• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

980ti SLI on wildlands not scaling properly?

Soldato
Joined
22 May 2010
Posts
12,894
Location
Minibotpc
So today i finally got my second 980ti for SLI. First game and only game i’ve tried so far is Ghost Recon Wildlands.

On single gpu i get around 70-80fps on high settings at 1080p

With SLI enabled, i get around 40-60fps on high settings at 1080p!

Somethings not quite right, does this game even support SLI? SLI is enabled in Nvidia Control panel and in programs its set to Force alternate frame rendering 2.

I ran with MSI OSD and its weird:

CPU: 85% avg usage
GPU1 & GPU2: around 50-60% usage

I was expecting higher usage, could this be down to SLI profiles? Im running the latest GPU driver.

System specs:
3770k 4.4ghz
980ti poseidon @stock speeds
16gb 1600mhz

Anyone suggest any profiles i could try or workarounds to get it working correctly?
 
So today i finally got my second 980ti for SLI. First game and only game i’ve tried so far is Ghost Recon Wildlands.

On single gpu i get around 70-80fps on high settings at 1080p

With SLI enabled, i get around 40-60fps on high settings at 1080p!

Somethings not quite right, does this game even support SLI? SLI is enabled in Nvidia Control panel and in programs its set to Force alternate frame rendering 2.

I ran with MSI OSD and its weird:

CPU: 85% avg usage
GPU1 & GPU2: around 50-60% usage

I was expecting higher usage, could this be down to SLI profiles? Im running the latest GPU driver.

System specs:
3770k 4.4ghz
980ti poseidon @stock speeds
16gb 1600mhz

Anyone suggest any profiles i could try or workarounds to get it working correctly?

SLI is dead long time now, barely any new games support it these days. You should have just upgraded card. However there are 2 things to check

a) If you are using Reshade SLI doesn't work outright. Uninstall Reshade or other injectors.

b) Many report SLI issues with last few months drivers.
https://www.nvidia.com/en-us/geforc...a-gtx-1080-classifieds-in-sli-mode-broken-in/

And yes I know quoting issues from Nvidia forums in here will get scoffed by some people :p
 
SLI is dead long time now, barely any new games support it these days. You should have just upgraded card. However there are 2 things to check

a) If you are using Reshade SLI doesn't work outright. Uninstall Reshade or other injectors.

b) Many report SLI issues with last few months drivers.
https://www.nvidia.com/en-us/geforc...a-gtx-1080-classifieds-in-sli-mode-broken-in/

And yes I know quoting issues from Nvidia forums in here will get scoffed by some people :p
Yeah i figured SLI was dead but i really enjoyed my last setup with my 680's which lasted me a good 4-5 years before i really needed to say goodbye. Thought i'd give it one last blast, there are some older games i still play so i guess i'll try those to make sure SLI is scaling properly first.
 
One reason I got rid of my last sli setup a few years ago now. It just don’t work right. Especially not the 9 series as they don’t have the nvlink bridge or high bandwidth bridge like the 10 series did.

Usually I found if it had poor scaling you could sometimes try different sli bit profiles in nvinspector. Sometimes adjusting AA and other settings would bring the performance back.

Just too much faffing around and in a lot of cases even with gpu’s running as they should you’d get better performance for some parts of the game and other parts the same or worse.

Quite a lot of games especially modern ones you’d even get worse performance. Some do run better but it’s quite a rare occurrence.

Sell both cards and get a decent single card in my opinion.
 
Looking back at my results again, i think my 3770k is bottlenecking my 980ti’s. My 3770k would hit 95% or more in some instances where as both my gpu’s would hit around 50-60% usage.
 
Shadow of the Tomb Raider 100% works. Test that first (free demo on steam), and test with higher resolutions using DSR for ruling out cpu bottleneck.
 
Shadow of the Tomb Raider 100% works. Test that first (free demo on steam), and test with higher resolutions using DSR for ruling out cpu bottleneck.

I'll give it a blast tonight, i tried GTA5 late last night and seemed to get higher fps than single gpu however again the usage for both gpu's was in the mid 40's avg. It's really weird, it's like both gpu's are just sharing the workload rather than pushing it to get higher fps.

Is there anything in the settings i could have missed? I'm running windows 10, anything in there that needs to be set?

In my previous SLI setup, was in windows 7 and never encountered these issues.
 
Nothing weird mate, you literally answered your own question....3550k is bottle-necking it. A faster CPU there means it would stop looking like it was "sharing the load" and usage would jump up on both cards.
 
SLI does scale in the game, I suspect your CPU limited as other's mentioned.

At 3440 x 1440 with Ultra preset and SMAA I can see decent scaling. For refrence, GPU's are 1080Ti's at stock doing boosting to constant 1886 Mhz. CPU is 7980xe @ 4.8 GHz 8 cores and 4.6 GHz 10 cores.

