• 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.

The dreaded microstutter

crinkleshoes said:
The only way to get rid of it is to switch to team green I'm afraid... their multi-gpu implementation is noticeably better for stutter.

It's still not as good as a single card, but it's close now at least :)

ATi still have a way to catch up in this department :(
Click to expand...

I would love a new card, but i paid 130 quid for the 2nd 6950 brand new, and the performance except for stutter i had before has been brilliant, i just dont think 350 quid for a 670 or 7970 is worth it when my 2 6950s which i got for dirt plays what i want on ultra with great fps. Nevermind getting 2 new 6x0 or 79x0.
 
crinkleshoes said:
I understand, it's frustrating... my last ATi cards were a pair of 5770s, almost put me off pc gaming for a while with the stutter :(
Click to expand...

I heard the stories of people saying xfire was horrid, nearly put me off, but the 2nd card popped up at 130 quid and i went for it, more than happy, and now ive eliminated the stutter with LTmatt and surveyors help, im back to being happy. Im going to skip this gen, and probably next gen, maybe see what Q3 or Q4 of next year brings gpu wise.
 
Surveyor said:
You're welcome.

It was a quick fix and if you can get even better perfomance with LtMatt's tweaks then all to the good.
Click to expand...

Not tried LTmatts yet but will later tonight. Im so glad it worked, felt really crest fallen when i had it, im 30 and it sounds silly but i was actually really gutted, like a kid with his sweets taken away. The whole rig is complete now, i know the 6950 is older gen gpu, but it works you know? why change it and spend more on something i dont need like a 7970 or 670/680
 
LtMatt said:
I would like to offer some alternative advice based on what ive found with crossfire.

Vsync removed microstutter for me. If i turn vsycn off, i get stutter.

I also tested all the render ahead settings when using crossfire using the PerfOverlayVisible 1 command in game.

The best render ahead setting in crossfire that keeps fps spikes to a minimum is 1. So create a user.cfg file and place it in the bf3 folder. You want this in the file

RenderDevice.TripleBufferingEnable 1
Render.DeviceVSyncEnable 1
GameTime.MaxVariableFps 59
renderdevice.forcerenderaheadlimit 1

The fps limit is 1 below 60, this removes any mouse lag. Try my settings and see what you think. I've tested all the methods, this is the best for my crossfire setup.

Good luck soldier.

EDIT

I just see you said surveyors advice worked. I emplore you to try my settings for even better performance. :)
Click to expand...

I'm amazed how you don't get input lag at 59, even with 58 locked it still feels awful. Feels like using etchasketch. When vsync is off for me everything is instantaneous with my Razer.
 
MichaelM said:
I think the whole microstutter thing is just down to low FPS. When you say 58 that's 29fps per GPU roughly and at that framerate I would count a game to be unplayable.

CF/SLI is mainly about pumping out as many FPS as possible im my opinion so i'd say lower something to make it smoother.
Click to expand...

That logic doesn't sound right, 60 FPS in Crossfire is a lot smoother than 30 FPS on a single card IMO/IME.
 
crinkleshoes said:
The only way to get rid of it is to switch to team green I'm afraid... their multi-gpu implementation is noticeably better for stutter.

It's still not as good as a single card, but it's close now at least :)

ATi still have a way to catch up in this department :(
Click to expand...

I don't know how you can say that when several people from both sides do and do not use this fix for the micro stutter issue.
 
chipachap said:
That logic doesn't sound right, 60 FPS in Crossfire is a lot smoother than 30 FPS on a single card IMO/IME.
Click to expand...

The logic is quite correct, not something I'd come to the conclusion of though until reading this and I notice any kind of nasty stutter.

Basically he's saying I believe, that because each card is rendering roughly 30fps if the game is running at 60fps. This is too low because if a card is only rendering at 30fps it must be heavily taxed and that frame rate would be much more volatile in terms of proportional increases or decreases in frame rate than a game running at 60fps (like you would maybe aim for on a single card by reducing detail\resolution).

