Crossfiring~ apparently orientation of cards matter

[OnionKnight]

Okay so I got my second HD 5770 which was a gigabyte brand (ver. 2) and crossfired it with my XFX HD 5770 (ver. 1). Everything went good.

 

 

Test:

 

Part 1:

I only use 1 monitor and had my XFX as my main display card (DVI plugs in there).

Did 3dmark vantage and got 14840

 

Part 2:

The I used my Gigabyte card as my main display card . Did 3dmark Vantage and got 14250.

 

Retest:

 

Lol there shouldn't be any difference in vantage scores so I did 2 more tests and got similar results.

 

Just to make sure there wasn't any difference in the performance of each card, I did 3dmark on each card separately.

 

XFX HD 5770 (ver. 1) ~ 9260

Gigabyte HD 5770 (ver. 2) ~ 9300

 

So ya they are almost identical. Lol I never seen this issue before (probably because no one ever mixes and matches models in any of the internet reviews =P).

 

Conclusion:

After reading more on how crossfire works, I assume it is because one card is better at transferring data than the other in a crossfire setup. Lol so I guess the lesson is that not all versions of the same card model are made equal (I believe my Gigabyte ver. 2 HD 5770 card might have faster transfer rates across the crossfire bridge). Depending on which version of the card you use as your primary display may actually effect its performance.

 

How crossfire works?

[kdoto]

ah yes the headaches of crossfire xD

[OnionKnight]

ah yes the headaches of crossfire xD

 

Ya I didn't know that shit about that the revisions of the card might affect performance =P (like crossfiring version 1 and 2 of HD 5770). But I was told that it wouldn't affect performance as long as I used the older version card as my primary display.

[NoGooD]

Onion,

 

I have not looked into this, but would the differences in X-fer rates cause any problems while in crossfire? It seems like eventually the faster card would have to wait for the slower one to catch up or throttle down to speed match..which as you have pointed out neither card is doing.

 

I'd be interested to see the results after gaming for several hours if any issues arise.

[OnionKnight]

Onion,

 

I have not looked into this, but would the differences in X-fer rates cause any problems while in crossfire? It seems like eventually the faster card would have to wait for the slower one to catch up or throttle down to speed match..which as you have pointed out neither card is doing.

 

I'd be interested to see the results after gaming for several hours if any issues arise.

 

I myself am not really sure how it works myself =/. I am new to crossfiring. But from my test I see that having the newer card as the secondary increases performance.

 

My Guess:

 

The secondary (newer card) is transferring the most data to the primary so it could be coordinated and displayed on the monitor, whereas I don't think there is much data flow from the primary --> secondary (so having the old card as primary means it will have less of a role in sending data along the cf wire). So this is just a guess because i seriously can't find shit on this on the internet lol (except for that diagram I posted)....

 

But ya I'll tell you how it goes. So far everything has been fine with crysis and just cause 2. (probably wait for a week at least)

[kdoto]

what s your average fps with the 5770s. i know i shouldnt but im looking for dual cards better than my 4890s lol xD

[OnionKnight]

what s your average fps with the 5770s. i know i shouldnt but im looking for dual cards better than my 4890s lol xD

 

I wouldn't bother k.o =P. HD 4890 crossfire are more powerful than my cards (they are probably just as powerful as a HD5830). Average fps depends on the game. Crysis goes from anywhere from 30-50 fps (lol its varies quite a bit depending on the scene), but i run it on maximum on all settings and 4x AA (i tried 8x AA but it impacted the performance lol)

[Number1Dad]

That's an interesting bit to remember. I wouldn't imagine it would affect your performance all that much though, not even enough for a whole frame per second difference in most games.

[OnionKnight]

That's an interesting bit to remember. I wouldn't imagine it would affect your performance all that much though, not even enough for a whole frame per second difference in most games.

 

Ya not much. 2-4 fps in crysis. But I am still not sure of the problems itself =P (maybe I just unlucky and got a crappy video card though lol? I tested both cards though and they perform the same, so maybe the crossfire bridge could be faulty? I am still guessing XD). Well whatever I guess this is why people don't mix brands and versions lol. As long as I use the older card as primary I don't have any problems at the moment

[kdoto]

yeah i average around 60 fps in crysis. all maxed. only problem is the game will randomly shut off after a while. if i turn AA off it wont do that. but bleh.

[detoren]

I have 2 4870 x2 from 1 brand and they give a diffrent score and on quad it wont run propperly cos 700 watt power supply 1 is from my son when i oc to 4 gig 1 does run on 2400 marks other almost 2500 with triple gpu ( my wife has normal 4870) it runs but hardly give a boost cos of the diffrent mem on card . btw vantage is giving a rate for your computer in total disc drives mem cpu and gpu 3d mark 06 is based on graffics and better to compare gaming quality and free to use .

[OnionKnight]

I have 2 4870 x2 from 1 brand and they give a diffrent score and on quad it wont run propperly cos 700 watt power supply 1 is from my son when i oc to 4 gig 1 does run on 2400 marks other almost 2500 with triple gpu ( my wife has normal 4870) it runs but hardly give a boost cos of the diffrent mem on card . btw vantage is giving a rate for your computer in total disc drives mem cpu and gpu 3d mark 06 is based on graffics and better to compare gaming quality and free to use .

 

LOL neither mark gives you a better rate for your computer. It depends on your video card models. For a direct x10/11 card you would want to use 3d mark vantage. 3d mark 06 uses direct x 9.0c and a much older shader model which will give you an unaccurate benchmark (unless you run all program under direct x 9.0c then I guess its accurate).

 

Also Vantage gives you two separate scores for your GPU and your CPU. They also give you an overall score (which is a combined average between the two). I am only using the GPU score

[kdoto]

well just ordered a 5970. let's see how this will do x]

[detoren]

5970 is not available here only for 150 euro above advise price so i wait a month or 2 , they ask 650 euro before its home lol advice price for shops was around 500 and its a bit overkill for et or cod ^^. running 300 fps is not realy good for your connection lol .

[OnionKnight]

well just ordered a 5970. let's see how this will do x]

 

Your video card is worth almost as much as my entire computer. Lets rob K.O.