ATI Radeon HD 5770 CrossFireX Performance

[Rob Williams]

Want to purchase a Radeon HD 5870, but can't find one in stock? One alternative to consider is instead purchasing two Radeon HD 5770's to take advantage of CrossFireX. Not only does this solution save you up to $80 at current pricing, but it proved in our results to offer even better performance in select titles, such as with Modern Warfare 2.

You can read our full article here and discuss it here.

[Doomsday]

love how technology works! keeps getting better and better!

can the Crysis issue be a driver problem?! i've noticed Nvidia and ATI releasing different patches to increase performances of particular games....

[Psi*]

I just bought 1 of these for the new build. I haven't yet received it yet or anything else. I am considering getting another in a week or 2 after things are running.

My selection choice was about anticipated compatibility, DirectX 11, & price.

And good to know about performance.

[Rob Williams]

I'm not sure about Crysis being a driver issue, because I was using the latest version of Catalyst. As I mentioned, no one else around the Web that I could see had nary a complaint, so I'm not sure of the story there.

Psi*, if you buy a second one, you won't regret it.

[gibbersome]

Great review Rob! 5770's in crossfire are legit alternatives to 5870!

The major downside I see (aside from Crysis) is the additional power requirements, but also that it won't future proof you like two 5870's will. Which isn't too bad at all!

[Rob Williams]

Thanks gibbersome, and yeah, definitely a legit alternative. I couldn't believe it when I saw better performance this way than with a single HD 5870. Certainly not a bad thing, though.

This is the number of comments to what is an important piece of work???

Hmmm, I just sold an HD 5870 that I bought the first week they came out as I am now traveling and thought I could make an extra buck or two (which I did), I was impressed by the technology and it was indeed a solid upgrade to the GTX 285 I previously owned.

Personally I am having difficulty believing that 2 5770's in crossfire can beat one 5870 in any of your test, after all even if crossfire scaled 100% then the 2 5770's would still be giving up half its memory bandwidth to the 5870 and at higher resolution and with FSAA enabled the mem bandwidth makes all the difference.
I for one do not believe that crossfire scales perfectly, so although I can believe that 3 5770's can potentially be faster than 1 5870 I am unable to see how 2 can beat it....

What are your thoughts behind this behavior?

[Rob Williams]

Hi Greg, and thanks for the post

Originally Posted by Greg

Personally I am having difficulty believing that 2 5770's in crossfire can beat one 5870 in any of your test, after all even if crossfire scaled 100% then the 2 5770's would still be giving up half its memory bandwidth to the 5870 and at higher resolution and with FSAA enabled the mem bandwidth makes all the difference.

You certainly aren't alone. When I looked at the results after testing, I didn't know what to make of it, but regardless of how it happens, those are the results we saw, and I'm confident anyone doing the same test will see the same thing. Could it be the improved architecture over the previous generation, or improved drivers? I don't know, but given that we use the exact same test machine and fresh OS install before each testing, not to mention completely manual benchmarking (no timedemos), our results are what they are.

What's somewhat interesting is that our 3DMark Vantage results backed me up a little bit:

http://techgage.com/article/ati_rade...performance/10

There's a wider variance at 1680x1050, but at both 1080p and 2560x1600, the HD 5770's in CrossFireX and the HD 5870 are scary close in performance... there's just a ~6% difference between the two configurations in the latter. For raw performance, the HD 5870 comes out ahead, but in real-world gaming, there are cases when they come out about even (it's easier to see differences when we're dealing with numbers in the thousands... not so much when talking about 40 - 50 FPS).

If I get a chance today, I'll load up that machine and test out both configurations again briefly just for the sake of it.

[Rob Williams]

Originally Posted by Greg

I for one do not believe that crossfire scales perfectly, so although I can believe that 3 5770's can potentially be faster than 1 5870 I am unable to see how 2 can beat it....

I re-ran both configurations (two runs each, averaged) with the latest Catalyst 9.12 drivers, and here are the results:

(3Dmark is 'Overall - GPU - CPU' and games are 'Minimum / Average')

ATI Radeon HD 5870

3DMark Vantage - Performance: 18698 - 17830 - 21896
3DMark Vantage - Extreme: 9155 - 8883 - 21925
Modern Warfare 2: 45 / 81.735
FEAR 2: 68 / 93.404
GRID: 83 / 103.221

ATI Radeon HD 5770 CrossFireX

3DMark Vantage - Performance: 17407 - 16327 - 21721
3DMark Vantage - Extreme: 8361 - 8099 - 21732
Modern Warfare 2: 43 / 85.35
FEAR 2: 58 / 87.895
GRID: 82 / 103.201

The results aren't far different from what we saw in the article, except that the "Extreme" 3DMark score has a slightly larger variance between the two, but otherwise, performance is either just under, exactly the same, or a bit better.