programs Used:
Fraps (for In-game benchmarking)
3dMark Time Spy Benchmark
Apex Legends
Battlefield V
Call of Duty: Black Ops 4
Overwatch
Star Wars Battlefront 2 (2017)


Testing methodology:
All games and benchmarks were done with the latest Nvidia driver release, which is 419.17 as of 2/25/2019. All in-game settings were set to their highest possible settings at 2k resolution (2560 x 1440), with Gsync disabled so that it doesn’t affect our FPS in any way. Fraps was used to measure 3 minutes of game time which was once started in-game. Each game uses multiplayer on a server with lower ping, whenever possible, and each game was tested with the same multiplayer mode and map.

For the game and benchmark choices, these are games I regularly play.

3dMark Time Spy Benchmark
Developed with input from AMD, Intel, Microsoft, NVIDIA, and the others, 3DMark Time Spy is one of the first DirectX 12 apps to be built "the right way" from the ground up to fully realize the performance gains that the new API offers. With its pure DirectX 12 engine, which supports new API features like asynchronous compute, explicit multi-adapter, and multi-threading, 3DMark Time Spy is the ideal benchmark for testing the DirectX 12 performance of the latest graphics cards.

Results:

Gigabyte 1080 TI Turbo
Score: 8443
Validation Link: https://www.3dmark.com/3dm/33897012?


Gigabyte 1080 TI Turbo SLI
Score: 14,210
Validation Link: https://www.3dmark.com/3dm/33897209?

Nvidia 2080 TI FE
Score 12,607
Validation Link: https://www.3dmark.com/3dm/33974203?

I knew going in that the 1080 Ti SLI setup would outperform everything else, although single card vs single card, the improvement was almost 50%.


Apex Legends
Apex Legends is a free-to-play Battle Royale game where legendary competitors battle for glory, fame, and fortune on the fringes of the Frontier.

Results:

Gigabyte 1080 TI Turbo
Min: 76
Average: 113.82
Max: 145

Gigabyte 1080 TI Turbo SLI
Min: 37
Average: 89.6
Max: 143

Nvidia 2080 TI FE
Min: 82
Average: 118
Max: 146

Between the single card and SLI 1080 TI setups, we can see that SLI scaling is broken, despite NVIDIA saying it received support for SLI in this title with the latest driver version. While playing it in SLI, it seemed like a hot mess – stutters, inconsistent frame drops. Switching to one card greatly improved the experience. In addition, we weren’t getting a whole lot better performance from a 1080 Ti vs a 2080 Ti. I’m guessing this is also due to newer drivers that aren’t optimized for this title at all. I’m sure future updates will bring some improvements.


Battlefield V
Enter mankind’s greatest conflict with Battlefield V as the series goes back to its roots with a never-before-seen portrayal of World War 2. Lead your squad to victory in all-new multiplayer experiences like the multi-map Grand Operations. Fight across the globe in the single-player War Stories campaign. Assemble your Company of customized soldiers, weapons, and vehicles – then take them on an expanding journey through Tides of War. This is the most intense, immersive, and innovative Battlefield yet.

Game was played in DirectX 11, as DirectX 12 does not support SLI.

Results:

Gigabyte 1080 TI Turbo
Min: 56
Average: 95.25
Max: 137

Gigabyte 1080 TI Turbo SLI
Min: 89
Average: 109.6
Max: 151

Nvidia 2080 TI FE
Min: 102
Average: 120.16
Max: 181

The 2080 TI clears all the other setups and appears to be the best choice. In regards to the 1080 TU SLI, again, we had issues with SLI, but this time it is because Nvidia has disabled it for this title. This can be worked around with Nvidia Inspector, a free tool you can download. However, within the last few Nvidia driver updates they did something to lock that option out, and the settings won't save with Inspector, thus making SLI completely useless. There's a work around that involves going to the Nvidia driver file folder and making two files – nvdrsdb0.bin and dvdrsdb1.bin read only. After doing this, SLI worked a bit, although with abysmal scaling at only 15%, and I was able to finish out the test. This scaling is probably the result of Nvidia trying to lock out SLI, as past drivers I’ve had better framerates with this setup, and it wasn’t as jittery in gameplay.

And because we have to, I had to try this game with RTX ON! Unfortuantly, in the multiplayer levels I saw a very slight performance dip but didn’t see much difference in quality. So, I decided to load up the first level, and you could see the difference. Here is how the 2080 TI performed:

Nvidia 2080 TI FE (RTX at ULTRA Settings with Direct X 12)
Min: 46
Average: 56
Max: 101

Call of Duty: Black Ops 4
Call of Duty: Black Ops 4 features gritty, grounded, fluid Multiplayer combat, the biggest Zombies offering ever with three full undead adventures at launch, and Blackout, where the universe of Black Ops comes to life in one massive battle royale experience.

Game was tested in Team Deathmatch.

Results:

Gigabyte 1080 TI Turbo
Min: 56
Average: 95
Max: 137

Gigabyte 1080 TI Turbo SLI
Min: 122
Average: 134.37
Max: 141

Nvidia 2080 TI FE
Min: 82
Average: 135
Max: 189

The 2080 TI barely takes the cake with the best average, nudging out the SLI setup slightly. However, minimum framerate was better on the SLI setup. In addition, we also saw much better scaling 1080 TI vs 1080 TI SLI this time around.

