Hi, I recently added another vid card for SLI. so I can do some testing and see how it reacts with SLI Enabled and SLI Disabled..

I also game in triplescreens and got the SLI setup for the benefits cause I also game in Stereoscopic 3D like a 3D Movie..

I was getting an out of memory error when enabling S3D and have gotten that resolved now plus i bought 8gb more memory for a total of 12gb so that should fix that for sure.

but I will definetely test the SLI enabled and Disabled and see how it goes..

and from what I gather when SLI is enabled 1 card will render the top half of the view and the 2nd card will render the bottom half resulting in each card rendering less then the full screen on 1 card.. so in theory it should be faster but who knows I will do a NON Scientific test with the results for all too see so stay tuned..

CrossFire works with the last patch. At first I got the strobe effect again using it but I got that to dissappear rolling back from 11.6 to 11.5b. For whatever reason 11.6 gave me that annoying strobe again. I could prolly have tweaked it away but rolling back solved it. Catalyst AI needs to be on for crossfire to work.

Again crossfire works just fine with the latest upgrade to RoF and has been verified running a diagnostic to confirm activity with both GPU's while in game.

The annoying crossfire LOGO is there in the top right corner. I turned off the Show Crossfire Status Icon switch.

Nephilim - unfortunately, this is another of the many misunderstandings about how SLI works (and in RoF, for those complaining about that...)

What you're referring to is called "Split Frame Rendering" (SFR) as I understand it, and I believe one of three modes (last I recall) that SLI can run in. There is also Alternate Frame Rendering (AFR), wherein each card tries to do every other line on the screen (I think) and AFR2 (an alternative for those apps that don't work well in AFR). At least that's how it was explained when I was studying it.

Jason has indicated RoF uses AFR, although he doesn't distinguish whether AFR or AFR2.

As an aside, all these modes can be enabled as I recall using the SLI configuration in Nvidia Control Panel. So, it might even be possible to disable SLI in RoF and still enable it through NVCP, and doing so might offer better performance or solve some of the problems reported with the 1.019+ support for SLI.

I can tell you for sure that SLI can be run on games without any 'support' built-in, and still have access to the configuration, various modes, and full benefit of SLI.

Thank you, sir. You've touched on one of the (several) points - all specific to SLI in RoF - that I have been trying to make.

Regards,

ok but technically I wasnt wrong it was 1 way of how sli does work just not in ROF thnx for the clarification.. I will try having sli enabled in the ROF config and in nvcp to see if that makes a diff..

the main reason I went with a 2nd vid card was the F@H Kicks but with 2 vid cards..

I don't know what point you are trying to make, I really don't.
It seems to me you think sli will produce a performance boost in RoF even without optimisation and that the findings of the developers and beta team are somehow flawed despite the hours of testing put in.

I suggest you try Sli in RoF before catagorically stating it will yeild improvements in RoF without the need for any optimisation first.

I'm on the beta team and was involved in testing the release candidates for 1.020.
Pre 1.019 neither SFR or AFR produced any performance increase over one card config on my my 480 GTX sli setup
Based on testing using a benchmark mission provided by the Rof devs. Which is run in auto pilot and reading taking from a predefined start stop point in the mission.
Using a single card and my Sli setup I produced identical min/max/avg readings for FPS.
Testing included use of AA/AF and SFR and AFR.
I tested this as part of my testing in the beta for multigpu support in 1.020.

On the various alpha/beta/RC's for 1.020 I did the same tests using the same benchmark mission.
There was a slight increase in fps, read very slight 2-3 fps at most using all manner of RoF/Sli settings and sli in both rendering modes (this is excluding in game super sampling) on my set up.

Some lower end cards, 260GTX etc. did produce some noticable gains but the higher end 480's and 580's didn't.

Only when checking the in game super sampling option did my PC show any signs of a performance boost using Sli. I gained around 30% increased frames, but even then it is not enough to make that option playable for me.

The bottom line is what Jason said in his graphics primer is based on a lot of testing, not just made up on the spot. The RoF devs did a lot of work on the code to implement some form of multi gpu support. The results of which are in game now and Jason's statement is 100% correct in what he says. Higher end NVidia cards will not see much benifit whereas lower end NVidia will.

