AMD Crossfire a scam - Almost no benefit over single card

Status
Not open for further replies.
oldmanbal

oldmanbal

2[H]4U
Joined
Aug 27, 2010
Messages
2,613
*I want to start off by pointing out that i have 2x7970s and 2x7950s in my rigs at the moment, and have no bias against Nvidia or AMD. I simply buy what I find to be the best performance/value for what my goals are each year when i build a new rig.*

PCPERSPECTIVE has finally given a repeatable and accurate representation of what so many people have been arguing over for quite some time. In most cases, crossfire delivers an identical gameplay experience as single card solutions. Nvidia has a substantial advantage in this article but still has many areas of improvement so it's not as one sided as people claim.

http://www.pcper.com/reviews/Graphi...ils-Capture-based-Graphics-Performance-Testin

Despite the damning information against crossfire in many games that simply renders useless and nonexistant frames that fraps picks up in benchmarks, I was specifically interested in their testing with vsync or frame limit implementation. Unfortunately they didn't use a frame limiter like afterburner's rivatuner in which you can manually set the frame cap to somewhere around your lowest fps which in my scenario fixes almost all of the issues as far as gameplay experience go, which is what Kyle trumps as the bottom line for end user experience.

I wrote Ryan Shrout an email last night playing off the use of vsync as a solution that many people recommend (and triple buffering for that matter) because of the substantial lag it adds to the game which is worse than all of the preceding crossfire/sli issues. He agreed that vsync can present additional issues that make gameplay worse. He also remaked that unfortunately they had already finished the article and didn't include any testing on frame limiting when set manually so I hope in the future they can provide data on it in a meaningful manner. I give Ryan as much thanks as possible and hope he continues to deliver groundbreaking gaming journalism.

As long winded and meandering the anandtech article was yesterday, seemingly purposeful in trying to give AMD some positive PR regarding their past transgressions, The new information provided by Ryan at PCPER has left me wondering if my second gpu has any value any more, and if AMD has been scamming the consumer purposefully for years? I know they are now admitting they have some work to do, but it's easy to admit you robbed a bank when you get pulled over with bags of cash and paint on your hands. So far AMD has only touched on latency issues and frame delivery, as far as I know they have yet to tackle the dropped frame and worthless frame argument.

Shame on you AMD, benching on fps avg is all but moot now on anything other than a single gpu.

I would love for AMD to start coming up with a solution to the disaster that is crossfire at the moment, but i'm not holding my breath as Nvidia has a newfound respect in my book.

What do other multi-gpu owners think? What settings do you use?

This is my preference:
no vsync
Frame limit to 60 or less depending on the game and min/avg fps performance.

Again, I'm really looking foward to a site like pcper running a bench using the frame limiting tools available as right now that seems to be the only viable option in mitigating the crossfire folly.
 
So...what you're saying is you see no difference in performance between one card and two in your own machines?
 
If you don't have time to read the whole article, take a look at the 'observed' FPS graphs on each of the games.
 
Yes, the question is are your games faster with 2 gpu than one gpu? What are you observing?
 
I've had nothing but excellent results with my crossfire set up. It's my very first CF setup. I had been scared off for years by people like YOU who have nothing but complaints.

I'm happy to say that you're wrong.
 
I'm going to do some research and testing this weekend. I'm so fucking tired of Multi GPUs anyway AMD or Nvidia both suck. Both an expensive, power sucking, house heating, pain in the ass.

I would not fall in love with Nvidia though don't worry something will come up. SLI 680s gets similar or worse performance to Crossfire 7970s so what does that mean?
 
What games have you tried benchmarking on? My experience is limited, but I wrote a post back in the day based on my experience with DCS: A10

http://forums.eagle.ru/showthread.php?p=1427563#post1427563

My performance was worse with crossfire than it was with a single card. Now this has been attributed by the game developer to AMD not optimizing crossfire for DCS, and on the other side AMD has said it's up to the game designer to optimize the graphics engine for crossfire. Whichever way it goes, the problem went unfixed.

Long story short, I've had a poor experience with crossfire, and will be sticking to single card ops in the future.
 
oldmanbal said:
'does not meet expectations?'
Click to expand...

Ok so you're not seeing a frame rate increase by equipping multiple cards. That's interesting to say the least. I'm not running Crossfire yet but was contemplating it as I'm going to need a faster AMD card for my 3D monitor. I'm not buying a new 27" 120Hz monitor to switch to switch to Nvidia. :)

I wonder how many others are not getting a higher frame rate from their cards? Anyone else experiencing this where you get the same fps with one card or 2?
 