Overwatch
The world needs heroes. Join over 40 million players as you take your place in the world of Overwatch. Choose your hero from a diverse cast of soldiers, scientists, adventurers, and oddities. Bend time, defy physics, and unleash an array of extraordinary powers and weapons. Engage your enemies in iconic locations from around the globe in the ultimate team-based shooter.

Results:

Gigabyte 1080 TI Turbo
Min: 113
Average: 180.37
Max: 208

Gigabyte 1080 TI Turbo SLI
Min: 144
Average: 172.7
Max: 214

Nvidia 2080 TI FE
Min: 129
Average: 216.32
Max: 276

Overwatch isn’t a demanding title, but still sees a lot of play, and numbers on this game with high-end gaming systems just don’t seem to be around. As you can see, all of these systems can run this game maxed out at 2K resolution with max settings easily, and with high enough refresh rates to make use out of these monitors with 144hz+ refresh rates.

The 2080 TI smokes both setups.

Star Wars Battlefront 2 (2017)
Hurtle between the rooftops of Theed in a speeding starfighter, cut swaths through armies of Clones with Darth Maul, and build your legacy on the battlefield. Your Star Wars fantasies come to life in Star Wars Battlefront II's multiplayer.

Results:

Gigabyte 1080 TI Turbo
Min: 50
Average: 76.45
Max: 99

Gigabyte 1080 TI Turbo SLI
Min: 63
Average: 103
Max: 141

Nvidia 2080 TI FE
Min: 70
Average: 120.8
Max: 146

One 1080 Ti felt smooth playing at 60fps, but on a higher refresh rate you could for sure feel the delay. Jumping to two 1080 TI SLI felt like a big improvement, but we were still not hitting the target of 120fps+ for the higher refresh monitor. The 2080 Ti was finally able to hit that benchmark, and just felt so much better.


Conclusion:
In real world applications, the 2080 Ti is either very close in performance or outperforms a 1080 Ti SLI setup. If you already have a 1080 Ti, at current market prices of $550, you’re better off selling it and paying the difference for a 2080 Ti vs getting another card for SLI; it’s a much better experience. If you can pay around $400-450 for the second 1080 Ti, that’s where things get interesting – go for the second card. However, do note that Nvidia seems to be focusing less and less on SLI scaling, and even a $400 add-on might not help the games you play.

If you already have a 1080 Ti SLI set up, this is where things can get a bit muddier. It just depends on if you feel like paying $300 after selling the two cards and getting a slight upgrade. I’m happy with this choice, but if you can wait, it may be better to hold off for the next generation of cards.

Overall, I’m very happy with my new card. The two Gigabyte Nvidia 1080 Ti’s have been sold on eBay.
 
Last edited:

cybereality

Supreme [H]ardness
Joined
Mar 22, 2008
Messages
5,001
Thanks for putting this together.

I agree that single faster card is the best option, and your numbers prove it.
 

Starrbuck

2[H]4U
Joined
Jun 12, 2005
Messages
2,533
Thanks for the comparison. The best things about 1080 Ti SLI vs. 2080 Ti are no SLI profiles to worry about, plus less heat, power consumption, and noise (if you are not on water).
 

WBurchnall

2[H]4U
Joined
Oct 10, 2009
Messages
2,620
Very interesting results. I'm surprised by the range. Sometimes one 1080ti is like 85 percent of the 2080ti with a price difference of almost 100 percent more for the later.

Other times the 2080ti seems to be like 35 to 40 percent faster than a single TI.

It seems depending upon the game that sometimes if you already have one 1080ti, it might not be with paying double the cost of your card for 15 percent more speed... If you have two cards to sell though seems like a decent upgrade
 

Stoly

Supreme [H]ardness
Joined
Jul 26, 2005
Messages
6,491
Pretty much what I expected.

This is further confirmation that SLI is dead

Not nvidia/amd nor developers are willing to push mGpu.
 

noko

Supreme [H]ardness
Joined
Apr 14, 2010
Messages
4,668
I’ve found sli at 2K weak and a lot of times cpu limiting. If at 2k I use DSR, render at higher resolution then downsampled. Pushing nearly 100FPS, Shadow Of The Tomb Raider, HDR at near 4K resolution on the Samsung monitor. Running at 1440p DSR, was the best visual gaming experience I’ve ever had. Scaling was over 90%. In this case it would blow away a 2080Ti. Too bad that is not common. I have a bunch of older games I am planning on playing which support SLI well.
 

Astrodave

[H]ard|Gawd
Joined
Apr 20, 2005
Messages
1,067
SLI, Nvidia 3D and Nvidia surround have been dying for years. They still work in older games and in some new games but Nvidia has clearly washed their hands of anything beyond a vanilla single card/monitor setup.
 

odditory

Supreme [H]ardness
Joined
Dec 23, 2007
Messages
5,872
SLI, Nvidia 3D and Nvidia surround have been dying for years. They still work in older games and in some new games but Nvidia has clearly washed their hands of anything beyond a vanilla single card/monitor setup.
Yep. SLI's been bleeding out for a long time. But Nvidia fading out from supporting SLI is merely symptomatic of a market fading out from demanding it (it also requires the game's developer to do the major part). Even if every game magically gained perfect SLI support tomorrow, there wouldn't be any surge of GPU buys to take advantage of it -- most people don't care so there's no incentive for Nvidia to waste manpower on it.