Queeg

I've said it in this thread so many times now but

is exactly right.
The higher spec the card the less the performance boost will be

From a NVidia point of view, what works best 2 x 260 or 1 x 480?, a single 480 everytime.
If you already have a 260 and can get a second 260 very cheaply, then it might be a good idea. If you have a single 480/580, then it is not worth getting a second card just for RoF.

Yeah, I know Catalyst AI needs to be on and yeah, I know how to turn off the Crossfire Status Icon.

I'd love to see a posting of your setup with ROF running AND the Crossfire Status Icon displayed in the upper right corner-at the same time. Maybe it's the 11.6 drivers.

I assure you it works, you need to have faith in people or start associating with people who are honest.

Off to get a print screen ...

***edit*** Sorry it took me like five tries to get posting pic's right but I finally got it.

Brigstock,

We are still discussing SLI and RoF (just for the record). The extensive testing you conducting was certainly...extensive. However extensive it may have been, though, the entire focus of the testing - as you've now described - was to determine increase in FPS. This is another of the points I've discussed earlier: Increased FPS is not necessarily the only improvement SLI provides, nor is that the intent. Performance consists of more than just FPS - and I'll say this again, high FPS doesn't necessarily equal best performance.

Specifically, and in fact, there are SLI modes devoted entirely to providing higher levels of AA (SLI8X, SLI16X) - this has nothing to do with increasing FPS. And, as before, all discussed on the Nvidia website. In short, your testing was flawed because it was limited to an effect that isn't necessarily the function of SLI.

Another few tidbits, directly from the Nvidia website:

"The amount of performance improvement will depend on the application and its ability to scale. ...In general, applications running at higher resolutions with higher image quality settings will benefit most."

This should augment your concession above, then, that if there's a problem with SLI in RoF, it's not with Nvidia or the SLI technology. And it also goes to show that overall quality and resolution, not necesarily higher FPS, are the areas where SLI offers the most improvements. Your tests don't seem to have been geared toward establishing the difference in anything other than FPS.

Again, from Nvidia:

"SLI technology can be enabled for every gaming application, including both OpenGL and Direct3D gaming applications. SLI technology provides either 3D performance scaling using alternate frame rendering (AFR) or split-frame rendering (SFR) or increased visual quality using the SLI Antialiasing mode."

This confirms, once again, that SLI can be applied to quality rather than just higher FPS. It also shows - conclusively, at that - that SLI can be used in any gaming app (though of course, benefit will vary).

I still would like to know why, if Nvidia knows about the problems, why there would've been all this "wasted time" we hear of trying to implement support for SLI, where - according to Nvidia themselves, it isn't necessary. I don't recall a lot of requests from the 'community' for this.



Following the above quote, I would still like to know what confirmation from Nvidia indicates their acknowledgement that their higher end cards don't benefit from SLI as much as their lower end cards. Certainly their website seems to imply otherwise. For me, this seems to be a limitation only with RoF's implementation.

My point here? Well, it's pretty straightforward: Whatever limitations, problems and "wasted time" there may be with RoF's implementation of SLI, it seems pretty obvious to me that they are a product of RoF's developers, and not a failing of the SLI technology itself. TBH, I was avoiding coming right out and saying that, because I'm going to get jumped on for it. But you've failed to get the point I've tried to make indirectly - so, there it is directly. All this business about graphics primers and microstutters just doesn't add up to (or explain) what people are generally reporting with regard to the SLI support in 1.019.

And I have serious doubts that Nvidia would admit their high-end cards don't function as well in SLI as their low-end cards. That alone, to me, indicates something isn't right in "RoF Land".

Incidentally, I did check, and have found at least one example (so far) of someone who said they were successfully using SLI in RoF before the 1.019 SLI support. It seems like what the changes did, more than perhaps anything else, was cause problems.

Regards.

Your posts never cease to amaze me, KB ...

