Crossfire 4870/4850

A

Avareyne

Limp Gawd
Joined
Jun 5, 2008
Messages
209
I asked on another forum but haven't gotten a clear answer. The new cards coming out, up to how many can be crossfire'd? It looks like up to 4, can 3 be done?

What Intel motherboards support this? Would there be a large performance increase? I could get 3 4850's for less than a GTX 280 it looks like so am interested in going that route, or maybe 2 4870's and add in 2 more down the line. :confused:
 
I am curious about this also. Currently running a single Nvidia 8800 GTS 640Mb (G80) on a Abit IP35-Pro motherboard. I will be going with an ATI board this time, barring any weird glitches when the NDA lifts. How well does crossfire work on the P35 chipset?

 
Three-GPU setup = low minimum framerates, poor scaling, and microstutter++++!
 
i probably wouldnt crossfire on a P35 cause one of the slots is x4 speed so that might not be enough. P45 w/ the PCIE2.0 x8/x8 config and x38/x48 with x16/x16 pcie 2.0 slots will be the best option

AFAIK, any Intel boards w/ 2 or more PCI-E x16 slots will have CrossFire work. In fact, 3-4 card CrossFireX works too if any Intel boards have that many PCI-E x16 slots
 
a good picture of the 48xx might reveal the answer. if there are two cf pegs then crossfire x maybe a go. drivers...who knows.
 
ZerazaX said:
i probably wouldnt crossfire on a P35 cause one of the slots is x4 speed so that might not be enough. P45 w/ the PCIE2.0 x8/x8 config and x38/x48 with x16/x16 pcie 2.0 slots will be the best option

AFAIK, any Intel boards w/ 2 or more PCI-E x16 slots will have CrossFire work. In fact, 3-4 card CrossFireX works too if any Intel boards have that many PCI-E x16 slots
Click to expand...

This is what I was wondering about, I would only xfire 2 cards tops, but it will probably not be worth it on my board.
 
I don't understand what microstuttering is. People have mentioned this with the GX2's and SLI setups (is it a problem with 280 SLI setups??)

Anyone care to explain?

I've been taking in too much info in the past couple weeks but not this.

If I can't crossfire them successfully then I guess I'll get the 280. I'm so confused and don't know what to buy anymore.
 
Avareyne said:
I don't understand what microstuttering is. People have mentioned this with the GX2's and SLI setups (is it a problem with 280 SLI setups??)

Anyone care to explain?

I've been taking in too much info in the past couple weeks but not this.

If I can't crossfire them successfully then I guess I'll get the 280. I'm so confused and don't know what to buy anymore.
Click to expand...

micro stutter: imagine the game running at a hundred frames per second clip and inexplicably it stalls for a half second then returns to normal. the stall is so brief that fraps might not pick it up if you were recording your gameplay. this is why the actual play testing performed by the editors of this site is so crucial.
 
Does the human eye even notice? I mean I've read we can only see something like 40 fps so if one stalls would it be noticeable to the average joe?

I'm assuming any SLI/crossfire setup does this? So even SLI 280's would be annoying if it's noticeable.
 
it doesn't have to stall from that high of a frame rate, but i used 100fps to illustrate how jarring it would be if you were running down a corridor then made a quick turn to the left to enter a room, but there's a hick-up and suddenly you're a few feet further into the room then you expected. like server latency issues in online shooters. if these stalls occur in succession, you can see how infuriating it can be.
 
wolff said:
I am curious about this also. Currently running a single Nvidia 8800 GTS 640Mb (G80) on a Abit IP35-Pro motherboard. I will be going with an ATI board this time, barring any weird glitches when the NDA lifts. How well does crossfire work on the P35 chipset?

Click to expand...

P35 crossfire will handicapp anything post X1900 series. The reason is in crossfire one card recieves a total of 4 PCI-e 1.1 lanes.
 
rennyf77 said:
micro stutter: imagine the game running at a hundred frames per second clip and inexplicably it stalls for a half second then returns to normal. the stall is so brief that fraps might not pick it up if you were recording your gameplay. this is why the actual play testing performed by the editors of this site is so crucial.
Click to expand...

rennyf77 said:
it doesn't have to stall from that high of a frame rate, but i used 100fps to illustrate how jarring it would be if you were running down a corridor then made a quick turn to the left to enter a room, but there's a hick-up and suddenly you're a few feet further into the room then you expected. like server latency issues in online shooters. if these stalls occur in succession, you can see how infuriating it can be.
Click to expand...