SLI was a solution to a problem in a time when games still ran circles around GPU's. Then GPU's caught up, and once a single card could play almost anything at any res, the die was cast.

Anyway, JMCB nice writeup.
 

IdiotInCharge

[H]F Junkie
Joined
Jun 13, 2003
Messages
13,269
Yep. SLI's been bleeding out for a long time.
A large part of this is that the render targets have stopped- while more exotic display types are hitting the market, 4k60 or 1440p144 seem to be the top end with most aiming for 1080p60.

We're still waiting for stuff like 4k120 and VR to become more 'mainstream', and alongside ray tracing, we'll likely see a resurgence of the need for multi-GPU.

For now though, I agree with the OP: buy the single fastest GPU you can.
 

psy81

Limp Gawd
Joined
Feb 18, 2011
Messages
508
It looks like your CPU bottlenecked at 2k. Battlefield V tends to scale well in SLI.

If you were using a 4k display I think the improvements in SLI will be bigger but still a single RTX 2080ti is the better option
 
Joined
Feb 9, 2013
Messages
594
It looks like your CPU bottlenecked at 2k. Battlefield V tends to scale well in SLI.

If you were using a 4k display I think the improvements in SLI will be bigger but still a single RTX 2080ti is the better option
Seems about right to me, I have a 7820x at 5GHZ w/ 1080 Ti SLI and get the same numbers as him. SLI just doesn't scale well anymore.
 

psy81

Limp Gawd
Joined
Feb 18, 2011
Messages
508
Seems about right to me, I have a 7820x at 5GHZ w/ 1080 Ti SLI and get the same numbers as him. SLI just doesn't scale well anymore.
Interesting... I actually don't have Battlefield V but have all the previous Battlefields which all scaled well. If its using the same Frostbyte Engine you'd think it would be the same. That said, I'm done with SLI as well. I paid $1,400 CAD for the 2XGTX1070s and most games my other card does nothing... It does look nice in my case though :p
 
Joined
Feb 9, 2013
Messages
594
Interesting... I actually don't have Battlefield V but have all the previous Battlefields which all scaled well. If its using the same Frostbyte Engine you'd think it would be the same. That said, I'm done with SLI as well. I paid $1,400 CAD for the 2XGTX1070s and most games my other card does nothing... It does look nice in my case though :p
Totally agree, it is strange. I think a lot of the engine was reworked when they added DX12.
 

JMCB

Limp Gawd
Joined
May 26, 2010
Messages
421
I played BF1 with 4x AMD r9 290x cards and it worked better than 1080 Ti SLI. I think it's mainly a driver issue.
 

lostin3d

[H]ard|Gawd
Joined
Oct 13, 2016
Messages
2,043
Thanks for you hard work on this.

If you should find yourself wanting to go thru all these hoops to add Metro Exodus to the list here's an SLI guide I just found. Guide is in the description and in the comments the author posts to an updated link. I originally went looking because someone really went off the deep end with a pair of 2080TI's running at 8k! I'm pretty sure they used this trick. I plan to try it on a laptop this weekend with a pair of 980m's at 1080p and 4k.


p.s. Nice thing about this video is 4k!

edit: Here's the link for the 8k vid 2080TI SLI test.

 

cybereality

Supreme [H]ardness
Joined
Mar 22, 2008
Messages
5,001
8K, omg!! I'm just finally comfortable with my 4K performance, now there is something else.
 

lostin3d

[H]ard|Gawd
Joined
Oct 13, 2016
Messages
2,043
8K, omg!! I'm just finally comfortable with my 4K performance, now there is something else.
Yeah initially I wanted to shrug it off. I'm still paying off my 2080TI and all the other things I upgraded during the holidays but I played that 8k stream on my 1440p 27" monitor and I could sill see quite a bit more crispness in details. Oh well.

DX11 was working, but last I heard the recent patch removed SLI support. That's the thing with SLI, it may work today may not tomorrow.
Too true. Same goes for PhysX. When I upgraded to my 2080TI from 1080 SLI I kept my cards just in case of a space invasion. Didn't happen, thankfully. One of the first things I did after install was reinstall one to try and use as a PhysX card for the 1st Metro's that I'd had such great experiences with in the beginning of 4k. Barely did anything using the latest RTX drivers even when I manually set it to the 1080(but NV CP did try to automatically set it there anyway). Tried it again with Exodus and turned on the PhysX indicator and it clearly shows it's only using CPU.

JMCB Think you'll part that GPU under liquid too? Could have some substantial gains.
 

Maddnotez

Limp Gawd
Joined
Nov 28, 2014
Messages
291
TL:DR - you have way too much money.

Jk, kind of. I read and enjoyed the info. I just got a used 1080 ti as an upgrade and I will check out RT in a few years.
 