You infer your conclusions from what nvidia says, and what ati says. Instead, the RoF devs practically saw before the latest patches that people with sli/x-fire setups didn't have the performance they expected to have, so they did 'something' to make this performance appear. And Brigstock clearly says that he talks about RoF, how it worked before in his case, how it works now, what are his observations, all inside Rise of Flight. Nobody gives a #%&*$# about what 'sli should work without any modifications' - in practice, that didn't happen, so it had to be addressed, and the devs did it how they could and stated what the results were, how the performance would increase. Again, nobody gives a dime about the theory - the inferring and the whole discussion you have built on it (actually a monologue) is entirely yours. Just pls don't come here to criticize somebody which is actually doing something really useful, just because 'you know' that 'in theory' (maybe the commercial one) things already work automatically.

If your problem is with 'why RoF doesn't have sli/crossfire' automatically working as the nvidia/ati theory says, than say so, and that's another question. But, while you seem to want to ask that, you don't ask it, but you presume somebody already gave you a negative answer, and start to argue with him.

Know breaks I wasn't just testing FPS I was testing AA/AF too. Please read my posts. The was no impact on either with two cards.



Nobody is saying that, you presume too much. This whole thread is about Sli/Crossfire support in RoF, not about how well sli works in general. The higher spec/lower spec thing is a purely a RoF thing not an NVidia thing. I've told you this before but you just aren't listening.



Who? I have seen others post saying the same without checking to see if it was infact working. Just because the game runs with 2 cards enabled does not mean anything.

I also remember being told Crossfire worked great in IL2 by the Ubi forum moderator, who was running 2 x 3750's at the time. On the strength of that I went and tried crossfire only to discover it wasn't supported at all.

Just for your reference my job title is senior desktop consultant. I work for a tier 1 investment bank in the city of London on desktop support. I know a little bit about computers, enough to make a handsome living. I can assure you that I have tested sli capability in RoF quite thoroughly.

Before you start going off on one again, I suggest you stop theorising and get yourself an sli setup and test yourself instead of patronising me.

CF works great for me as well.

11.6 drivers, 2x6950 2Gb

Pic of the CF logo -



Settings I use -

KB:

What setup and settings are you using for ROF? I'd appreciate hearing your experience with SLI/Crossfire in the game. Thanks.

Brigstock, Queeg et al:

As life would sometimes have it, I have learned today that I must leave town on an urgent business matter. Although I'm sure you will not miss me, I will definitely look forward to continuing the dialogue at the earliest opportunity. Please feel free to leave any further comments here, and I will gladly respond as I can. Just didn't want you guys to think I'm 'running out' on this discussion. I may or may not be able to get back to this while I am indisposed.

Brigstock/Laser - if you'll check, you'll note I mentioned earlier in the thread I actually have used SLI before, in three separate builds. So, this isn't a theoretical discussion. It's based upon my own expeeriences with SLI and the facts readily available at the Nvidia website.

Queeg, as I explained above (and as shown in my sig) I'm not currently using SLI. I do have an old SLI set (two identical cards); real life permitting I do intend to do some study with/without SLI in RoF. However, a few things to note here:

1. I'm honestly not sure - given the 'forced' updates from RoF - how I can go back and forth between pre-1.019 and post to do proper testing. Maybe you could tell me something that would help.

2. It's been made abundantly clear that the higher-end Nvidia cards don't offer any improvement in RoF with SLI. But, I'm being urged to do this and try it for myself....now, I can't imagine why anyone would spend money to build such an arrangement. Note I use a 570 in my sig, which performs plenty well enough in the few games I play. If there's no SLI improvement in RoF, tell me: where's the motivation to spend $250-300 for a 2nd 570? Even if I do spend this money strictly for this testing, if it works like you say then a. I'll have wasted the money, and b. it won't change the fact that it's a problem with RoF, not SLI.

On the other hand, if I don't spend the money, you guys will claim I can't possibly have anything to contribute here because I'm not currently running 1.019 *and* SLI *in* RoF. Although it's not true, that's what you're claiming. I know enough about SLI to know it's not causing the problems, and I know it's not 'typical' SLI behavior for low-end cards to increase performance more than high-end cards. Again, something's wrong in RoF (and nowhere else).

