CrossFire and SLI frame rates do not reflect reality because of lack of synchronization!

[CP5670]

As I remarked earlier, my experience has been that the actual impact of this issue increases with the framerate, and you get more performance "chopped off" at high (constant 60+) framerates than low ones even if the end result is still greater. That might explain why more people haven't brought it up. Many multi-GPU users prefer to turn up the AA or other settings to get very high levels of IQ while maintaining a tolerable level of performance instead of getting a very high framerate at the same settings.

Also, from the comments I've seen around here in the past, there are many people who cannot make out differences beyond 30 or 40 fps, and many more who have 60hz LCDs and won't see over 60fps anyway. These users may not notice the problem quite as much. IIRC I saw it most prominently at 70fps or more, where the perceived level of smoothness was nowhere close to what a single card provides at the same framerate.

is this problem getting a relief if vsync and triple buffer is used?

That is how it appeared to me. If you notice this problem, you should certainly try turning those on, although you may have trouble getting them to work.

look, i have played with Single GPUs and Crossfire .. there IS a definite advantage to playing with one - IF it is fast enough for you. BUT if you need to go faster, the disadvantages of Crossfire are far outweighed by the advantages and i can handle the little "oddities" for overall faster [and smoother - for me] FPS; and i won't compromise on detail.

The question is exactly how much of an impact this effect has. I think the raw framerate increase is usually enough to make things substantially faster in the end anyway, but on SLI I've seen at least one situation (in SC: Chaos Theory) where two cards were pulling about double the framerate of one but I couldn't see any improvement at all in the smoothness.

 

[apoppin]

Originally posted by: BFG10K

agreed .. what they do now works OK .. but if you are "sensitive" ...
. . . get an Xbox

A better idea is to get a GT200.

count me "in"

Day One

i am very sensitive to Xfire 'oddities' - i would MUCH prefer a faster single GPU - and would die with an X2 i think

The question is exactly how much of an impact this effect has. I think the raw framerate increase is usually enough to make things substantially faster in the end anyway, but on SLI I've seen at least one situation (in SC: Chaos Theory) where two cards were pulling about double the framerate of one but I couldn't see any improvement at all in the smoothness.

i have seen that also - *not usually* - but even in canned benchmarks like 3DMark06 when a single old 8800GTX will look smoother than 2900 Crossfire running about the sames and yet score less !
.. and this sounds like the basis for an entire Review Article on the Main Site

*cough*



. . . for someone else to do .. i have done three-in-a-row half-ass "reviews" here in video and i like to actually PLAY some freaking games for a change.

You think we have to wait for Derek or Anand to explore it?
- it looks like a Big One

 

[CP5670]

count me "in"

Day One

Same here. I've been waiting on this card since last August.

.. and this sounds like the basis for an entire Review Article on the Main Site

*cough*



. . . for someone else to do .. i have done three-in-a-row half-ass "reviews" here in video and i like to actually PLAY some freaking games for a change.

You think we have to wait for Derek or Anand to explore it?
- it looks like a Big One

Someone should definitely point Derek to this thread. It would make for a very interesting article and would clear up this issue for good.

[edit] man, you're editing your post every time I reply to your last edit.

 

[apoppin]

The Only Thing that will derail my plans is if r700 literally blows my mind with exceptional performance and features that make it compelling.
- i must qualify this before i get into trouble with "predictions"

and well .. #2 i could be broke ,,, the economy is going to hell right now
= but strangely, i am also hopeful - beyond hoping

exciting times indeed!

 

[CP5670]

I'm not that hopeful of the R700. All the indications at this point are that the top end card of that line will be a multi GPU solution. Unless their single GPU version provides comparable performance to the GT200 (in which case they wouldn't need to release the X2 version right away), I can't see myself considering it.