JMCB

Limp Gawd
Joined
May 26, 2010
Messages
421
TL:DR - you have way too much money.

Jk, kind of. I read and enjoyed the info. I just got a used 1080 ti as an upgrade and I will check out RT in a few years.
We all have those money sinks. Luckily I've kept the costs down by selling what I've had while it maintained price. Tax returns also help haha.

Sadly, the 1080 Tis have been sold off on eBay. If in the event my card starts to play space invaders, I've got the integrated graphics on the 8700k to hold my over on videos...
 

Maddnotez

Limp Gawd
Joined
Nov 28, 2014
Messages
291
We all have those money sinks. Luckily I've kept the costs down by selling what I've had while it maintained price. Tax returns also help haha.

Sadly, the 1080 Tis have been sold off on eBay. If in the event my card starts to play space invaders, I've got the integrated graphics on the 8700k to hold my over on videos...
I'm a few gens behind you. Upgrading from the 970 which I am going to put in a second build for my kid. In case mine ever shits the bed I will reclaim lol.
 
  • Like
Reactions: JMCB
like this

vulcan78

n00b
Joined
Feb 10, 2020
Messages
8
I made an account just to post this, yeah kind of fail considering that youre using an M.2 PCI-E SSD that takes up a limited number of PCI-E lanes on Z370, leaving the 2nd GPU slot at x4 speed. This actually has quite a detrimental effect on SLI performance. You also only list games that have poor or negative SLI scaling, but even among them, in BF5 1080 TI SLI is actually faster than single 1080 Ti when the second card is at x8 or higher, way faster in fact (40%):


I'm tempted to go through and basically debunk all of the rest of your benchmarks but yeah, 960 EVO M.2 SSD taking up PCI-E lanes rendering 2nd GPU limited to x4 bandwidth speed, I'm surprised no-one else caught this.

I found this thread as I'm with a single 1080 Ti under water-block contemplating adding a 2nd used 1080 Ti FE for $400 vs paying ~$1200 for a used 2080 Ti before Ampere refresh. I may try this out and dump both 1080 Ti's when 3080 Ti arrives early 2021. Ray Tracing is out of the question for me as I'm @ 3440x1440 and RT maxed has FPS in the 50 FPS range on avg. in all 5 titles that support the feature (/s) as per HardwareCanucks recent RT in 2020 verdict:

Here's what 1080 Ti vs 2080 Ti actually looks like without the 2nd card hobbled at 4x speed and the games compared not carefully curated titles that mostly exhibit negative scaling (except BF5, 1080 Ti SLI is 40% faster than single 2080 Ti with both cards at 8/16x speed as per the video above).



Post script:

Not only that but the 1080 Ti variant you used to compare against the 2080 Ti FE is probably the worst 1080 Ti you could possibly go with in terms of cooling. All of the "Turbo" variants ran as hot as FE on avg., leading to a loss of 50-100 MHz vs open air cooler design. You probably had both cards at 1800-1850 MHz when in actuality they can do 2000 MHz simply bringing the core down to 60-65C (not 85-90C). This is a null argument because you have them in SLI so in this case blower may actually perform better as it's pushing the heat out of the case. But I've seen testing where this doesn't matter if your case has ample airflow, primary GPU may run 5C hotter but on avg both GPU's will run 10-15C (i.e. 65-70C vs 80-85C) cooler than blower design. But in your case, no pun intended, you are trying to compare two GPU's with an inferior cooling solution and higher temps to a single card with a superior cooling solution and lower temps. Not only does 2080 Ti FE have an open air cooler design that allows it to run ~10C cooler but it is factory overclocked by 100 MHz to skew things in the other direction. When both 1080 Ti and 2080 Ti are running at the same freq. there is only about a 25% on avg difference in performance but when you have both 1080 Ti's running about 150 MHz under what your 2080 Ti FE is running at and then you have 2nd card running at 4x bandwidth, I mean what kind of comparison is this? It's great if you want to tell yourself that the $1300 "upgrade" was worth it but yeah, really lousy testing methodology and youre not fooling anyone with discernment.
 
Last edited:

vulcan78

n00b
Joined
Feb 10, 2020
Messages
8
Thanks for you hard work on this.

If you should find yourself wanting to go thru all these hoops to add Metro Exodus to the list here's an SLI guide I just found. Guide is in the description and in the comments the author posts to an updated link. I originally went looking because someone really went off the deep end with a pair of 2080TI's running at 8k! I'm pretty sure they used this trick. I plan to try it on a laptop this weekend with a pair of 980m's at 1080p and 4k.


p.s. Nice thing about this video is 4k!

edit: Here's the link for the 8k vid 2080TI SLI test.

"Hard work". OP has 2nd GPU hobbled at 4x speed because boot drive is M.2 SSD occupying precious few PCI-E lanes on Z370 chipset. See my post above where 1080 Ti SLI is actually 40% faster in BF5 than single 2080 Ti when both cards are at 8/16x. Entire thread should be tossed out purely on these grounds. All of the benchmarks are basically meaningless except for the few titles that have negative scaling, I mean there's a point to be made here, but not all games have negative scaling. I'm not saying that SLI is a smart move in 2020, but it's far from dead:

If youre in my boat with a single 1080 Ti and can pick up a 2nd one used for $400, I mean even if one or two titles youre playing doesn't have SLI support that's still far preferable than forking out $1200 before taxes for 2080 Ti that itself shows variable gains (40% in some games sure, but as low as 22% in others, it's almost as bad as SLI in this regard).