Brigstock - the poster who said they used SLI prior to 1.019 is above, in this very thread. That's only one, but I'm sure there have been others; I'll have to get back to you. (You could look for yourself, though I don't expect you'd be interested in that, right?) What I *do* expect, though, is you'll dismiss (yet another's) experience simply because it doesn't fit your design. Nonetheless, he reports what he does.

Also, re: your testing: You explained above that the strict focus of your test was FPS. In fact, as you described it, FPS was the *only* actual measurement. And, without measurement, testing is useless. You can't go back and change your story now to claim you also tested the various forms of SLI AA, various resolutions, etc - all of which are various improvement SLI is proven to offer (except in RoF, apparently). You indicated you were testing for FPS increase (which doesn't surprise me, because it's a typical misconception that SLI only offers improvement in the form of higher FPS - which I've now demonstrated simply isn't true). You said "Testing included use of AA/AF and SFR and AFR.", not that any comparisons were done between the modes, and not that anything that was measured (and these are not things you can measure with FPS, regardless). Specifically, you omitted any mention of the unique SLI AA modes. Sorry, but as I said earlier, your testing - *any* testing that only measures FPS - is not a valid full test of SLI performance.

All the best, gentlemen.

KB

Is that the problem, you have made a statement and just can't back down, even when the evidence is against you. Looks like you want to run this right into the ground. The real issue here though is misinformation. I say one thing based on real life testing and you say another based on theory and it is theoretical until you actually try sli in RoF. I've had Crossfire systems and Sli systems in the past but those systems are only relevent to the titles I ran back then. To say sli worked well in IL2 on my old PC and then to state it "should" also work well in RoF is not a sound statement.

My testing, it wasn't a strict FPS test. Testing was gauged using FPS to mark performance, but all aspects and graphic settings where tested and gauged by fps benchmarking. eg: a single card with 8xAA 8xAF and a dual setup using 8xAA and 8xAF, which produced the same results in RoF 1.019 and only very marginal improvements in 1.020 *on my system. All setting were tested, no aa, 2xaa 4xaa etc. Any change in graphic setting would have been tested sli off and sli on and a few tests carried out with one card in the PC and both cards in the PC.

As for why I'd spend extra on an Sli setup, I used to have sli in the past and it worked well for me at that time. In this instance though I bought the extra card purely to test sli in the beta and becuase I can afford to indulge my whims now and again. Which is why I'm being so vocal on this issue. Normally I'd not get involved in threads like this. But when someone like you says Sli will work in RoF and yield improvements, then I have to step in and warn that higher end NVidia will not see any real benefit. So people don't waste money

On pre 1.020 sli support, I saw that post, but it is the only one I have seen that reported sli to work in RoF pre 1.020. But as I said, did the user test in single and dual to see if there was an improvement. I've seen posts before where people say sli works because they have sli and the game runs, without testing to see if there is in fact any improvement in dual as opposed to single card.

I look forward to seeing your results, if you can get more from sli than the game developers manipulating the game code and the beta team testing, then I'd welcome it.

After two unsuccessful attempts with a GTX 590 (bad stuttering both times), I borrowed a friend's GTX 580 to try with mine. So far, the 2x580 setup is working well. No stutters and FPS at 50-60 (40-50 on the ground) in career mode with everything maxed (except HDR set at Med) and 4xSGSS via Nividia Inspector. Basically, the addition of a second card allows me to bump the SGSS up from none to 4XSGSS at the same FPS.

Not sure that's worth the cost of a second 580, but it is improvement.

My findings too. A boost when using super sampling and as you say does that warrant the cost of a new card

Running dual cards was always about the ability to have crazy high AA or max settings or crazy AA and max settings. Granted if you're happy with what you see on screen already there was never any need to upgrade but for those that wanted to max out their games then getting the best they can get is the only option. If you can afford it, why not?

Hmmm... might have to try SS. Already running max AA, maybe I'll tick SS and lower the AA a bit.

That's awesome guys (regarding Crossfire) and I'll try the posted settings. I'm running 11.6 and the damned Crossfire thingie in the upper right corner never came up. Seen it displayed in all of the other "Crossfire ready" games I run... WTF?????