Let's say frame rate drops from 30->20fps - 50% drop, if frame rate was at 60->50fps that's still 10fps lower but the proportion of decrease is much lower.

I suppose when a proportionately high number of fps decreases it's hard for the driver to synchronize everything up again and give a smooth flowing gameplay.

...Convoluted explanation, not technical and more than likely wrong but I would guess very true in principal just to get idea across.
 
chipachap said:
I don't know how you can say that when several people from both sides do and do not use this fix for the micro stutter issue.
Click to expand...

It's an accepted fact amongst most that nvidia SLI does a better job at avoiding micro stutter than Xfire. Don't have a specific source but spending years over internet forums has shown that to be the case beyond any doubt at all.

Same as the old "AMD can't make drivers" (or ATI rather). There's no factual evidence but a shed load of circumstantial and examples if you look around to undeniably show that as being true, in so far as nvidia make substantially better drivers for their graphics cards (won't comment on chipset - they are god damn awful).

I'm not saying "AMD can't make drivers", but that's the feeling if you put it into an exaggerated literal short sentence.
 
hernaldo said:
It's an accepted fact amongst most that nvidia SLI does a better job at avoiding micro stutter than Xfire. Don't have a specific source but spending years over internet forums has shown that to be the case beyond any doubt at all.

Same as the old "AMD can't make drivers" (or ATI rather). There's no factual evidence but a shed load of circumstantial and examples if you look around to undeniably show that as being true, in so far as nvidia make substantially better drivers for their graphics cards (won't comment on chipset - they are god damn awful).

I'm not saying "AMD can't make drivers", but that's the feeling if you put it into an exaggerated literal short sentence.
Click to expand...

That's so true, I don't have the vocabulary to be able to make that kind of statement, and the previous post of yours also. All very correct in my opinion.
 
chipachap said:
That logic doesn't sound right, 60 FPS in Crossfire is a lot smoother than 30 FPS on a single card IMO/IME.
Click to expand...

Perhaps that is because CF/SLI 60fps is 1:1 with the 60Hz refresh rate while 30fps single GPU is showing 2:1 so the refresh rate is showing the same frames twice and therefore not as smooth a transition.

So a single GPU running 30fps with 30Hz refresh should seem relatively as smooth.


I'm not saying I know any of this for sure but I'm just trying to make sense of it.
 
shawsy109 said:
I tried yours, and its fine, smooth as silk on caspian border which is were the stutter was, funny it was the only map i get it on. All good now though thnx.
Click to expand...

Yes, my advice worked lol. :D

Take note fellow crossfire bf3 users. GG all. :cool:
 
hernaldo said:
The logic is quite correct, not something I'd come to the conclusion of though until reading this and I notice any kind of nasty stutter.

Basically he's saying I believe, that because each card is rendering roughly 30fps if the game is running at 60fps. This is too low because if a card is only rendering at 30fps it must be heavily taxed and that frame rate would be much more volatile in terms of proportional increases or decreases in frame rate than a game running at 60fps (like you would maybe aim for on a single card by reducing detail\resolution).

Let's say frame rate drops from 30->20fps - 50% drop, if frame rate was at 60->50fps that's still 10fps lower but the proportion of decrease is much lower.

I suppose when a proportionately high number of fps decreases it's hard for the driver to synchronize everything up again and give a smooth flowing gameplay.

...Convoluted explanation, not technical and more than likely wrong but I would guess very true in principal just to get idea across.
Click to expand...

The reasons why 60fps with 2 cards might not feel great is down to a number of issues.

If the CPUs are rendering in scissors/split frame mode then its usually less of an issue as they are both working on part of the same frame. Beyond that its more complicated - if your at 60fps due to the vsync cap then each card is building a frame very quickly and under utilised you will see some input latency from the vsync but it should otherwise be smooth. But in other cases you might be seeing something like: 0ms card one renders a frame, 3m card two renders a frame, then a pause of around 30ms before they both spit out a couple of frames each which might equal 60fps but the massive gaps between frames will make it feel sporadic and a lot lower framerate.
 
You must log in or register to reply here.
Back
Top Bottom