I have done crossfire and sli and Im like meh! not worth the money but it might be If I had:) 3 30's to play with.
 
depends on the games, depends on the cards, they all have different scaling, Nvidia is no different here, some scale very well, some scale very poorly.

There is a reason why I use single card and always have, neither AMD nor Nvidia offer great scaling all the time like it should be, seeing as both SLI and Crossfire have been out for years and years.

Nvidia offers more games that scale however AMD has more scaling % with games that do :)

Yeh and I have a friend who has more then enough $ to get the best gear available who simply put, gave up on SLI as to him the stuttering, improper scaling, extra heat/power became more trouble then it was worth, so he says "now I have to turn things down a bit, but I still enjoy the same games"
 
I would say the biggest issue for me, is going from 1920x1080 to 2560x resolution. At the higher resolution, crossfire seems to really just struggle to keep a consistent experience. The frame rate looks like it should be playable, but often times is choppy, laggy, and feels off. The number of dropped and runt frames identified in Ryan's testing finally puts some science into why this is happening. There is almost ALWAYS a substantial increase in this when moving up to higher resolutions, which is why we buy multi-gpu setups for in the first place.

This is the current advertising on AMD's crossfire page on their website:
"Tired of dropping frames instead of opponents? Find a CrossFire™-certified graphics configuration that’s right for you."

Hilarious.
 
Pcper said:
You may notice that there is a lot of “my” and “our” in this story while also seeing similar results from other websites being released today. While we have done more than a year’s worth of the testing and development on our own tools to help expedite a lot of this time consuming testing, some of the code base and applications were developed with NVIDIA and thus were distributed to other editors recently.
Click to expand...

Raises Eyebrow?

is this an Nvidia advertisement?
 
I just went back sli after running my card single for months now coming off of a asus 7970 crossfire setup . . and all I got say is I see then end of dual gpu configurations unless both developers can get over the hurdle of making the gpu frame evenly on all occasions ,My gaming on my single gpu 680 is much smoother than my dual card 680 gpu setup , even if Im framing at a lower fps . . but Id love to see more user info on this
 
This is why the User Experience's Reviews that [H] does is so much better than strictly benchmark and numbers and pretty much discounts the conclusion pcper makes.

if you look at the two reviews from the [H] Both the 7970 and the Crossfire review.. if what PCPer is saying is true, the highest playable settings for the crossfire vs single card would be the same

[H] does just look at the FPS but the gaming experience as a whole and their conclusion is that Crossfire does indeed increase the user experience, and not just by a little.
 
So Dual Card issues.. that have been going on since the VOODOO2 days are all of a Sudden NEW issues? some of you guys must have just started gaming on the PC with WOW or something
 
H-street said:
Raises Eyebrow?

is this an Nvidia advertisement?
Click to expand...

ryan at anandtech has confirmed the new FCAT softwear/code is unbiased in analyzing data, specifically complimenting it for better demonstrating the end product that the user sees and experiences. IMO AMD would have released something similar if it wasn't so damning.

Nvidia released it to the public and said have at it, they knew what reviewers would find.
 
*I want to start off by pointing out that i have 2x7970s and 2x7950s in my rigs at the moment, and have no bias against Nvidia or AMD. I simply buy what I find to be the best performance/value for what my goals are each year when i build a new rig.*
Click to expand...

You owning AMD cards kinda sounds like the old, "I have black friends..." thing. And what is it, man? A scam or a folly? One's intentional and the other isn't. Also, I have never expected consistent performance increases... not in 1998, not today. If all games were on the same game engine, sure... but there are so many different engines and configurations, blah blah blah...

I've always seen a noticeable increase in performance when adding in a second card, AMD or Nvidia. Sometimes even if it's only 10 FPS, that's just enough to make the game more playable. 2 cards is a luxury, and lately not much worth it in either camp.
 
I remember when running Nvidia 9600GTs in SLi was the hot thing to do..and there were PLENTY of games that didnt run right in Sli or didnt have a profile at that time.... but i knew the risks going in... there are too many variables when running 2 cards to not have issues....
 
H-street said:
This is why the User Experience's Reviews that [H] does is so much better than strictly benchmark and numbers and pretty much discounts the conclusion pcper makes.

if you look at the two reviews from the [H] Both the 7970 and the Crossfire review.. if what PCPer is saying is true, the highest playable settings for the crossfire vs single card would be the same

[H] does just look at the FPS but the gaming experience as a whole and their conclusion is that Crossfire does indeed increase the user experience, and not just by a little.
Click to expand...