 

[apoppin]

[edit] man, you're editing your post every time I reply to your last edit.

i am sorry .. but i also see others have replied and it just looks like i am slow
--actually, i am pretty quick .. my connect is so slow .. 50 frickin' six dialup :|

i should learn to "preview" .. but that ALSO takes too long
-its' lose-lose for me .. satellite is $80 a month and the ping sucks worse than 56K.

i try to make up for what i lack in HW with technical brilliance and analysis
---well, i try .. i say nothing about succeeding - or emoticon overload


:laugh:

i am not the slightest bit hopeful about r700 either - EDITED!!

 

[phexac]

This is interesting, but the reason I have stayed away from SLI is that 1 high-end card is enough to run pretty much every game out there at max settings at a good framerate. So why do I need a 2nd card? The one exception is Crysis. Good job making a game that no one can play at max settings for the next 1-2 years, SLI or not. I'll be buying that game when there are cards that can actually run it at max.

 

[nRollo]

No offense, but I don't buy this.

The whole "framerates are much higher but the actual performance is much lower" thing just isn't true. I've been running SLi since there's been SLi, on lots of different computers. I know for a fact that I've run many high AA, high res settings that no sinlge GPU solution could come close to, and run them silky smooth.

I imagine there is deviation from the time of rendering the frames between AFR and single card, but I don't think there is any "scandal" going on.

Has it occurred to anyone if AFR actually performed worse one reputable tech site would have noticed it by now? This may be being discussed by some guys "on the German forums", but all I have to do is disable GPUs or take out a card to see things run choppy at my settings with single GPU and run great with Multi.

Edit:
One small correction- back in the days of the ATi MAXX- I did notice, and see reported, problems with synchronization, but that was a long time ago. I'd bet good money Crossfire AFR behaves like the SLi AFR I'm used to these days.

 

[Datenschleuder]

Guys, if you don't believe me, than make the simple test based on frame time measurements like I did.
Just claiming that everything is fine and that this problem is no big deal is nothing but ignorance.

Every second frame on SLI/CrossFire is much slower than the frame rate claims.
That this (of course) matters, because it is destroying smooth frame rates, is demonstrated in this video.
And even if you don't notice it in your scenarios - most of the benefits by the second graphics card are eliminated - the frame rate values are much lower in reality.

Log the frame times with AFR and post them as I did if you want to make a point.
Everything else is just slip slop talking.

I don't want to offend anyone, but we really have to get back to a discussion based on facts.

 

[apoppin]

Originally posted by: Datenschleuder
Guys, if you don't believe me, than make the simple test based on frame time measurements like I did.
Just claiming that everything is fine and that this problem is no big deal is nothing but ignorance.

Every second frame on SLI/CrossFire is much slower than the frame rate claims
That this (of course) matters, because it is destroying smooth frame rates, is demonstrated in this video.
And even if you don't notice it in your scenarios - most of the benefits by the second graphics card are eliminated - the frame rate values are much lower in reality.

Log the frame times with AFR and post them as I did if you want to make a point.
Everything else is just slip slop talking.

I don't want to offend anyone, but we really have to get back to a discussion based on facts.

no i don't believe you and i think your "theory" is all wrong
-neither do i *care* what your video shows .. i can make anything look bad

i run CrossFire

Perhaps there is something wrong with your set up

List all the HW components in your rig please including OS and drivers; perhaps we can help you

are you now calling me ignorant? i got a little testy with your "science" earlier because it IS B .. unfounded .. sorry for my crusty manner

 

[semisonic9]

Originally posted by: apoppin

If you don't play twitch shooters and don't notice microstutter then SLI + large LCD is not a bad way to fly. For the rest of us I'm glad there are single card solutions and CRTs.

i do notice and i also play twitch shooters ..
... everything is compromise .. but the *cure* for most of us - except to buy a faster single GPU [which for example, could not happen if you had purchased a 8800GTX-Ultra back last - last November. You still can't buy a single faster GPU - it will be TWO years!! .. So ..
If you want "faster" then your choices are - (1) to lower your resolution, (2) sacrifice details - OR (3) buy a second one

Problem Solved for most of us by #3
- the least compromise

/\ This.

The OP has a point about how stats are reported, but I believe most of this information has already been widely available. It's unrealistic for most sites/magazines/etc that review hardware to take up time, energy, and column inches breaking down the basics of SLI again and again.