Wildlands has a nice extract showing CPU / GPU usage details in full if you follow it through to My documents. My results:

Single GPU

FPS:
Average:
51.31
Min: 46.31
Max: 57.09

CPU usage:
Average:
8.8%
Min: 5.8%
Max: 30.5%

GPU usage:
Average:
98.1%
Min: 98.0%
Max: 98.8%

Dual GPU

FPS:
Average:
76.99
Min: 62.44
Max: 83.83

CPU usage:
Average:
12.3%
Min: 9.4%
Max: 23.9%

GPU usage:
Average:
96.9%
Min: 86.5%
Max: 97.8%

I will add though, CPU usage is a slight red herring as with 36 threads, usually 1-2 are heavily loaded, so while CPU usage may be low overall, its focused. Not too much of an issue as the overall average GPU usage in both runs is fairly high. Scaling can be exaggerated by employing DSR and enabling higher resolutions, drop in FPS but as a % SLI scaling does go up.
 
SLI does scale in the game, I suspect your CPU limited as other's mentioned.

At 3440 x 1440 with Ultra preset and SMAA I can see decent scaling. For refrence, GPU's are 1080Ti's at stock doing boosting to constant 1886 Mhz. CPU is 7980xe @ 4.8 GHz 8 cores and 4.6 GHz 10 cores.

Wildlands has a nice extract showing CPU / GPU usage details in full if you follow it through to My documents. My results:

Single GPU

FPS:
Average:
51.31
Min: 46.31
Max: 57.09

CPU usage:
Average:
8.8%
Min: 5.8%
Max: 30.5%

GPU usage:
Average:
98.1%
Min: 98.0%
Max: 98.8%

Dual GPU

FPS:
Average:
76.99
Min: 62.44
Max: 83.83

CPU usage:
Average:
12.3%
Min: 9.4%
Max: 23.9%

GPU usage:
Average:
96.9%
Min: 86.5%
Max: 97.8%

I will add though, CPU usage is a slight red herring as with 36 threads, usually 1-2 are heavily loaded, so while CPU usage may be low overall, its focused. Not too much of an issue as the overall average GPU usage in both runs is fairly high. Scaling can be exaggerated by employing DSR and enabling higher resolutions, drop in FPS but as a % SLI scaling does go up.

So it does work, i suspect the same as well wit the cpu being the limiter here.

But here's an interesting find!

I ramped up the visual scaling in options from 1.0 to 2.0 and behold my gpu's ramped up to nearly 90% usage while maintaining above and beyond 60fps?! CPU usage was running around the 85-90% mark.

What is going on here? Is this normal, because if my gpu's can scale up with the visual scaling, why can't it increase its usage and increase the fps??? I'm confused!
 
So it does work, i suspect the same as well wit the cpu being the limiter here.

But here's an interesting find!

I ramped up the visual scaling in options from 1.0 to 2.0 and behold my gpu's ramped up to nearly 90% usage while maintaining above and beyond 60fps?! CPU usage was running around the 85-90% mark.

What is going on here? Is this normal, because if my gpu's can scale up with the visual scaling, why can't it increase its usage and increase the fps??? I'm confused!

I imagine its still a case of your CPU being the limit. It's not just the GPU that dictates FPS, its the CPU which actually has to prepare the frames to deliver them to the GPU's.

By x2 the resolution scale you would normally get to the point at which your GPU's act as the overall limit to system performance. But given you had such low usage on GPU's before and high CPU usage, you had a lot of headroom which your GPU's in effect are now using which is why your GPU's ramped up to 90% usage now.

Check your CPU usage on a core by core basis using something like MSI AB, you may see that some of your cores are maxed out limiting things. the 85%-90% you usage you see for the CPU is an average across all your cores, It does not account for a few cores being maxed out which is acting as a limit.
 
So today i finally got my second 980ti for SLI. First game and only game i’ve tried so far is Ghost Recon Wildlands.

On single gpu i get around 70-80fps on high settings at 1080p

With SLI enabled, i get around 40-60fps on high settings at 1080p!

Somethings not quite right, does this game even support SLI? SLI is enabled in Nvidia Control panel and in programs its set to Force alternate frame rendering 2.

I ran with MSI OSD and its weird:

CPU: 85% avg usage
GPU1 & GPU2: around 50-60% usage

I was expecting higher usage, could this be down to SLI profiles? Im running the latest GPU driver.

