• 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.

Prices make me tempted to go sli for fun

You'd be better off selling that 980ti, and putting what you would've spent on another into getting a 2nd hand 1080ti.
 
I've had sli 980 Ti and a single 1080 Ti with the 980s stock an overclocked 1080 Ti is just about ahead when sli is working however you can get an extra 20-25% when overclocking the 980s.

Though when sli didn't work the 1080 Ti was a huge jump in performance. Personally I'm waiting to see what the 2070 can do, and try and bag a 1080 Ti for less than £500 sold mine for £700 at the beginning of the year. Plus you'll get 11gb Vram which is helpful if you're running a higher res I.e 4k like me
 
ALXAndy said:
I am just waiting for Nvidia to pull all support for SLi. AMD sort of did with Crossfire, by declaring the Radeon Duo thing (Fury) was not meant for gaming.. They never seemed to mention Crossfire again after that.

SLi IMO was always better than Crossfire but even on my Titan Blacks it was starting to lose support and I ended up trying to play on a 4k screen on one, which was not sufficient. Since then? it's not gotten any better.

Hopefully this NVlink thing can make the cards work together (by adding the RT units together) because I know Navi is supposed to be something similar (like as an example four small polaris cores in one die tied together..). IDK, we will see, but I would avoid using more than one GPU for anything now unless you are rendering and it is worth the extra electricity.
Click to expand...

It works very well currently as it’s essentially what stitches all those massive deep learning boards together. The two cards would be seen as one by the application and communicate fully instead of just splitting rendering which tends to lead to all the glitches that people are talking about.

We shall see, it’s clearly the future over SLI for Nvidia.
 
Been running SLI permantly since 2006. 2x 7600GT/2x7900GT/2x7900GTO/2x8800GT/9800GX2/3xGTX280/3xGTX470/3xGTX580/3xGTX670/3xGTX780/3xGTX970/3xGTX980Ti/2xGTX1080/2xGTX1080Ti. All through those years I've only seen SLi getting better untill we hit 2016 it went downhill, 2017 was a blurry mess, and in 2018 it's actually back on track. The issues related to SLi I have had can be counted on one hand.
I am exited about the new NVLink tech and what it will mean ?... I hope we finally get stacked Vram so that 11GB +11Gb = 22GB and not 11GB again.
 
I have used SLi for a long while (2 x 4gb GTX680's) and while I have only just got rid of them last year they did my proud. Saying that for all the good they did they were not perfect. Hot, noisy and didn't work with all the games. Nvidia SLi support has just got worse. I own a 1080Ti Strix OC and am sooo happy with its performance at 2560x1440, everything turned up to max on 95% of my games and its just superb. I don't think I would go back to SLi again.
 
Gripen90 said:
Been running SLI permantly since 2006. 2x 7600GT/2x7900GT/2x7900GTO/2x8800GT/9800GX2/3xGTX280/3xGTX470/3xGTX580/3xGTX670/3xGTX780/3xGTX970/3xGTX980Ti/2xGTX1080/2xGTX1080Ti. All through those years I've only seen SLi getting better untill we hit 2016 it went downhill, 2017 was a blurry mess, and in 2018 it's actually back on track. The issues related to SLi I have had can be counted on one hand.
I am exited about the new NVLink tech and what it will mean ?... I hope we finally get stacked Vram so that 11GB +11Gb = 22GB and not 11GB again.
Click to expand...

From what I saw about Navi it is apparently "scaleable" even though that word nor scalable seem to exist. I would imagine what it means is AMD can basically put four little dies together and have them seen as one. Possibly using the Infinity Fabric they already have, possibly something else. However, that is what we expected going into the future as it made it possible and cheaper to create "big" dies without actually creating big dies and the risks involved.

Now I could swear Nvidia also said Volta was "scaleable". So they may be doing it externally for now using NVlink.

Now bear in mind I could be *completely* wrong but we all know that multi GPU support in DX12 is a bust. It just hasn't happened in the way MS said it would, not at all. If, however, Nvidia and AMD want it to truly work? then they need a way of making it work without requiring any bother from the devs at all.
 
Kered_Dub said:
For DX12 - if Devs now have responsibility for coding in SLI (mGPU) support its probable they wont bother in most cases...
Alas for console ports too, SLI support is unlikely these days.
Click to expand...

Yes exactly. This is why they need a way of it all being done "at metal" as it were. They need to stop making technologies that are game dependent.

I think in all honesty that is why I love my XB1X so much, because I have realised why PC gaming graphics have gone stale. Because all they were doing was making games that would run on the two consoles, then making them run in Windows. But, I would imagine "making them run" was literally about all they did. They certainly haven't done anything PC specific with them IMO.

So if I am going to be trapped playing console ports I may as well get a bloody console. At least there are less problems on day one.
 
You must log in or register to reply here.
Back
Top Bottom