1080 ti SLi problem

[zizo3434]

I just got myself an EVGA SC 1080 ti Black Edition in addition to my Existing MSI gaming X. Problem is that their heights are way different so that hard HB bridge is not possible . I read that 2 Flexible SLi bridges would do the trick but my problem is that one card is being 100% utilized the other can't go about 75% utilisation no matter what I do ! ??‍♂️ Although When using each card separately everything is ok. What Can I do ? Could it be one of the ribbon Bridges is faulty? Should I get more bridges and test?  Or am I doomed because an HB bridge is a must ? Thanks guys. 

[Zando_]

You should just be able to use a standard HB bridge, they go off slot spacing, not card height. And the SLI tabs are on the PCB which is always level with the PCIe slot, no matter the card. 

[i_build_nanosuits]

6 minutes ago, Zando Bob said:

You should just be able to use a standard HB bridge, they go off slot spacing, not card height. And the SLI tabs are on the PCB which is always level with the PCIe slot, no matter the card. 

no that's not true, if the card is wider then it means the SLI tabs are further out from the PCIe slot...like this card for example...

 

[Zando_]

1 minute ago, i_build_nanosuits said:

no that's not true, if the card is wider then it means the SLI tabs are further out from the PCIe slot...like this card for example...

Oh.... the PCBs themselves are wider. I thought you mean the cards were thicker than each other. I guess stick with the flexible ones for now, IIRC most of the time SLI doesn't scale to exactly 2x the performance, I'd assume because it doesn't use 100% of the second card, but then I've never used SLI. 

[i_build_nanosuits]

1 minute ago, Zando Bob said:

Oh.... the PCBs themselves are wider. I thought you mean the cards were thicker than each other. I guess stick with the flexible ones for now, IIRC most of the time SLI doesn't scale to exactly 2x the performance, I'd assume because it doesn't use 100% of the second card, but then I've never used SLI. 

yeah that's what he meant...if you have one card wider than the other then the rigid bridge wont work.

 

Also, i'm with you...you can't go in there and expect to get 100% load on both cards...that's just not happening, and having card #2 running at 75% is already pretty darn good for SLI.

[Mick Naughty]

Have you tested it in a benchmark or just in games?

[zizo3434]

2 minutes ago, Mick Naughty said:

Have you tested it in a benchmark or just in games?

I watch YouTubers testing games with the same settings and there results are better than mine . I'm really confused and in my country you can't find sli bridges locally unless you buy a full mobo! So before ordering internationally I need to know if it might be a Faulty/Partly functional ribbon sli bridge. ??

[i_build_nanosuits]

Just now, zizo3434 said:

I watch YouTubers testing games with the same settings and there results are better than mine . I'm really confused and in my country you can't find sli bridges locally unless you buy a full mobo! So before ordering internationally I need to know if it might be a Faulty/Partly functional ribbon sli bridge. ??

ohhh god...that's not how you test.

Go download some graphics benchmarks (Unigine heaven for example) and test it and see if you can max out your cards.

[Mick Naughty]

9 minutes ago, zizo3434 said:

I watch YouTubers testing games with the same settings and there results are better than mine . I'm really confused and in my country you can't find sli bridges locally unless you buy a full mobo! So before ordering internationally I need to know if it might be a Faulty/Partly functional ribbon sli bridge. ??

That didn’t answer my question what so ever. Trying to see if it’s a game/driver setting or a hardware issue. Thus, runnng benchmarks will show that. 

[zizo3434]

1 minute ago, Mick Naughty said:

That didn’t answer my question what so ever. Trying to see if it’s a game/driver setting or a hardware issue. Thus, runnng benchmarks will show that. 

Yoooooo! I tested the "heaven benchmark" and the usage is identical "mostly in the high 90s"! That's a good sign right??? But why not in games? I see YouTubers getting better utilisation. Here take a look the usage chart

[i_build_nanosuits]

Just now, zizo3434 said:

Yoooooo! I tested the "heaven benchmark" and the usage is identical "mostly in the high 90s"! That's a good sign right??? But why not in games? I see YouTubers getting better utilisation. Here take a look the usage chart

so many things will affect utilization of your cards in SLI...your system works as it's suposed to...SLI is what it is, you can't change it.

[Mick Naughty]

Need to turn up setting in the games. Or check the usage of the cpu. 

 

I just started having a similar issue. Same game I’ve played for years won’t max out my cards now. 

[Lathlaer]

OP, let's deconstruct your problem.

 

What are your specs - CPU in particular?

What settings are we talking about - what resolution? Type of AA (TAA has a notoriously bad effect on SLI scaling for instance)

What games? Give few examples. If I have them, I might be able to comment on the utilization.

Do you have NVIDIA Inspector installed? Often this is the crucial point, if someone is running a custom SLI profile for a game then they might reach higher utilization. Again - depending on the game, need those examples.