I have my primary 1080 Ti under water block, so heat uptake by primary card isn't a concern as both cards would be in a loop with 1kw rad surface area and I have a 1kw PSU ready to go (I built my system so that I could add a 2nd GPU down the road if I wanted to).
 
Last edited:

vulcan78

n00b
Joined
Feb 10, 2020
Messages
8
I forgot to add this, M.2 SSD is not even worth it considering it takes up PCI-E lanes, I know this from personal experience. I picked up a 970 EVO during black friday of 2018 and was completely underwhelmed. Not only did Windows not really boot any faster (until I made 15 different changes in BIOS) but games only loaded 1-2 seconds faster on avg:


What OP could have done instead of paying $1300 after taxes for what is in reality a ~25-30% bump in performance over single 1080 Ti is they could have purchased two GPU waterblocks, a larger radiator or two, and migrated entire system to a larger chassis that is more ideal for watercooling, such as Thermaltake View 71, which is what I'm using. Instead of buying that expensive M.2 SSD (960 EVO) they could have purchased a conventional SSD at half the price.

2x GPU water-blocks: $200-300
1-2 large radiators: $100-200
Thermaltake View 71: $150

$450-$650

vs

$1200 for 2080 Ti
$250 for 960 EVO

$450-650 vs $1450

Actual 1080 Ti SLI vs 2080 Ti performance with all cards running at 8/16x (in case the previous examples weren't enough)


So basically, this is an example of a misconfigured system and a waste of money.
 

Furious_Styles

[H]ard|Gawd
Joined
Jan 16, 2013
Messages
1,573
I forgot to add this, M.2 SSD is not even worth it considering it takes up PCI-E lanes, I know this from personal experience. I picked up a 970 EVO during black friday of 2018 and was completely underwhelmed. Not only did Windows not really boot any faster (until I made 15 different changes in BIOS) but games only loaded 1-2 seconds faster on avg:


What OP could have done instead of paying $1300 after taxes for what is in reality a ~25-30% bump in performance over single 1080 Ti is they could have purchased two GPU waterblocks, a larger radiator or two, and migrated entire system to a larger chassis that is more ideal for watercooling, such as Thermaltake View 71, which is what I'm using. Instead of buying that expensive M.2 SSD (960 EVO) they could have purchased a conventional SSD at half the price.

2x GPU water-blocks: $200-300
1-2 large radiators: $100-200
Thermaltake View 71: $150

$450-$650

vs

$1200 for 2080 Ti
$250 for 960 EVO

$450-650 vs $1450

Actual 1080 Ti SLI vs 2080 Ti performance with all cards running at 8/16x (in case the previous examples weren't enough)


So basically, this is an example of a misconfigured system and a waste of money.
You are really reaching hard here. Trying to go open loop is no minor task for someone who has never done it before. Plus how about all those games where SLI just doesn't work? Yeah you're SOL. I'd rather buy the fastest card available than dick around with SLI settings and hope they keep supporting it in future games.
 

vegeta535

2[H]4U
Joined
Jul 19, 2013
Messages
3,698
You are really reaching hard here. Trying to go open loop is no minor task for someone who has never done it before. Plus how about all those games where SLI just doesn't work? Yeah you're SOL. I'd rather buy the fastest card available than dick around with SLI settings and hope they keep supporting it in future games.
I don't agree that open water loop is such a daunting task for people that visit this site. Just for most people the cost isn't justifiable. Everything else you said is spot on.
 

cybereality

Supreme [H]ardness
Joined
Mar 22, 2008
Messages
5,001
I get what you are saying vulcan78 but I think you're mistaken.

I've ran SLI for years, even had 2x 2080 Tis at one point, and it's just not worth it.

Many games don't support SLI (espeically newer games in DX12 and Vulkan) and even when it "works" there are bugs and annoying things you have to try to fix.

I sold my second 2080 Ti and honestly performance seems about the same (meaning it's more than fine with 1 card for what I'm doing). YMMV.
 

vulcan78

n00b
Joined
Feb 10, 2020
Messages
8
You are really reaching hard here. Trying to go open loop is no minor task for someone who has never done it before. Plus how about all those games where SLI just doesn't work? Yeah you're SOL. I'd rather buy the fastest card available than dick around with SLI settings and hope they keep supporting it in future games.
OP already has a loop, look in the image, I see a D5 pump / reservoir and a CPU block.

I get what you are saying vulcan78 but I think you're mistaken.

I've ran SLI for years, even had 2x 2080 Tis at one point, and it's just not worth it.

Many games don't support SLI (espeically newer games in DX12 and Vulkan) and even when it "works" there are bugs and annoying things you have to try to fix.