~Semi

 

[apoppin]

Originally posted by: semisonic9

Originally posted by: apoppin

If you don't play twitch shooters and don't notice microstutter then SLI + large LCD is not a bad way to fly. For the rest of us I'm glad there are single card solutions and CRTs.

i do notice and i also play twitch shooters ..
... everything is compromise .. but the *cure* for most of us - except to buy a faster single GPU [which for example, could not happen if you had purchased a 8800GTX-Ultra back last - last November. You still can't buy a single faster GPU - it will be TWO years!! .. So ..
If you want "faster" then your choices are - (1) to lower your resolution, (2) sacrifice details - OR (3) buy a second one

Problem Solved for most of us by #3
- the least compromise

/\ This.

The OP has a point about how stats are reported, but I believe most of this information has already been widely available. It's unrealistic for most sites/magazines/etc that review hardware to take up time, energy, and column inches breaking down the basics of SLI again and again.

~Semi

That IS what i am saying .. yes, there could be an *entire article on the Main Site* devoted to explaining it

or we could do it ourselves

personally i don't care .. i "know" and i much prefer CrossFire to a single weak-ass 2900xt for DX10 games

and it will be "moot" for me in a couple of months, anyway - the Companies also "know" and evidently are working on it [my personal belief is that AMD is using AFR but also something "else" - undocumented] - and i will be buying GT200 GTX .. so .. i am not going to do the review
[i hope]

 

[Datenschleuder]

It doesn't matter, but I run a 8800 GTX SLI system.
And this is not just a theory. This was replicated by dozens of SLI and CrossFire users in the linked German forums and the linked German hardware review site as well.

Simply test it out on your own.
It only takes 5 minutes to do this, so what's the big deal?

 

[apoppin]

Originally posted by: Datenschleuder
It doesn't matter, but I run a 8800 GTX SLI system.
And this is not just a theory. This was replicated by dozens of SLI and CrossFire users in the linked German forums and the linked German hardware review site as well.

Simply test it out on your own.
It only takes 5 minutes to do this, so what's the big deal?

i have; thank-you - Yes, yours is not just a "theory". I believe your entire premise - and "math" in the OP is simply fallacious. That is not how it works in practice nor theory.
- you have the burden of Proof - not a "video"

i believe we can help you and the others on the German forums
- my German is not as good as your English

List your HW components/OS and Drivers .. or stop bugging us, please
-- we are already well aware of the oddities of multi-GPU and have been discussing it. There are several threads devoted to micro-stutter as one issue. Some of us *hate* CrossFire and SLi; most of us can "live" with it

i would *die* if i regularly have your issues and i think we can help you if you stop calling us stupid for not agreeing with you.

I'm sorry if i did not give you a better welcome. Is it too late?
-welcome to discuss - not to challenge

 

[Datenschleuder]

i have; thank-you - your premise is fallacious

So where is your frame time log?
Post it please.

i believe we can help you and the others on the German forums

List your components .. or stop bugging us

I am a graphics programmer and looked into this much deeper than just figuring out possible configuration problems.

I even got a reply by a Nvidia developer support person and he confirmed the existence of the problem and he said: "we can't fix this issue at the moment because of kernel space issues" - that was two years ago!


EDIT:

and "math" in the OP is simply fallacious. That is not how it works in practice nor theory.
- you have the burden of Proof - not a "video"

Dude, just claiming that someone is wrong, without delivering any arguments is quite annoying!

And you clearly haven't read my posts thoroughly either, because else you would know that I did post a link to a video several times already.

 

[apoppin]

Originally posted by: Datenschleuder

i have; thank-you - your premise is fallacious

So where is your frame time log?
Post it please.

i believe we can help you and the others on the German forums

List your components .. or stop bugging us

I am a graphics programmer and looked into this much deeper than just figuring out possible configuration problems.

I even got a reply by a Nvidia developer support person and he confirmed the existence of the problem and he said: "we can't fix this issue at the moment because of kernel space issues" - that was two years ago!

