Reviews - Double trouble: Sapphire Radeon HD 4670 in CrossFire

[HEXUS]

We pit two Radeon HD 4670 GDDR3 cards in a CrossFire configuration and see if they can beat out a Radeon HD 4850 and GeForce 9600 GT.

Read more.

[Jmatt110]

Interesting review

One thing though, the colors of those graphs looked almost identical. I think they need to be changed ASAP, it was a little confusing looking at them at first.

[cptwhite_uk]

But would you risk the possibility of crippled performance in some games for the sake of saving maybe £10 and getting a (probably) unperceivable performance increase?

I highly doubt it.

This comes close to breaking the rule of "SLI/CF is only for the ultra high end rigs at monster resolutions" but shouldn't be enough to tempt someone away from buying a 4850 or 9800GT in place of it.

Give me a 20% performance difference for the same cost as a single card and then I'll talk.

You also don't compare idle / load power useage and noise levels. Both of which I can only imagine would be worse than a single card solution...(even though it's a lower powered card). However the possibility to have two lower power card both passively cooled instead of one larger card which is impossible to passively cool...now that's an interesting thought

Edit: Typo bottom of page 5, should read "£100 to £110" I suspect.

[GaryRW]

I was suprised it didn't mention the possibility of buying one £50 card now, and then pairing a 2nd one in xfire in a few months time. Helps you spread the cost and in theory, the second card will be cheaper as costs drop.

I have no idea how quickly card prices drop though at this end of the market - is is just that by the time you buy the 2nd card, they'll be something newer and better value out??

[Tarinder]

GaryRW,

You've answered your own question.

The market changes so quickly that recommending buying a card six months down the line is fraught with potential problems.

We wanted to see what the performance of two cards was like, compared to a Radeon HD 4850.

[Tarinder]

Interesting review

One thing though, the colors of those graphs looked almost identical. I think they need to be changed ASAP, it was a little confusing looking at them at first.

We're still experimenting with the graphs. I thought that this method, where the GPU companies are clearly defined, would be better, as the cards position on the graphs doesn't change.

Let me know if you still think differently and we'll change it.

[Kerotan]

Decent review- impressive to see how well Xfire is scaling these days, I would seriously consider this having a P45 mobo, but I think single-card solutions will always be better in terms of stability and game support. Passive x2 versus active x1 is an interesting idea though, would certainly make for a quieter PC.

[Tetras]

The market changes so quickly that recommending buying a card six months down the line is fraught with potential problems.

The most obvious drawback being in six months it is quite likely a faster single card will be available for the same money.

I really have no faith in SLI/Crossfire as a long-term platform, sure it is fun to mess around with and useful in a few circumstances but is it ever going to be a viable mainstream option? I'd pick a 4850 every time.

[will101]

Has anyone told Scan they are meant to be selling the 4670 for £50-55? Only they are listing them at £68+ which is a shame as I was considering eBaying my 8800GTS 320mb and buying a 4670 with the proceeds...

[Webby]

Don't suppose you want to test this again using a x38 or x48 board do you? Using P45 you are limiting the crossfire setup to run at 8x rather than 16x. Then again if your buying a ~£100 graphics card are you likely to be buying a ~£110+ (X38) or ~£130 (X48) motherboard.

[loki]

The graphs are a nice touch, but personally I'd prefer each item to be a different colour and the colour scheme to be the same throughout the review.

It does look like crossfire is reaching quite a mature stage now, with some 80%-100% increase through 2 cards. I actually thought the technology would just die out when I first saw SLI and Crossfire because there was so little benefit.

[RoganJosh]

Different colours for different cards would be useful on the comparison graphs - or maybe it's just my eyes?


er, sorry - didn't see the last post before I put my comment in - anyway, I agree with loki

[Aranthos]

Different colours for different cards would be useful on the comparison graphs - or maybe it's just my eyes?

The graphs are a nice touch, but personally I'd prefer each item to be a different colour and the colour scheme to be the same throughout the review.

One thing I accidentally came across was that if you click on a card in the key at the top-right of each graph it hides its bar. Made it a bit easier to work out what was what, but still...

[EDIT] Ooh, just had an idea! How easy would (or wouldn't) it be to add an arrow from each item on the key to its appropriate bar on the graph? Or even return to the old method of having the item name on the bar itself? That would be easier I imagine but I've not had a whole lot of experience with Flash so...

[Jmatt110]

We're still experimenting with the graphs. I thought that this method, where the GPU companies are clearly defined, would be better, as the cards position on the graphs doesn't change.

Let me know if you still think differently and we'll change it.

It's not that, I like the companies being clearly defined However having different shades of the color so they are a little more distinguishable would be nice.

[IronWarrior]

Great review, I got the 4670 IceQ Turbo version and been wanting to CF it ever since I got it, seeing that you can double the FPS in some games is awesome.