I sold my second 2080 Ti and honestly performance seems about the same (meaning it's more than fine with 1 card for what I'm doing). YMMV.
I ditched SLI years ago as well (680M SLI replaced with a single 980M, Alienware M18xR2 and 780 Ti SLI replaced with a single 980 Ti, now I have a single 1080 Ti) but looking for more performance if one is with a single 1080 Ti in a loop with enough radiator surface adding a second 1080 Ti may actually be a preferable alternative to getting price gouged with "2080 Ti" (in actuality the entire Turing stack was renamed one GPU higher, i.e. "RTX 2070" is in actuality the 60 card: TU-106 SKU, no SLI, only as fast as the outgoing 80 card, not the 80 Ti card like every new 70 card going back to Kepler).

Right now I can pick up a used 1080 Ti FE with an EK block for $400. Add another $30 for a terminal and I'm at performance that is easily 50% faster than single 2080 Ti in a vast majority of titles.

Looking at my Steam library, the games that are the most demanding at the resolution I play at (3440x1440) all have SLI support:

Assassins Creed: Odyssey
Watchdogs 2
No Mans Sky
Kingdom Come: Deliverance (AFR 2)
The Witcher 3
Rise of the Tomb Raider
Shadow of the Tomb Raider (near perfect scaling, see video above)
Fallout 4
Fallout 76 (should I ever elect to pick this title up)
Forza Horizon 4
Far Cry 5
Resident Evil 2 (and upcoming 3)
Devil May Cry 5
Project Cars 2
GTA 5
Rage 2
Wolfenstein: New Colossus

And a slew of titles I don't even have but may pick up in the future:


The titles that aren't listed above are not demanding enough to require more GPU compute. I can run them with one GPU.

Additionally, I mine crypto (NiceHash), solely to heatt my apartment at this point @ $ .70 a day with the 1080 Ti I have @ 65% PT (200W). It gets cold here (San Francisco), and I find that I'm often using a space-heater at the foot of my desk in addition to the ambient heat my PC is producing (like when I'm perusing the web and typing comments such as this during the day, I primarily game for 2-3 hours during the evening). I could offset my heating bill with an additional 1080 Ti by about $15 a month. $400 purchase becomes a $300 purchase 6 months hence and at that point I will probably sell both 1080 Ti's and upgrade to a single 3080 Ti.

Yeah, there will always be that one demanding title or two that doesn't support SLI but I can live with that vs spending $1200 before-taxes for 25% bump on avg. up and over 1080 Ti and Ray Tracing that I can't even turn on because RT is too demanding at 3440x1440 (see HardwareCanucks recent appraisal of RT in 2020, where they were averaging 50 FPS @ 2560x1440 with a single 2080 Ti with RT maxed).

I'm just tired of waiting to upgrade and I'm not about to "upgrade" (sorry, ~25% isn't an upgrade) to 2080 Ti. Simply from an ethical, moral perspective I'm disgusted with the entire Turing line-up and want no part in providing fiscal positive feedback to NGreedia for their decision to essentially double the price of their products with their rename scam.

In fact I'm downright disgusted with them and I do honestly hope AMD gives them a serious run for their money this year with Big Navi. Hell if I wasn't in love with my UW panel (AW3418DW @ 120 Hz) or if anything should happen to it I may go with a Free-sync panel and just ditch NGreedia altogether. I might go with an all AMD build next year. Tired of both Intel and NGreedia, this is what a monopoly in the industry looks and feels like: $1300 80 card, $900 70 card, and $600 60 card when one acknowledges they simply renamed the entire lineup one GPU higher and 14+++++++++++++++++ iterative garbage where 9900k has same IPC as 6700k some 4 years on. Zero incentive to actually innovate, just double the prices and milk the mindless consuming public. No thanks.

Anyhow, still mulling over adding a second 1080 Ti for $400 and some change for 50% more performance up and over single 2080 Ti, the last vid above nearly pushed me off the fence in that direction. Scrutinizing that video, it seems that 1080 Ti SLI (stock clocks) is on avg 25-30% faster than single 2080 Ti (overclocked). Considering 1080 Ti can OC another 15%, if there is perfect scaling 50% faster isn't unreasonable.

Anyhow, I may wait until the 5th of March to see whether or not AMD will announce Big Navi (leak shows an RX card 17% faster than 2080 Ti in VR benchmark), if they do, then NGreedia will definitely follow suit and announce Ampere at GTC later that month. Now way is NGreedia going to let AMD sell a card faster than 2080 Ti for $700 until Computex in June, the expected announcement window for Ampere. Remember, both manufacturers likely have their respective cards ready to go now (AMD just received Korean certification for a new GPU and that happens 1-2 months before releasing a new GPU historically).

RTX 3080 will probably be 25-30% faster than 2080 Ti, solely in rasterization, and possibly 50% or faster in RT for $800.

$400 for 1080 Ti SLI is very tempting though, but I may wait, I have an extensive back-log of less demanding titles to play through that I should probably get out of the way now and save the more demanding titles for after I upgrade my GPU.
 
Last edited:
  • Like
Reactions: noko
like this

Furious_Styles

[H]ard|Gawd
Joined
Jan 16, 2013
Messages
1,573
OP already has a loop, look in the image, I see a D5 pump / reservoir and a CPU block.