...and i am a HW reviewer, so what?
- i think i look "deeply"

Show us your frame log .. i am not the one complaining about "CrossFire issues"



i think something is wrong with your system or you are just very sensitive .. i am also very sensitive but i report for 95% of people - not the elite few who are like "audiophle Tweaks" and i think you should have your own forum

actually there is one -- and i love them like brothers as they discuss the *esoteric* .. i believe you would fit right in .. but please don't tell them i sent you if you are going to be aggressive. The are even more "sensitive" to 'annoying' than we are

http://forum.beyond3d.com/

everything in Video Graphics is *compromise* and it presents an ILLUSION of smoothness and realism. From my personal observations .. i can *see* this microstutter .. i can demonstrate it and i can replicate it .. but it is something i can IGNORE in favor of the OVERALL improved experience with Faster FPS and USUALLY it is OK - even for me

TWO YEARS AGO it was worse!!!

and i said i believe AMD is no longer strictly limited to AFR .. i can see the IMPROVEMENT with my OWN EYES over the last 3 or 4 driver sets including 8.4

i have ZERO recent experience with SLi - but plenty of us do and would also be qualified to assist you

are you going to cooperate by posting your specs including OS and drivers? We can run you thru some exercises to pinpoint it and maybe make it tolerable for you

it is NO "scandal"

 

[Datenschleuder]

Show us your frame log

Apparently you don't read my posts at all.
I did this in the first post already my friend.

I won't waste my time anymore here on replies that only contain wild claims.

THANKS for keeping this thread intelligent and reasonable!

 

[Keysplayr]

Originally posted by: Datenschleuder

Originally posted by: Sylvanas
I can't notice anything like this so it doesn't phase me....

You don't care that most of the money you spent on SLI/CrossFire is going down the drain because of this?

So your saying that Crossfire/SLI offers no benefits in performance in both framerate and Image quality settings? Are you saying that people who buy a second card to Crossfire or SLI are in fact wasting there money because they are not getting any performance gains?

How does this affect people who do Xfire/SLI? Since when does 1 second have to equal 30fps? Especially in Crysis with minimum framerates dropping to teens on almost any graphics card setup out there?

I agree with Apoppin here. Full of "BS theory".
Placebo effect? Give me a break dude. That's like saying people who Xfire or SLI only "think" they're getting better performance, and the increase in fps, res, IQ, don't actually exist.

Holy **** is all I'm going to say here.

I'm currently switching back and forth from a single 9800GTX to SLI'd 9800GTX's.
I see the differences first hand, and must say, they are rather large. But I guess I must be imagining it if all I am doing is swallowing this placebo here. What was I thinking?

 

[Datenschleuder]

So your saying that Crossfire/SLI offers no benefits in performance in both framerate and Image quality settings? Are you saying that people who buy a second card to Crossfire or SLI are in fact wasting there money because they are not getting any performance gains?

How does this affect people who do Xfire/SLI? Since when does 1 second have to equal 30fps? Especially in Crysis with minimum framerates dropping to teens on almost any graphics card setup out there?

There are still some real gains with SLI/CrossFire of course.
But the point is that these gains are immensely lower than the frame rate improvements over single card setups suggest.

 

[apoppin]

Originally posted by: Datenschleuder

Show us your frame log

Apparently you don't read my posts at all.
I did this in the first post already my friend.

I won't waste my time anymore here on replies that only contain wild claims.

THANKS for keeping this thread intelligent and reasonable!

You are wasting our time

you are the one making outrageous claims that is suited for a tin-foil hat conspiracy

you have no standing here; no history .. and now - no credibility

YOU have posted a single video that could be your own HW/SW issues and we don't know if you are really running XP/XP64/Vista32/Vista64 or even a Win2K variant

you talk about 2 year old info from NVIDIA and there is Kernel space available now unless you are not keeping up or do not understand

ATI does Multi GPU different from NVIDIA

We already are aware of the issues, we understand it and we can usually help someone MINIMIZE it

Now - if you are not trolling, what are you doing please?

There are still some real gains with SLI/CrossFire of course.
But the point is that these gains are immensely lower than the frame rate improvements over single card setups suggest.