Exactly. Brent and Kyle picked up on the qualitative evidence of AMD's deficiency here long before anyone else.
 
TheGamerZ said:
I've had nothing but excellent results with my crossfire set up. It's my very first CF setup. I had been scared off for years by people like YOU who have nothing but complaints.

I'm happy to say that you're wrong.
Click to expand...

i love people like you. you cant even look at the data and say "hmmm, there might be something there".

H-street said:
Raises Eyebrow?

is this an Nvidia advertisement?
Click to expand...

this is the thing that has a little nervous, but the data looks solid.
"amd has higher fps, well lets benchmark this way. oh look, now we're faster"

H-street said:
This is why the User Experience's Reviews that [H] does is so much better than strictly benchmark and numbers and pretty much discounts the conclusion pcper makes.

if you look at the two reviews from the [H] Both the 7970 and the Crossfire review.. if what PCPer is saying is true, the highest playable settings for the crossfire vs single card would be the same

[H] does just look at the FPS but the gaming experience as a whole and their conclusion is that Crossfire does indeed increase the user experience, and not just by a little.
Click to expand...

you pick a multiplayer bf3 review from jan 2013 and compare it to a sp review from dec 2011.

from the [H] fc3 review
http://www.hardocp.com/article/2012/12/17/far_cry_3_video_card_performance_iq_review/4#.UVMcYhzFX_o

"We have a lot to talk about in regard to multi-GPU performance and gameplay experience in Far Cry 3. The first thing to keep in mind is that raw faster framerates doesn't always mean a better gameplay experience. There are other factors involved that framerate cannot measure. We experience these other variables as we play the game, and subjectively explain the experience to you.

In the table and graph here, you will find that the AMD Radeon HD 7970 GHz Edition CrossFire setup had the highest framerates, beating GTX 680 SLI in raw framerate. The minimum FPS is higher, 42 vs. 37, the maximum is higher 62 vs. 53, and the average is 14.5% higher at 51.2 versus 44.7 FPS. This is at the same playable settings of 2560x1600 with 4X MSAA, Enhanced Alpha to Coverage, HDAO and Ultra in-game settings.

While the framerates are faster under CrossFire, the gameplay experience is noticeably different between GTX 680 SLI and HD 7970 GHz Edition CF in Far Cry 3. We experienced a horrible amount of stuttering under CrossFire. It was as if the game was pausing every few frames, making the game choppy, and the performance inconsistent. The gameplay experience was close to awful using CrossFire."
 
JustinCorrigible said:
I've always seen a noticeable increase in performance when adding in a second card, AMD or Nvidia. Sometimes even if it's only 10 FPS, that's just enough to make the game more playable. 2 cards is a luxury, and lately not much worth it in either camp.
Click to expand...

I know many people, like everyone does obviously, that think that if you don't almost or at least double your performance with SLI or Crossfire there is no value in either solution. I'm starting to agree. to bad for all these guys with Triple and Quad setups. gives new meaning to "money to burn"
 
Last edited:
That title really distracts from what is a really interesting article. I kept having "hitching" in FC3 with my 690 but the framerate was not terrible. It just is something sort of a barely perceptible glitch and really stands out when you spin around quickly like you tend to do a lot in an FPS.

Going from the 690 to the titan was a huge improvement in FC3 and Tomb Raider.
 
harmattan said:
Exactly. Brent and Kyle picked up on the qualitative evidence of AMD's deficiency here long before anyone else.
Click to expand...

What are you on about?

Read H-street's post again. He clearly says that [H]'s review points out that there IS a qualitative difference between 7970 single card and 7970 CF, DESPITE what the OP of this thread, PCPer's article, and that article's concerns may suggest.

Also, like so many initial responders to that PCPer piece, I felt that the article was in dire need of an editor :eek:
 
tricky0502 said:
i love people like you. you cant even look at the data and say "hmmm, there might be something there".



this is the thing that has a little nervous, but the data looks solid.
"amd has higher fps, well lets benchmark this way. oh look, now we're faster"



you pick a multiplayer bf3 review from jan 2013 and compare it to a sp review from dec 2011.

from the [H] fc3 review
http://www.hardocp.com/article/2012/12/17/far_cry_3_video_card_performance_iq_review/4#.UVMcYhzFX_o

"We have a lot to talk about in regard to multi-GPU performance and gameplay experience in Far Cry 3. The first thing to keep in mind is that raw faster framerates doesn't always mean a better gameplay experience. There are other factors involved that framerate cannot measure. We experience these other variables as we play the game, and subjectively explain the experience to you.