Excellent explanations, and it's also especially noticeable at medium framerates, where dips are common and microstuttering happens a LOT. It also makes for a choppier-looking image even with equal framerates. Also, since frames come at different times and aren't evenly-spaced throughout each second, it doesn't look as fluid.

For instance with 100FPS on a normal 1-card setup you would get 1 frame every 1/100th of a second, whereas with 100FPS on a 3-card setup you would see frames sometimes within 1/1000th of a second of eachother, essentially making a "skip" where they overlap and don't even get displayed, resulting in a far worse experience where many are getting slapped together next to eachother and spaced unevenly, causing jerkiness.

Lost! said:
Yes, touching your GTX 280 revealed that. :D
Click to expand...

No, kid, it's known fact on all SLI or CF solutions due to how they work.
 
rennyf77 said:
it doesn't have to stall from that high of a frame rate, but i used 100fps to illustrate how jarring it would be if you were running down a corridor then made a quick turn to the left to enter a room, but there's a hick-up and suddenly you're a few feet further into the room then you expected. like server latency issues in online shooters. if these stalls occur in succession, you can see how infuriating it can be.
Click to expand...

Ok, that'd be quite problematic in mmo's. Obviously I need not end up running further into a room and killing my entire raid due to my video card. Cue me getting /gkicked and me throwing the GPU out the window. :eek: lol.

Does this happen with the SLI 280's? (yeah I know this is the ATI forum, but I really need a solution soon here). The GX2 can't be an option for me if that sort of thing occurs (which people say it does). I play mostly mmo's and the reason I am building a new rig is because I cannot lag, at all, in high end raiding or else it becomes problematic over time. :mad:
 
resolutions and game settings play a part in how frequently you might encounter stuttering. last night while i was futzing around with my p45 build, i fired up cod4 with two x2's loaded in the system. while i didn't play for very long, forcing 8x adaptive super sampling aa at 1680x1050 revealed the stuttering phenomena when turning around quickly.
 
GoldenTiger said:
It is an inherent problem with how SLI/CF work, so yes, it does :(. However, some users are reporting improvements with the 280/260s.
Click to expand...

the only ways to solve it is to
-Vsync, if your never dip below 60 then you get smooth even 60fps. But if you have a set up that deliver such fps your single cards will probably do just fine in the first place.

-Ditch AFR as a way to achieve CF/SLI. AFR is used currently as it's the easiest way.

-Have the cards sync to each other to even out the time between frames more. From a B3D thread iirc a programme exists for this for NV cards. It basically injects small delays into the cards that runs ahead of the other one.
 
GoldenTiger said:
No, kid, it's known fact on all SLI or CF solutions due to how they work.
Click to expand...

And we've seen the legions of 8800U Tri-Sli users whining then. Nope, didn't see any.

Stop taking the rather lame 8800GT SLI/3870 CF examples. Everytime people report microstuttering with those 2 configs it's almost always incredibly high resolutions with AA, which make the video memory buffer to overflow and causes data swapping over PCI Express which makes the frames more prone to not sync with each other.

Considering how easy the G92 uses video RAM I'm clearly not surprised too.
 
Lost! said:
And we've seen the legions of 8800U Tri-Sli users whining then. Nope, didn't see any.

Stop taking the rather lame 8800GT SLI/3870 CF examples. Everytime people report microstuttering with those 2 configs it's almost always incredibly high resolutions with AA, which make the video memory buffer to overflow and causes data swapping over PCI Express which makes the frames more prone to not sync with each other.

Considering how easy the G92 uses video RAM I'm clearly not surprised too.
Click to expand...

It's not at high resolutions only, do your research. It occurs at ALL resolutions, microstuttering is a fact of life due to how AFR works.
 
Lost! said:
And we've seen the legions of 8800U Tri-Sli users whining then. Nope, didn't see any.

Stop taking the rather lame 8800GT SLI/3870 CF examples. Everytime people report microstuttering with those 2 configs it's almost always incredibly high resolutions with AA, which make the video memory buffer to overflow and causes data swapping over PCI Express which makes the frames more prone to not sync with each other.