 

SLI is a hungry beast. You might say that it doesn't matter whether you have a top CPU because of resolution. You might say that x16/x16 isn't needed though even with top Pascal cards this has not been my experience in some situations. You might say you don't need a HB bridge and 2 ribbons are enough. 

 

But if you start adding all of that, you are essentially creating a ground for suboptimal experience with SLI.

 

Last but not least, yeah - a faulty SLI ribbon is not something that can be discarded as impossible. It does happen and yes, there are people who are binning SLI bridges.

[zizo3434]

58 minutes ago, Lathlaer said:

OP, let's deconstruct your problem.

 

What are your specs - CPU in particular?

What settings are we talking about - what resolution? Type of AA (TAA has a notoriously bad effect on SLI scaling for instance)

What games? Give few examples. If I have them, I might be able to comment on the utilization.

Do you have NVIDIA Inspector installed? Often this is the crucial point, if someone is running a custom SLI profile for a game then they might reach higher utilization. Again - depending on the game, need those examples.

 

SLI is a hungry beast. You might say that it doesn't matter whether you have a top CPU because of resolution. You might say that x16/x16 isn't needed though even with top Pascal cards this has not been my experience in some situations. You might say you don't need a HB bridge and 2 ribbons are enough. 

 

But if you start adding all of that, you are essentially creating a ground for suboptimal experience with SLI.

 

Last but not least, yeah - a faulty SLI ribbon is not something that can be discarded as impossible. It does happen and yes, there are people who are binning SLI bridges.

6700K @4.5 GHz . All games tested @2160p native. Games are rise of the Tomb Raider and crysis 3 . Utilisation go all over the place although as you can see in heaven benchmark usage was much better. So I'm confused. And starting to think the new Nvidia drivers made things worse to make touring look better maybe? I'm starting to feel it's not the ribbon Bridges although I won't completely discard the possibility. 

[Mick Naughty]

Well that’s a simple thing to check. Swap the drivers. They have them stored up to the past year. 

[Lathlaer]

I launched Rise of the Tomb Raider since I have a save there after finishing the game and you are right, one card does have a lower utilization. Thing is, I can't really say whether that has any effect on the performance since the FPS seemed about right for me, though I can't really remember well. 

 

It's late for me here but when I have a moment tomorrow I will swap the drivers for 398.11 and will check again.

 

EDIT: oh btw. I'm sitting on 399.24 now, not the 411.51

[pas008]

Did you use ddu?

 

If not I have to run into that not completely fixing it too

Windows reinstall fixed everything

Especially on couple creators shit 

But now i just backup my windows disk

After i have things running correctly

For new starting point before something fucked it up

 

 

And yes sli bridges can be bad 

Had a couple through the years

Both were ribbon ones though

Sli since 8k series

8600gt,8800gtx,250gts,gtx260,gtx460,gtx560ti,gtx670,gtx970,gtx1080ti

And prolly missing some short lived sli setups too

Inspector is important tool too

 

[zizo3434]

14 hours ago, Lathlaer said:

I launched Rise of the Tomb Raider since I have a save there after finishing the game and you are right, one card does have a lower utilization. Thing is, I can't really say whether that has any effect on the performance since the FPS seemed about right for me, though I can't really remember well. 

 

It's late for me here but when I have a moment tomorrow I will swap the drivers for 398.11 and will check again.

 

EDIT: oh btw. I'm sitting on 399.24 now, not the 411.51

Please let me know if you find any improvement so I'd reverse the drivers too. Thank you so much man . ??

[Lathlaer]

1 minute ago, zizo3434 said:

Please let me know if you find any improvement so I'd reverse the drivers too. Thank you so much man

Here is the thing - I am sitting constantly on 399.24 since the premiere of Shadow of the Tomb Raider. Previously I sat on 398.11 when I was playing RotTR. I haven't upgraded to the 411.70 at all.

 

I checked benchmark yesterday and it gave me about the same amount of FPS as expected (4213x1764, everything on ultra/very high, MSAA - 112 FPS in official benchmark). During the benchmark my GPU utilization was a bit wonky for a moment in scene 1 but then it was 100% on both.

 

When I loaded my save game from Geothermal Valley after finishing the game, I had one card sitting on 100% and the other about 75%. Hard for me to say whether this is normal or not because it's not like I played the game with MSI Afterburner OSD constantly visible. Maybe it's just that location.

 

When I get back from work I will test the previous driver which I used for playing RotTR (398.11) and then the newest one (411.70).

[Lathlaer]

4 hours ago, zizo3434 said:

Please let me know if you find any improvement so I'd reverse the drivers too. Thank you so much man

I tested 399.24 vs. 411.70. Didn't test the earlier because the 399.24 gave me the same expected results.