Thank-You

not immensely - something is probably wrong with your set-up - now if you don't want help and just want to argue - without providing ANY more evidence or even your own specs [at an absolute minimum], i respectfully suggest you go away on this topic. I also accept that you and some of the other German-speaking tech forum members are also "overly sensitive" - but you will not accept ANY PoV other that your own - you cannot believe that we are also very experienced? We are arguably the overall No1 Video Tech forum certainly top 5! I am also finished here unless you please cooperate.

 

[Keysplayr]

That makes zero sense. Are you talking scaling? But that doesn't even make sense.
Lets break this down.

"These gains are immensely lower" What gains are you talking about? More fps is more fps no matter how you slice it. There is generally never a 100% gain when going crossfire/SLI, but 80% improvement seems to be at least the norm under titles that support Xfire/SLI.

"The frame rate improvements over a single card" Huh? As opposed to zero cards?

 

[Datenschleuder]

Originally posted by: keysplayr2003
More fps is more fps no matter how you slice it.

The whole point is that the FPS gain becomes insignificant when the frames are not updated homogeneously.

Frame #0: 00.0 ms
Frame #1: 00.1 ms should be: 16.6 ms
Frame #2: 33.3 ms

So in this (exaggerated example):

- the frame rate says 60 FPS (1000 ms / (33.3 ms / 2))
- but the reality is that it is only 30 FPS (1000 ms / 33.3 ms)

Because Frame #1 update is not significant (it follows almost immediately after the previous frame), and Frame #2 takes 33.2ms to update, which equals ~30 FPS.

Read my first post, where the issue is explained with a real example.

 

[Jax Omen]

damn I wish I had the "don't feed the trolls' ASCII to post here...

 

[v8envy]

Well, you're still getting the higher frame rates, on average. You still have a better chance of having a frame be completed in time to be shown on an LCD monitor @ 60hz than you would with a single card. Your worst case performance won't be any better than a single card, but your likely and best case performance is much better. The only problem is the worst case happens regularly, as do the likely and best cases -- causing you to perceive stutter.

AFR issues have been known and described and analyzed and well understood for years. You knew (or should have known) multi-GPU solutions were not perfect, so I don't understand why you're trying to stir up a crusade now.

My solution to not being able to run a game at acceptable frame rates with a single GPU card is... to wait until I can. That's a completely different compromise than apoppin e.g. would make, but that's the beauty of this world. We can all do what makes us happy.

And btw, apoppin, the 8800GTX/Ultra was not a 2 year wait for a better card -- that GPU is still capable of manhandling ALL games except Crysis even at obscene resolution. So it's more of a ~6 month wait for a better single GPU, depending on when NV coughs up the GT200. And yeah, put me down for one more body not expecting miracles from the R700. It should compete nicely with the 8800GTS and 9800GTX once in the $200-250 AR range though.

 

[CP5670]

Well, I think the OP is correct and am pretty sure I've seen this issue myself, but I do agree that the way he presented this topic and is arguing with everyone is hurting his credibility.

Has it occurred to anyone if AFR actually performed worse one reputable tech site would have noticed it by now? This may be being discussed by some guys "on the German forums", but all I have to do is disable GPUs or take out a card to see things run choppy at my settings with single GPU and run great with Multi.

They probably have not actually looked. It's fairly subtle if you don't know about it and not something that would be readily apparent while just letting the benchmarks run. The reviewer would have to be sitting at the computer the whole time and looking for things like this. It's certainly worth bringing to Derek's attention in any case. I would like to see one of the major hardware sites look into this and related issues, whether they confirm or debunk them.

/\ This.

The OP has a point about how stats are reported, but I believe most of this information has already been widely available. It's unrealistic for most sites/magazines/etc that review hardware to take up time, energy, and column inches breaking down the basics of SLI again and again.

I've never seen this reported by any hardware site until that PCGH article. I had my suspicions for a while and some other posters here had made comments to the effect over the years (most recently n7 in his 3870 X2 thread), but that was about it.