CrossfireX – 57xx Series

thumbCrossfireX – 5750 & 5770

Not everyone can get a pair of matched cards, especially if you bought them seperate times. We quickly put a 5750 and a 5770 in CrossfireX just for the hell of it, and to give you some idea of what to expect you should you need to get a similar set going.

Manufacturer:
Price:



Recently, I’ve reviewed two different cards from the 57xx line up; a reference 5770 1GB and the HIS HD 5750 IceQ+ 1GB (both cards supplied by HIS). This has given me the opportunity to test out something I’ve tried before on previous generations but not as yet seen first-hand with the new 5xxx series. CrossfireX. In this case, I’m going to be showing you some results from testing a 5750 and 5770 in CrossfireX formation.

I’m not going to go overboard with the tests as this is supposed to be a quick article, but I did want to show a little of what this mismatched set-up can do. It’s entirely feasible that you may find yourself with either a single 5750 or 5770, and if you decide to go Crossfire you may not be able to get a matching card at the time, or perhaps a non-matching card might be on sale/a bargain/2nd hand. Hopefully, this article can give you an idea as to what to expect.

 
card_5750 card_5770




The two cards we are going to be using are both supplied by HIS. The first is the HIS HD 5750 IceQ+ 1GB. This card runs at the standard 5750 clocks of 700MHz Core and 1100MHz (4600MHz) Memory. The second card is the HIS 5770 1GB, a reference model also running at standard clocks of 850MHz Core and 1200MHz (4800MHz) Memory.

 

crossfire crossfireUV2 crossfireUV1


Set up was utterly painless. You simply plug the cards into their respective PCIe slots, attach a CrossfireX Bridge to both cards, install the drivers and reboot. Upon reboot, the system detected the two cards, flashed the screen a couple of times and enabled CrossfireX before loading the desktop.

 

cccx1

Testing

Test Setup: Intel Core 2 Duo 6420 @ 3.20GHz, 4GB of OCZ PC2-6400 Ram @ 960MHz, Asus Blitz Formula, Maxtor Diamondmax 10 7200 250GB HDD, Asetek Waterchill Watercooling, Hyper Type M 730w PSU. All latest drivers as of December 2009 (Catalyst 9.12) and the OS is Windows 7 64bit.

For comparison, we are using a reference Radeon HD 5770 card, the HIS HD 5750 IceQ+ 1GB and the HIS HD 4890 iCooler X4 1GB.

Software

Left 4 Dead 2 – Recording a custom demo on the Dead Center – Hotel level (inside in the inferno), we used FRAPS to record frame rates as we played back the demo on all cards at same settings.

Batman Arkham Asylum – We used a combination of the in game benchmark and FRAPS to gather our numbers for this game. All cards were set to the highest possible settings for that card.

Assassin’s Creed – We headed for the nearest tower from the bureau roof in Acre and repeatedly climbed to the top. With 2 leaps of faith and a good look of the city from on high, we once again used FRAPS to record our frame rates. Settings for each card were set to highest possible for that card.

Crysis Warhead – We used the Framebuffer benchmark tool to run through the Ambush demo and recorded the results with FRAPS. Settings for each card were set to highest possible for that card.

Colin McRae DiRT2 – DiRT2 has some very good looking visuals and provides us with our first DirectX 11 test. We used FRAPS with the games inbuilt benchmark to test a quick run around a London track.

 


 

Left 4 Dead 2

l4d2

l4d2AVG

All cards here were set to the maximum in game settings, with 4xMSAA, 8xAF @ 1680×1050. A very clear win for the Crossfire setup, surpassing the 4890 with ease. This particular part of the game is quite intesne graphically but the CrossfireX set up handles it well.

Batman Arkham Asylum

batman

batmanAVG

Towards the beggining of the test, all of the cards throw out similar numbers but as the test progresses, then we see the 4890 and the 5750+5770 stretch their legs. All 4 setups here put out nice numbers, but the the two 57xx single cards are running without any MSAA at all. Forcing 2xMSAA (Edge-Detect for 6x Sampling), the 4890 and the Crossfired cards still maintain high framerates as well.

Assassin’s Creed

asscreed

asscreedAVG

All of the cards were able to provide a playable experience at maximum settings, however you can see where the frame rates spike that the 4890 and the Crossfired pair reach higher peaks. A faster CPU would be benificial here to help maintain a higher minimum when there is a lot on screen at once.

 


 

Crysis Warhead

crywar

crywarAVG

You’ll have to forgive the lack of 4890 results; I did the 4890 testing first, misplaced the results, and since this is a quick and dirty article … I couldn’t be bothered to go back and do them again :p Anyway, the results for Crysis Warhead were gathered at Enthusiast settings. The 5750 couldn’t handle any MSAA, the 5770 allowed for 2x but was borderline unplayable, while the pair of cards was just right for 2xMSAA at 1680×1050.

Colin McRae DiRT2

 

dirt2

dirt2AVG

All settings where on high for this test, but do keep in mind that the 4890 is a DX10 only card, and therefore didn’t get a performance hit from running the game in DX11. The single 57xx cards didn’t use any MSAA, however the 4890 did. The Crossfire setup was also able to use 2xMSAA and put out similar numbers to that of the 4890 but in DX11. In all honestly, the only real noticeable difference while driving is shinier cars, and I must admit I would be tempted to drop back to DX9 with the pair of 57xx’s to try and maintain a frame at 60, as this for me personally gives a more positive experience with this game.

Overclocking

Both cards were overclocked independently for their reviews, but none the less I overclocked them in a pair until I had either reached their limits and/or no descernable difference in average frame rates could be seen.

overclock

Interestingly I was able to get a higher overclock out of the 5770 than I had previously, reaching final paired numbers of 940/1425 (default is 850/1200). The 5750 reached the same stable 865/1365 (default of 700/1100) I’d got in the review. Both cards have turned out to be good overclockers, and it’s nice to see the 5770 stretch it’s legs a little more than previously.

The DiRT2 results while overclocked are of particular interest, because while it is only a 5 FPS difference, that difference is enough to make me personally say “ok, go on then, we’ll play with DX11″. When not overclocked the frame rates some times dropped a little too low for comfort and I seriously considered dropping back to DX9. Left 4 Dead 2 also gets a substantial boost, putting the average well above the 60 FPS we aim for, and even in the inferno of the first level, the minimum never went below 45.

Final Words

So is it worth it? Yes, without a doubt. Would I have preferred two 5770′s? Of course, but with the overclocks, I’m not going to be missing out on much.

Set up was very simple, I didn’t have to do anything at all except install the hardware and then the drivers; Crossfire was enabled upon a reboot. The only trouble I’ve had using the setup day to day is due to a bug in Left 4 Dead 2 where on every map change I have to alt-enter in and out of the game or suffer an almost halved frame rate for the rest of that map. Apart from that things have just plain worked.

Gaming experience is on par or above that of a 4890, but things are cooler and quieter than your average 4890 even with the two cards running.

So if you can’t get a second 5770 to go with your existing card, if you can get a good deal on a 5750, you won’t be missing much, especially if you overclock the cards. If you already have a 5750, perhaps you can find a 5770 at the right price to pair with it. It all boils down to price and availability, but if you can’t get a matched pair, if our experience is anything to go by, then you won’t be sorry with a 5750 and a 5770 in the same machine.

Questions? Comments? Talk to us in the Forums.

HOME