Considering how easy the G92 uses video RAM I'm clearly not surprised too.
Click to expand...

I call BS. Source please.
 
Ok, so what's my best solution here? The GTX 280 solo and see if the microstuttering improves with it?

I cannot have any microstuttering.
 
Avareyne said:
Ok, so what's my best solution here? The GTX 280 solo and see if the microstuttering improves with it?

I cannot have any microstuttering.
Click to expand...

you seem to be missing something here. microstuttering only happens when you have more than one card. getting anything "solo" would avoid the problem. though depending on what you're trying, your FPS might be so low it doesn't matter.

GTX 260 IMO. pretty sure that's what i'm getting, anyway.
 
Back on topic a bit :)
Has anyone seen a 4870 Crossfire test?
 
dystopic said:
you seem to be missing something here. microstuttering only happens when you have more than one card. getting anything "solo" would avoid the problem. though depending on what you're trying, your FPS might be so low it doesn't matter.

GTX 260 IMO. pretty sure that's what i'm getting, anyway.
Click to expand...

But to get the performance of a 280, I'd definitely need 2x ATI cards it looks like (from the tests already out) but with microstutters.

The 260 is not much of an improvement over the 4850 (at twice the price, not worth it) and is not on my consideration list at all. It's either 280 or 4870, I don't care about brand! Whichever solution gives the best performance on a single card I'll take, cause I can't have this thing called microstuttering. I almost bought the GX2 but did not know/understand what microstuttering was.

This GPU thing is putting a damper on buying a mobo, so I gotta figure it out.
 
How do the microstutters compare between ATI and NVidia dual cards?
 
Nenu said:
How do the microstutters compare between ATI and NVidia dual cards?
Click to expand...

I've been running 2 HD3870's for about 2 months now I'd say, and I play:

HL2
HL2:E1
CS:S
UT2004
UT3
MOH:Airborne
COD:4

I have yet to experience ANY problems, just smooth gameplay and excellent visuals.
 
RADEoN said:
you're not gonna see a legit one until a day before the release of the card.
Click to expand...
It has been reported that the 4870 is likely to be launched at the same time as the 4850 so hopefully there will be some techies that can get their hands on some early samples.
 
I play Age of Conan mostly right now. Anyone with SLI/Xfire play that and can tell me about the microstuttering???
 
Nenu said:
I wondered which is more noticeable?
Click to expand...

both pretty much the same. 8800gtx sli and 3870x2 cf will stutter at 16x10 with tons of aa; 8x or higher. at 19x12 the difference between 4x aa and 4x super sampled aa is pretty drastic. i like using cod4 because it can run fps in the hundreds at 1900 res. by 2500 res when your maybe running 2x aa, stuttering can cloak in between predictable periods of slow down. the game matters as well. gow stutters all the time, perhaps because of the tight 62 frame ceiling.
 
rennyf77 said:
it doesn't have to stall from that high of a frame rate, but i used 100fps to illustrate how jarring it would be if you were running down a corridor then made a quick turn to the left to enter a room, but there's a hick-up and suddenly you're a few feet further into the room then you expected. like server latency issues in online shooters. if these stalls occur in succession, you can see how infuriating it can be.
Click to expand...

Have you considered time travel? Look at your clock when this happens next time.
 
Are (2) 8x lanes of PCIex 1.0 going to bottleneck with these in crossfire? Speaking of what we get on our 975x intel boards of course.
 
Hypernova said:
the only ways to solve it is to
-Vsync, if your never dip below 60 then you get smooth even 60fps. But if you have a set up that deliver such fps your single cards will probably do just fine in the first place.
Click to expand...

Correct me if I am wrong....

With Vsync off and you never drop below 30 fps, you should be able to turn vsync on not experience microstuttering, since the game will just lock you into 30 fps.

I have no problem with 30 fps in any game I play as long as its constant. The problem comes in with Vsync if you drop below 30 fps.
 
Ok well since I would have to xfire to get near the performance of a single 280 (according to the only review, so I'm not setting this in stone) then I guess I'll be going green because I don't want microstuttering. Still waiting to hear about any xfire/SLI experiences in AoC.
 
Oh and will [H] at the very least record if there is microstuttering or not? Simply because I have never seen a review that really says.
 
You must log in or register to reply here.
Back
Top