System specs:
3770k 4.4ghz
980ti poseidon @stock speeds
16gb 1600mhz

Anyone suggest any profiles i could try or workarounds to get it working correctly?
You're CPU bottlenecked! Both my 1080tis steam along at 98-99% usage in GTA 5, SLI isn't dead it still scales really well in some games. I'm on 4.5Ghz 5960X. It shouldn't be set to Force alternate rendering unless SLI isn't supported and if that's the case (SLI not supported) then
you'll get crap scaling. I had Ghost recon wildlands for 48hours and it scaled well on my titan XPs even though I didn't think it looked much better so get a new CPU and those 980tis will shine. I also used 980ti in SLI in GTA 5 and again GPU usage was always maxed with very good scaling.
That SLI connector are you using e.g. single ribbon? Ooops seems others have said the same, yeah increasing the visual settings means the GPUs have to work harder producing a better image but they're always going to be limited by how fast the CPU can feed them the raw data.
 
You're CPU bottlenecked! Both my 1080tis steam along at 98-99% usage in GTA 5, SLI isn't dead it still scales really well in some games. I'm on 4.5Ghz 5960X. It shouldn't be set to Force alternate rendering unless SLI isn't supported and if that's the case (SLI not supported) then
you'll get crap scaling. I had Ghost recon wildlands for 48hours and it scaled well on my titan XPs even though I didn't think it looked much better so get a new CPU and those 980tis will shine. I also used 980ti in SLI in GTA 5 and again GPU usage was always maxed with very good scaling.
That SLI connector are you using e.g. single ribbon? Ooops seems others have said the same, yeah increasing the visual settings means the GPUs have to work harder producing a better image but they're always going to be limited by how fast the CPU can feed them the raw data.

Picking up a new board, cpu and ram tomorrow. I've gone for the following as an upgrade... (purely because im using an intel waterblock otherwise i would have gone for a AMD 3600.

9600KF (just over £200 from a competitor)
Gigabyte Z390 Gaming SLI Board (i'm on a budget)
Corsair Vengeance RGB Pro 3200mhz 16gb

See what improvements i'll get from this gpu usage wise.. if you're right then i should see 90-100% usage and a dramatic increase in frames.

I'm using single ribbon, is it true i can use the HB connectors even though there wont be much if any improvement? I'm using one out of convenience.
 
Picking up a new board, cpu and ram tomorrow. I've gone for the following as an upgrade... (purely because im using an intel waterblock otherwise i would have gone for a AMD 3600.

9600KF (just over £200 from a competitor)
Gigabyte Z390 Gaming SLI Board (i'm on a budget)
Corsair Vengeance RGB Pro 3200mhz 16gb

See what improvements i'll get from this gpu usage wise.. if you're right then i should see 90-100% usage and a dramatic increase in frames.

I'm using single ribbon, is it true i can use the HB connectors even though there wont be much if any improvement? I'm using one out of convenience.

That should be a solid improvement, be interested to hear back what you see :)

Yes a HB bridge is capable of higher throughput and it does make some difference in some cases vs ribbon bridges: https://www.pcworld.com/article/3087524/tested-the-payoff-in-buying-nvidias-40-sli-hb-bridge.html Though not sure if the 9xx series can take advantage of it.

the LED hard bridges which preceded the HB bridge though is also capable of higher throughput in most cases so can usually grab one of those for cheap for use with 9xx as that is more capable the the ribbon stuff.
 
That should be a solid improvement, be interested to hear back what you see :)

Yes a HB bridge is capable of higher throughput and it does make some difference in some cases vs ribbon bridges: https://www.pcworld.com/article/3087524/tested-the-payoff-in-buying-nvidias-40-sli-hb-bridge.html Though not sure if the 9xx series can take advantage of it.

the LED hard bridges which preceded the HB bridge though is also capable of higher throughput in most cases so can usually grab one of those for cheap for use with 9xx as that is more capable the the ribbon stuff.

Thats interesting to know, does it matter on brand as i've seen lots of different branded ones from ROG to Aorus LED hard bridges.

Thanks
 
Thats interesting to know, does it matter on brand as i've seen lots of different branded ones from ROG to Aorus LED hard bridges.

Thanks

All just for looks and brand matching etc really. I seem to remember that using a decent bridge on my old 980Ti SLI setup did help.

Just have to make sure you get one with the correct spacing.
 
So its in, frame rates have more than doubled! Able to crank most settings to ultra or ultra high on wildlands now.

Went for the 9600KF running stock and on 3200 ram. What a difference! GPU’s are now running near 100% and cpu sitting around 90%.
 
So its in, frame rates have more than doubled! Able to crank most settings to ultra or ultra high on wildlands now.

Went for the 9600KF running stock and on 3200 ram. What a difference! GPU’s are now running near 100% and cpu sitting around 90%.

Nice increase. Confirms you were very much chocked beforehand hand.
 
Nice increase. Confirms you were very much chocked beforehand hand.

Wasn't that expensive of an upgrade either, managed to sell my old cpu, board and ram for £230 so all in all this upgrade cost me £230! Bargain!
 
Back
Top Bottom