In the table and graph here, you will find that the AMD Radeon HD 7970 GHz Edition CrossFire setup had the highest framerates, beating GTX 680 SLI in raw framerate. The minimum FPS is higher, 42 vs. 37, the maximum is higher 62 vs. 53, and the average is 14.5% higher at 51.2 versus 44.7 FPS. This is at the same playable settings of 2560x1600 with 4X MSAA, Enhanced Alpha to Coverage, HDAO and Ultra in-game settings.

While the framerates are faster under CrossFire, the gameplay experience is noticeably different between GTX 680 SLI and HD 7970 GHz Edition CF in Far Cry 3. We experienced a horrible amount of stuttering under CrossFire. It was as if the game was pausing every few frames, making the game choppy, and the performance inconsistent. The gameplay experience was close to awful using CrossFire."
Click to expand...

And this is why evaluating the real world gameplay experience of games will always trump whatever data is collected. Frame rate numbers and frame time numbers are surely good indicators as to what is going on, but the fact of the matter is that we play games for the experience and if that experience is sullied by the hardware or game settings being used, then the gaming experience is worthless to the person spending their hard earned money on games and hardware.
 
H-street said:
if you look at the two reviews from the [H] Both the 7970 and the Crossfire review.. if what PCPer is saying is true, the highest playable settings for the crossfire vs single card would be the same

[H] does just look at the FPS but the gaming experience as a whole and their conclusion is that Crossfire does indeed increase the user experience, and not just by a little.
Click to expand...

Benchmarks, frametimes, and framerates have been lies for GPU evaluation for a while now if looked at alone. This data NEEDS interpretation and analysis as to how it relates to the gaming experience for it to be valuable.
 
How much of this is due to Crossfire, and how much of this is due to GCN?

The 2x 4870x2 setup that I ran in my main computer for years, and is still in use in my secondary computer runs very good for what it is. A single 4870 would be pretty slow, but 4 4870s delivers performance approaching that of a single GTX680 in most situations. That is a huge difference, and one that is immediately noticeable. Games that would have run like a slideshow get 45+ fps instead and that is a night and day difference.

It really seems to me like most of these AMD issues began with GCN. SLI being smoother than crossfire isn't really new, but no one was previously claiming that Crossfire was no better than a single card.
 
this thread started out as pure flame bait....OP is saying one AMD card vs 2 AMD cards in X-Fire = same exact performance....
 
GotNoRice said:
It really seems to me like most of these AMD issues began with GCN. SLI being smoother than crossfire isn't really new, but no one was previously claiming that Crossfire was no better than a single card.
Click to expand...

Again this comes down to interpretation of the data and what is being compared in any review. If CrossFire was no better than a comparable single GPU card, then real world gaming testing of highest attainable settings and resolutions would expose this easily. Again, it is why we were first in the industry to start this tremendously resource intensive testing years ago.

We have been talking to NVIDIA about frametime testing and collection for a long time now and there is good information back from inside the NVIDIA organization that HardOCP GPU reviews was the catalyst for this coming about. We had the opportunity to help develop the program tools with NVIDIA but chose not to. PCPer has put an incredible about of time and money into this program that we were simply not comfortable with spending. PCPer has done a great deal of needed work on this with NVIDIA, which is commendable, but I am not sure data collection on this front will prove to be the end all be all in GPU reviews. It all still comes down to evaluating the end user gaming experience and how well the hardware allows you to achieve you wants and needs on this front. Frame time data collection will never be something that any users can use at home easily so it will never be more than a review data point. Focus on the user experience will still have the most impact on video card sales making sure the end user gets what he wants and needs.
 
GotNoRice said:
How much of this is due to Crossfire, and how much of this is due to GCN?
Click to expand...

They are going to be publishing articles for a variety of setups in the coming weeks @ pcper. Anandtech is currently in the process of using the same type of setup to create a new look for their readers with the release of Nvidia's FCAT.
 
I don't have any issues with crossfire performance on my 6850+6870 setup at either 1680x1050 or 1680x1050x3 (except when I run out of video ram with too high of AA on...)

