Crossfire
So you heard that the X16xx series can plug and play into CrossFire mode? You say you want to see what this buys you, besides more on your credit card? I am here to please. To that end, HIS was nice enough to send me a 2nd HIS X1600Pro IceQ Turbo.
Adding the 2nd X1600Pro was easy enough but it felt strange however, not to have to connect the two cards together; none the less there they both were.
Once I was back up and running, I went into CCC and looked for my CrossFire tab, it was no where to be found. I checked to make sure I had in fact downloaded / installed Cat 6.4, check. I proceeded to un-install the drivers, and re-download / install them just to make sure. A reboot later, the CrossFire tab was there (I did note that the entire CCC number was in fact the same as previous). I selected the tab and hit apply, the screen went blank for a second and then came back. Nothing truly noticeable in basic windows, so lets re-run those tests I had previously performed and see what CrossFire does for my performance.
A few notes about CrossFire while we are here, CrossFire has 3 modes of rendering, Tiling (preferred method and default for DX in X18xx and X19xx series only), Scissors (OpenGL / DX (X16xx series) default) and Alternate Frame Rendering. In order for a game to use one of these methods, it must be profiled by ATI (although a simple google can get you around this limitation). If a game is not profiled, it moves to SuperAA mode, which in effect uses the 2nd card to give you no performance hit when you select 2x, 4x or even 8x AA. In essence you have one card driving your frame rate and the other making everything pretty.
With that out of the way, lets see how CrossFire faired when compared to a single X1600Pro IceQ Turbo using the same benchmarks as previous
Half Life 2
It's interesting that HL2 is my first test, as it gained the least by having CF enabled. My only thought here is that maybe HL2 is not profiled for the X16xx series cards? With that thought in mind, I went ahead and renamed the hl2.exe to FEAR.exe, and re-ran the test again.
|
Frames Per Second
|
Min
|
Max
|
Avg
|
|
HIS X1600Pro IceQ Turbo Crossfire
|
39
|
118
|
60.89
|
|
HIS X1600Pro IceQ Turbo Single
|
34
|
93
|
54.37
|
Well what do you know, thats more like what I would have expected. Of course, the tell tale sign is that when I increased to 4xAA and 8xAF, there was almost no performance hit.
Call of Duty 2
|
Frames Per Second
|
Min
|
Max
|
Avg
|
|
HIS X1600Pro IceQ Turbo Crossfire
|
14
|
63
|
44.78
|
|
HIS X1600Pro IceQ Turbo Single
|
13
|
24
|
19.65
|
Looking at CoD2, we see more then a 100% improvement over a single X1600Pro. CF mode makes this game not only playable, it makes it down right enjoyable at 1280x768, Moving to 4xAA and 8xAF we still see a greater then 50% improvement, however still not exactly playable.
Quake 4
|
Frames Per Second
|
Min
|
Max
|
Avg
|
|
HIS X1600Pro IceQ Turbo Crossfire
|
48
|
127
|
84.09
|
|
HIS X1600Pro IceQ Turbo Single
|
24
|
78
|
50.41
|
Next is Quake 4, which sees a nice benefit as well from using CF. Even when applying 4xAA and 8xAF we see the CF solution barely take a hit, and continue to provide ample frame rates for enjoyable play.
NEXT