I ditched SLI years ago as well (680M SLI replaced with a single 980M, Alienware M18xR2 and 780 Ti SLI replaced with a single 980 Ti, now I have a single 1080 Ti) but looking for more performance if one is with a single 1080 Ti in a loop with enough radiator surface adding a second 1080 Ti may actually be a preferable alternative to getting price gouged with "2080 Ti" (in actuality the entire Turing stack was renamed one GPU higher, i.e. "RTX 2070" is in actuality the 60 card: TU-106 SKU, no SLI, only as fast as the outgoing 80 card, not the 80 Ti card like every new 70 card going back to Kepler).

Right now I can pick up a used 1080 Ti FE with an EK block for $400. Add another $30 for a terminal and I'm at performance that is easily 50% faster than single 2080 Ti in a vast majority of titles.

Looking at my Steam library, the games that are the most demanding at the resolution I play at (3440x1440) all have SLI support:

Assassins Creed: Odyssey
Watchdogs 2
No Mans Sky
Kingdom Come: Deliverance (AFR 2)
The Witcher 3
Rise of the Tomb Raider
Shadow of the Tomb Raider (near perfect scaling, see video above)
Fallout 4
Fallout 76 (should I ever elect to pick this title up)
Forza Horizon 4
Far Cry 5
Resident Evil 2 (and upcoming 3)
Devil May Cry 5
Project Cars 2
GTA 5
Rage 2
Wolfenstein: New Colossus

And a slew of titles I don't even have but may pick up in the future:


The titles that aren't listed above are not demanding enough to require more GPU compute. I can run them with one GPU.

Additionally, I mine crypto (NiceHash), solely to heatt my apartment at this point @ $ .70 a day with the 1080 Ti I have @ 65% PT (200W). It gets cold here (San Francisco), and I find that I'm often using a space-heater at the foot of my desk in addition to the ambient heat my PC is producing (like when I'm perusing the web and typing comments such as this during the day, I primarily game for 2-3 hours during the evening). I could offset my heating bill with an additional 1080 Ti by about $15 a month. $400 purchase becomes a $300 purchase 6 months hence and at that point I will probably sell both 1080 Ti's and upgrade to a single 3080 Ti.

Yeah, there will always be that one demanding title or two that doesn't support SLI but I can live with that vs spending $1200 before-taxes for 25% bump on avg. up and over 1080 Ti and Ray Tracing that I can't even turn on because RT is too demanding at 3440x1440 (see HardwareCanucks recent appraisal of RT in 2020, where they were averaging 50 FPS @ 2560x1440 with a single 2080 Ti with RT maxed).

I'm just tired of waiting to upgrade and I'm not about to "upgrade" (sorry, ~25% isn't an upgrade) to 2080 Ti. Simply from an ethical, moral perspective I'm disgusted with the entire Turing line-up and want no part in providing fiscal positive feedback to NGreedia for their decision to essentially double the price of their products with their rename scam.

In fact I'm downright disgusted with them and I do honestly hope AMD gives them a serious run for their money this year with Big Navi. Hell if I wasn't in love with my UW panel (AW3418DW @ 120 Hz) or if anything should happen to it I may go with a Free-sync panel and just ditch NGreedia altogether. I might go with an all AMD build next year. Tired of both Intel and NGreedia, this is what a monopoly in the industry looks and feels like: $1300 80 card, $900 70 card, and $600 60 card when one acknowledges they simply renamed the entire lineup one GPU higher and 14+++++++++++++++++ iterative garbage where 9900k has same IPC as 6700k some 4 years on. Zero incentive to actually innovate, just double the prices and milk the mindless consuming public. No thanks.

Anyhow, still mulling over adding a second 1080 Ti for $400 and some change for 50% more performance up and over single 2080 Ti, the last vid above nearly pushed me off the fence in that direction.
Wow, pass the bud my friend!
 

vulcan78

n00b
Joined
Feb 10, 2020
Messages
8
Wow, pass the bud my friend!
Nice jab, where do you feel that I'm mistaken with the information that I've presented or are you another 2080 Ti owner who feels the need to justify how stupid a $1300 for 25% performance bump decision is?

Here you go buddy, go ahead and light up on your end with your 2080 Ti, youre going to need it:

Please remind yourself while youre puffing away that the 2080 Ti is overclocked to 2200 MHz and the 1080 Ti's are at stock clocks. Imagine the performance difference with the 1080 Ti's under full water block running at 2000 MHz instead of 1800 MHz.
 

AltTabbins

Fully [H]
Joined
Jul 29, 2005
Messages
19,744
I ran 1080 (non ti) and 2080ti's in SLI for a while and while I don't have numbers, I think the overall experience with just 1 of each was better than with 2. For titles that supported and scaled well with SLI, it was great. For titles that didn't, sometimes I would actually have issues like lower frame rate than a single card, or stuttering, or frame dips. Performance might have averaged to be better with SLI when you considered the ups and downs, but the overall experience for me is better with a single card. It might be the fault of SLI, it might be the fault of only have 16 pcie lanes talking to the CPU at one time.. I don't know the tech details of it. I just know that I have a better overall experience on a single card.
 