 

Official built-in benchmark from RotTR:

 

Settings: 4213x1764, DX12, everything maxed, SMAA

 

399.24

GPU usage in scenes: 1. Mountain Peak (65-70% in the beginning going up to 99% on both GPU's in the second half) 2. Syria (94-99%) 3. Geothermal Valley (99-100%)

Average FPS: 113.41

 

411.70

1. Mountain Peak (65-70% in the beginning going up to 99% on both GPU's in the second half) 2. Syria (94-99%) 3. Geothermal Valley (99-100%) -> the same

Average FPS: 113.77

 

I loaded an autosave from Geothermal Valley and in both instances I got 99% on both GPU's and about 110-120 FPS. 

 

As shown in benchmark, there is probably few locations where simply the utilization isn't that high, depending on the level of detail in scenery and whether it's CPU intensive or not.

 

Conclusion: the drivers are not to blame. If you are not getting the right performance then you need to look elsewhere for a reason.

 

Might be easier if we compared a specific location. You can tell me where should I go to check or smth (which camp etc) and I can do that.

[zizo3434]

3 hours ago, Lathlaer said:

I tested 399.24 vs. 411.70. Didn't test the earlier because the 399.24 gave me the same expected results.

 

Official built-in benchmark from RotTR:

 

Settings: 4213x1764, DX12, everything maxed, SMAA

 

399.24

GPU usage in scenes: 1. Mountain Peak (65-70% in the beginning going up to 99% on both GPU's in the second half) 2. Syria (94-99%) 3. Geothermal Valley (99-100%)

Average FPS: 113.41

 

411.70

1. Mountain Peak (65-70% in the beginning going up to 99% on both GPU's in the second half) 2. Syria (94-99%) 3. Geothermal Valley (99-100%) -> the same

Average FPS: 113.77

 

I loaded an autosave from Geothermal Valley and in both instances I got 99% on both GPU's and about 110-120 FPS. 

 

As shown in benchmark, there is probably few locations where simply the utilization isn't that high, depending on the level of detail in scenery and whether it's CPU intensive or not.

 

Conclusion: the drivers are not to blame. If you are not getting the right performance then you need to look elsewhere for a reason.

 

Might be easier if we compared a specific location. You can tell me where should I go to check or smth (which camp etc) and I can do that.

Dude believe it or not. Something is seriously weird going on . On Rise of the Tomb Raider I only test in geothermal valley btw. The thing is when I crank the settings all the way up but I keep AA @SMAA, the utilisation of both gpus is only around 75% on each as if my pc is trolling me "nananana no 60 fps @4K for you". But when I crank up AA all the way up to MSAA X4 The frame rate drops which is expected but the surprise is that the utilisation of buth GPUs becomes 99% !!!! How ??? Why ??? It's so confusing. Like why not just keep the 99% when I turn down AA and keep the high frame rate. ??‍♂️??‍♂️

[Lathlaer]

Can you run the benchmark and see what kind of FPS you are getting on average and in Geothermal Valley specifically?

 

Check if you have V-Sync on or off. If you have V-Sync on either in game or in NVIDIA Control Panel and you set your refresh rate in game to 60Hz, the game will not allow you to exceed 60FPS and this might be a bottleneck.

 

Unless you are not getting 60fps in 4K? Because you totally should.

[zizo3434]

4 minutes ago, Lathlaer said:

Can you run the benchmark and see what kind of FPS you are getting on average and in Geothermal Valley specifically?

 

Check if you have V-Sync on or off. If you have V-Sync on either in game or in NVIDIA Control Panel and you set your refresh rate in game to 60Hz, the game will not allow you to exceed 60FPS and this might be a bottleneck.

 

Unless you are not getting 60fps in 4K? Because you totally should.

No I don't get 60 fps @4K . I just get around 55 fps Average running around geothermal valley and utilisation of both GPUs is around 75% when V-sync is on with exclusive full screen on too ...... And btw when I turn off both v-sync and full screen "windowed mode" i actually get over 60 fps but it becomes a choppy stuttry mess. Then again if it was the bridges or a faulty one of them then why 99% on both when AA is cranked all the way up ?? You see where I'm coming from ? 

[Lathlaer]

3 minutes ago, zizo3434 said:

No I don't get 60 fps @4K . I just get around 55 fps Average running around geothermal valley and utilisation of both GPUs is around 75% when V-sync is on with exclusive full screen on too ...... And btw when I turn off both v-sync and full screen "windowed mode" i actually get over 60 fps but it becomes a choppy stuttry mess. Then again if it was the bridges or a faulty one of them then why 99% on both when AA is cranked all the way up ?? You see where I'm coming from ? 

Well, yeah this explains it a bit. How is the utilization when you turn v-sync off? And how is the FPS in fullscreen then?

[zizo3434]

12 hours ago, Lathlaer said:

Well, yeah this explains it a bit. How is the utilization when you turn v-sync off?

Problem solved man . It was the stupid Windows 10 "Game mode" hindering and crippling SLi performance instead of improving it. LMAO Can you believe it? I tried everything possible and that was just the problem. ??????‍♂️ The one you activate pressing Windows Key + G .