4850x2 smacks GTX280

[Unknown Soldier]

According to AMD, even its watered down he version of the R700 with DDR3 memory will make short work of Nvidia'a flagship GTX280.

The company claims a performance lead of anywhere from 8 to 34 percent in current titles against the GTX280. As the HD4850 X2 is set to sell at $399, or around €270-290 in the EU, this sounds like an impressive score. The card runs at 625MHz, and unlike the HD4870 X2 it features 2GB of cheaper DDR3 memory, which goes a long way in keeping the cost down.

There's also a chance vendors will offer a 1GB version, but this doesn't seem very likely, although it would allow vendors to offer it at an even more attractive price. Even with 2GB the new card will probably provide you with the best price/performance ratio in the high end.

Much has been said about the R700 lately, but the DDR3 flavored HD4850 X2 might just be the card to hit the sweet spot. It will end up significantly cheaper than the HD4870 X2, but it will still pack enough punch to outperform the GTX280 while selling for some €50 less.

The only downside is that you'll have to wait a bit more to actually get one. We're exepecting it to hit retail sometime in the second half of September, but it looks like the wait will be worthwile indeed.

News Source: http://www.fudzilla.com/index.php?option=com_content&task=view&id=8899&Itemid=1

 

[Freak'n Big Panda]

Pricing seems off. Two 4850s are cheaper, granted you don't get the 1GB VRam but AIBs should rememdy that situation shortly for a very minimal price premium. $349 would have been profitable and really would have rubbed salt in NV wounds, but I suppose thats dangerously close to 4870 territory.

 

[Davros]

how does it compare to a 9800gx2 ?

 

[compres]

I would guess within 5% of 4850 CF.

 

[Moloch]

more power consumption (and heat) should atleast result in a faster card
Of course when crossfire doesnt scale it's going to be a lil faster than a 9800gtx..
That's the problem with ati's multi gpu for the high end thing they got goin..

 

[Tchock]

more power consumption (and heat) should atleast result in a faster card
Of course when crossfire doesnt scale it's going to be a lil faster than a 9800gtx..
That's the problem with ati's multi gpu for the high end thing they got goin..

The TDP is in the same league as the GTX 280.

230W for the 4850X2 vs 230 (236?) W for the GTX280.

Plus, flashing the BIOS to the 160Mhz mode for the 4850s put them below GTX 280 idle usage, although crossfire makes it exceed the 280 I'm sure things shouldn't be that far. Should be much lower than the 4870X2 though.


BTW I'm rather tired of the profile thing going on. That AFR friendly exe hack or FEAR.exe works with most modern titles. In fact, I don't think any non-modern titles without CF scaling would need the extra boost...

 

[Moloch]

The TDP is in the same league as the GTX 280.

230W for the 4850X2 vs 230 (236?) W for the GTX280.

Plus, flashing the BIOS to the 160Mhz mode for the 4850s put them below GTX 280 idle usage, although crossfire makes it exceed the 280 I'm sure things shouldn't be that far. Should be much lower than the 4870X2 though.


BTW I'm rather tired of the profile thing going on. That AFR friendly exe hack or FEAR.exe works with most modern titles. In fact, I don't think any non-modern titles without CF scaling would need the extra boost...

Think again http://techreport.com/r.x/radeon-hd-4870-x2/power-load.gif
I dont think having xfire on a stick is going to make up for that power consumption deficit betweem the two...

 

[Dave Baumann]

Power utilization will vary from app to app between architectures (and even differing configurations between architectures) so a single snapshot doesn't necessarily tell you much.

 

[Tchock]

Think again http://techreport.com/r.x/radeon-hd-4870-x2/power-load.gif
I dont think having xfire on a stick is going to make up for that power consumption deficit betweem the two...

http://images.anandtech.com/graphs/amdradeonhd4870x2launch_081208022737/17230.png

Hmm?

Bins, bins, biiiiiins...

 

[ChrisRay]

The only way to properly test power consumption is to get the latest game with the highest fillrate and shader requirements to make any sort of valuable conclusion. According to techreport they used 4xAA/16xAF @ 2560x1600 in HL2. Not the most demanding game. So you can get a pretty good idea of how they measured power consumption.

http://techreport.com/articles.x/15293/10

http://techreport.com/articles.x/15293/6

Nothing there looks all that surprising considering how they are scaling from 1 to multi GPU.

 

[CarstenS]

Power utilization will vary from app to app between architectures (and even differing configurations between architectures) so a single snapshot doesn't necessarily tell you much.

So, which app would you recommend to get the highest load known so far? I know for a fact, that 3DMark isn't it.

 

[Tchock]

I nominate Furmark (8.6 drivers for Radeons). Eeeeeebil.

 

[I.S.T.]

Crysis DX10 Very High settings with max AF, resolution and 4x4 supersampling?

 

[Moloch]

http://images.anandtech.com/graphs/amdradeonhd4870x2launch_081208022737/17230.png

Hmm?

Bins, bins, biiiiiins...

Well using AT' s numbers xfire on a stick is using more power than regular xfire and with that in mind the 4850 x2 will use more power than xfired 4850 which will put it using more power than th 280 perhaps?
TR has the 4870 x2 and 4870 CF using nearly the same ammount of power, perhaps a consequence of using an older, less demanding game than AT?

 

[pjbliverpool]

BTW I'm rather tired of the profile thing going on. That AFR friendly exe hack or FEAR.exe works with most modern titles. In fact, I don't think any non-modern titles without CF scaling would need the extra boost...

The key word there being hack.

I don't want to have to hack my games just to get them to work properly with my GPU. Especially when such hacks may have other unwanted side effects.

Don't get me wrong, faster + cheaper is clearly a great thing but its not a clear win because its xfire.

 

[BRiT]

Well using AT' s numbers xfire on a stick is using more power than regular xfire and with that in mind the 4850 x2 will use more power than xfired 4850 which will put it using more power than the 280 perhaps?

The 4870 X2 uses 2Gigs of DDR5. Using two 4870 crossfire uses only 1Gig of DDR5. That may well be the difference in power consumption between Crossfire and X2.

Do we know if 4850 X2 will use more memory than two 4850 crossfire? Perhaps only then will we know if 4850 X2 will draw more than two 4850 Crossfire.

 

[Moloch]

The 4870 X2 uses 2Gigs of DDR5. Using two 4870 crossfire uses only 1Gig of DDR5. That may well be the difference in power consumption between Crossfire and X2.

Do we know if 4850 X2 will use more memory than two 4850 crossfire? Perhaps only then will we know if 4850 X2 will draw more than two 4850 Crossfire.

I know but he brought up an interesting point, TR's numbers and AT's numbers for power are quite different.. as I commented on..

 

[Freak'n Big Panda]

The 4870 X2 uses 2Gigs of DDR5. Using two 4870 crossfire uses only 1Gig of DDR5. That may well be the difference in power consumption between Crossfire and X2.

Do we know if 4850 X2 will use more memory than two 4850 crossfire? Perhaps only then will we know if 4850 X2 will draw more than two 4850 Crossfire.

it's not that the quantity of vram is lower it's that the 4850x2 uses gddr3 not gddr5, that's what ends up lowering power consumption. Well that along with the lower core clock.

 

[AlphaWolf]

it's not that the quantity of vram is lower it's that the 4850x2 uses gddr3 not gddr5, that's what ends up lowering power consumption. Well that along with the lower core clock.

Is GDDR5 higher consumption?

 

[Freak'n Big Panda]

yea by a pretty big margin