Last edited:

cybereality

Supreme [H]ardness
Joined
Mar 22, 2008
Messages
5,001
Correct AltTabbins . FPS numbers don't tell the full story, I was actually tricked as well looking at the FPS meter not realizing the whole experience sucked.

I had read about microstutter before, and knew it was an issue, but didn't realize exactly what it was for a while. My eye-opener was when I got a Radeon VII to replace 2x Vega 64 cards.

On paper, the Vega Crossfire was no question getting higher FPS. But the VII card was leaps and bounds smoother, even with a lower FPS. That is not even considering issues like the shadow flickering in Far Cry 5, which was now gone.

Though Nvidia was not as bad, that is a good example of how higher FPS does not mean smoother or a better experience. Even G-Sync/FreeSync does not overcome the issue. After that I sold my second 2080 Ti and haven't looked back.

Unless there are some huge redesign with how SLI/Crossfire works (I thought that was coming with mGPU, but that hasn't panned out) then I don't think I'd ever consider multi-GPU again.

Also, don't you think it's strange to be calling the company "NGreedia" yet you run their hardware and are promoting people to buy their product? I mean, you can just buy AMD if you don't like their practices.

I mean, I have a 2080 Ti, but I'll probably go AMD next round if their ray tracing solution pans out. Also might consider running Ryzen on my main machine, even though everything is working fine now so I won't upgrade until necessary (or I can't fight the bug any longer).
 
Last edited:

serpretetsky

[H]ard|Gawd
Joined
Dec 24, 2008
Messages
1,721
I made an account just to post this, yeah kind of fail
Hi vulcan78! Welcome to these forums!

I mean what kind of comparison is this? It's great if you want to tell yourself that the $1300 "upgrade" was worth it but yeah, really lousy testing methodology and youre not fooling anyone with discernment.
"Hard work"
What OP could have done instead of paying $1300 after taxes for what is in reality a ~25-30% bump in performance over single 1080 Ti is they could ...
So basically, this is an example of a misconfigured system and a waste of money.
People might be more receptive to you if you don't insult and then continuously point out perceived failures of the OP in their personal test.
The OP is not, after all, a professional reviewer.
A simple request if he could rerun the test with a SATA drive, for example, might work. Or whatever other recommendation you might have.

(I'm not an admin, and this is simply a suggestion)
 

Furious_Styles

[H]ard|Gawd
Joined
Jan 16, 2013
Messages
1,573
Nice jab, where do you feel that I'm mistaken with the information that I've presented or are you another 2080 Ti owner who feels the need to justify how stupid a $1300 for 25% performance bump decision is?

Here you go buddy, go ahead and light up on your end with your 2080 Ti, youre going to need it:

Please remind yourself while youre puffing away that the 2080 Ti is overclocked to 2200 MHz and the 1080 Ti's are at stock clocks. Imagine the performance difference with the 1080 Ti's under full water block running at 2000 MHz instead of 1800 MHz.
No, I'm not justifying anything, and I did not spend 1300 on a 2080 ti, or even 1000 for that matter. You're the one jumping through hoops to justify 1080 ti SLI with the "ngreedia" comments for added effect.
 

Dayaks

Supreme [H]ardness
Joined
Feb 22, 2012
Messages
7,566
SLI isn’t even an option in my mind so the choice was easy for me. The last thing I need is the inconsistency of a second GPU. My gaming time is limited enough.
 

Armenius

I Drive Myself to the [H]ospital
Joined
Jan 28, 2014
Messages
20,700
I was big into SLI for a good while. My last SLI setup was a pair of GTX 970. I went to a single GTX Titan X. In pure FPS the Titan was a little slower but the experience was a lot better. With the declining SLI support I have not even considered doing it again since.
"Hard work". OP has 2nd GPU hobbled at 4x speed because boot drive is M.2 SSD occupying precious few PCI-E lanes on Z370 chipset. See my post above where 1080 Ti SLI is actually 40% faster in BF5 than single 2080 Ti when both cards are at 8/16x. Entire thread should be tossed out purely on these grounds. All of the benchmarks are basically meaningless except for the few titles that have negative scaling, I mean there's a point to be made here, but not all games have negative scaling. I'm not saying that SLI is a smart move in 2020, but it's far from dead:

If youre in my boat with a single 1080 Ti and can pick up a 2nd one used for $400, I mean even if one or two titles youre playing doesn't have SLI support that's still far preferable than forking out $1200 before taxes for 2080 Ti that itself shows variable gains (40% in some games sure, but as low as 22% in others, it's almost as bad as SLI in this regard).

I have my primary 1080 Ti under water block, so heat uptake by primary card isn't a concern as both cards would be in a loop with 1kw rad surface area and I have a 1kw PSU ready to go (I built my system so that I could add a 2nd GPU down the road if I wanted to).
That's not how PCI-E lanes on Intel work. The video cards get 16 lanes dedicated to them from the CPU. The M.2 slot get 4 lanes from the PCH. Running 2 video cards and an M.2 NVME drive gives x8/x8 to the video cards and x4 to the SSD.
 
Top