Then again I don't play any of the games that were tested in that article and I have last gen cards....
 
the_real_7 said:
I just went back sli after running my card single for months now coming off of a asus 7970 crossfire setup . . and all I got say is I see then end of dual gpu configurations unless both developers can get over the hurdle of making the gpu frame evenly on all occasions ,My gaming on my single gpu 680 is much smoother than my dual card 680 gpu setup , even if Im framing at a lower fps . . but Id love to see more user info on this
Click to expand...

single card is the better option any time-
either it is amd or nvidia.

One reason I OC a single 7970 for my 5040x1050 set up and lower settings.
much smoother experience.
 
GunzofNavarone said:
some of you guys must have just started gaming on the PC with WOW or something
Click to expand...

Comments like these aren't necessary.

Enthusiasts new to the dual card scene might not have any idea about some of the problems associated with them muchless xfire vs sli. When I first got my second 570 I noticed a huge improvement in some games. For other games the micro stuttering was just to much.

JustinCorrigible said:
You owning AMD cards kinda sounds like the old, "I have black friends..." thing.
Click to expand...

Are you implying there is something wrong with stereotyping xfire performance? I mean, it's ignorant to do with people because they are all different based on personalities... But video cards of the same model and using the same driver are basically identical...
 
Last edited:
GunzofNavarone said:
this thread started out as pure flame bait....OP is saying one AMD card vs 2 AMD cards in X-Fire = same exact performance....
Click to expand...

But he also said this..

"While some people might have assumed that this new testing methodology would paint a prettier picture of NVIDIA’s current GPU lineup across the board (due to its involvement in some tools), with single card configurations nothing much is changing in how we view these comparisons. The Radeon HD 7970 GHz Edition and its 3GB frame buffer is still a faster graphics card than a stock GeForce GTX 680 2GB GPU. In my testing there was only a couple of instances in which the experience on the GTX 680 was faster or smoother than the HD 7970 at 1920x1080, 2560x1440 or even 5760x1080."
 
Lets get this discussion back to ground level and focus in on one metric in specific that I personally found to be the most powerful out of all the testing data Ryan came up with.

Take a look at the charts that demonstrate the observable fps in the article. These are taking the 'bogus' frames, ie dropped frames that do not even render or show up in the tests, but bloat the fps score by being counted in tools like fraps, and the 'runt' frames, that are being counted, but are essentially nonexistent due to the fact that they are incomplete frames and do not deliver anything productive to the end user experience. If you wanted to slander AMD, you could say that these are ways of cheating benchmarks and bloating your score by kicking out ghost frames that essentially don't exist and aren't being rendered to the user. Now they may actually be somehow rendered in the gpu's pipeline, but essentially nothing is being delivered to the end user in terms of visual presentation which equates them to being a 'lie' when measured by a collection service such as fraps.

While i almost always see great scaling from my crossfire setups, what has been uncovered was the perceived truth that what was going on was much more complicated than just providing the same experience at higher framerates. We are seeing a situation where crossfire is disastrously muddled, delivering frames that are not screened, and frames that when screened are a few pixels high that don't add to the gameplay experience thus hampering the quality of the video feed. Crysis 3 observable framerates demonstrate this as with crossfire enabled, you are actually getting much worse performance comparatively to a single gpu. This is the opposite of what AMD advertises and claims almost completely. Not to mention we are not even adding in the latency issues and consistency of each frames delivery which AMD is currently working on ways to improve (kudos to them).

I would be quite happy to see this review metric in future video card and game reviews because it demonstrates the most honest representation of (latency aside) true performance being displayed to the user.

I would like to change topics later in this thread to cover lag/stuttering produced by vsync, triple buffering, and AA as well because they also have a substantial impact on the playability of certain titles since some games just don't run well with any of these settings turned on with AMD hardware, and crossfire compounds the issue as well. What are your thoughts about the 'observed' fps metric that Ryan Shrout is using?
 
Meh, I gave up on multi-gpu's a long time ago. When I went SLI, I was expecting a game to go from medium settings to ultra or high settings at least. But instead it went from medium to medium high. From that day forth, I decided money is better spent by selling the one you have and upgrading to a new card rather than trying to add another card.

I've got to locate this article but are they testing with the flagships? Are they testing them on PCIe 3.0 boards? Are the CPU supporting PCIe 3.0?

I do believe that multiple flagship GPU's are saturating the PCIe 2.0 bandwidth finally aren't they?
 
Status
Not open for further replies.
Back
Top