Palit Radeon 4870 X2 CrossFire Video Card Review

[Apoptosis]

Palit Radeon 4870 X2 CrossFire Video Card Review

The Palit Radeon HD 4870 X2 graphics card is finally released now that AMD has lifted the embargo on the fastest graphics card in the world. See how this card is redefining the world of computer graphics with its pair of GPUs and 2GB of GDDR5 memory! See how the Radeon HD 4870 X2 performs in CrossFireX versus a number of leading NVIDIA GeForce GTX products.

If you want the absolute best when it comes to gaming graphics performance a pair of Radeon HD 4870 X2 cards are what you need. They are super fast graphics cards and at $549 each they aren't priced too high compared to where other cards have launched in the past few years. AMD has done a great job with these cards and they were able to take back the performance crown with the release of these cards. NVIDIA can no longer say they have the fastest graphics card and that has to hurt with the NVIDIA NVISION LAN party just a couple weeks away.

Article Title: Palit Radeon 4870 X2 CrossFire Video Card Review
Article URL: http://www.legitreviews.com/article/766/1/
Pricint At Time of Print: MSRP of $549

[stopthekilling77]

Just strange how those physix capabilities inflate the scores for nVidia.
Good read though, looks like a great card!

[Apoptosis]

Just strange how those physix capabilities inflate the scores for nVidia.
Good read though, looks like a great card!

Just 3DMark Vantage!!! None of the other benchmarks use the PhysX SDK, so it does nothing for performance or features!

[skier]

Too bad this was kinda spoiled by HotHardwares review i read earlier, they had a GTX280 SLI and TRI-SLI setup, but their max powerdraw was more like 850W most of the results were the same though

edit: 4870X2 Crossfire: 828W
TRI SLI GTX280 : 718W

[stopthekilling77]

Just strange how those physix capabilities inflate the scores for nVidia.
Good read though, looks like a great card!

Just 3DMark Vantage!!! None of the other benchmarks use the PhysX SDK, so it does nothing for performance or features!

Yeah, inflation of ego more than score xD

[geokilla]

Great Review Nate! However, I have a questions, which I asked before on another graphic card review.

Why is the 4870 X2 the slowest out of all the graphic cards tested for CoD4? According to Hardware Canucks, the 4870 X2 absolutely destroys the competition. At 2560X1600 on highest detail, 4X AA and 16X AF, they got average frame rates of 122.5 and a minimum of 54, while you guys got only 92FPS at 19X12. That's a huge difference of frames.

For Crysis, AnandTech managed 40FPS at 19X12, Hardware Canucks managed average of 36FPS and minimum of 21.6FPS at 19X10 with High Detail, 2X AA, and 64 bit while you guys got only 28FPS at 19X12. Now AnandTech doesn't state what quality they used, but it's probably something like high and above right? That we can only guess....

The test systems, such as CPU, RAM, etc have an impact on the GPU performance, but is the performance difference that big? I compared the test systems between you guys and Hardware Canucks, and the systems are basically the same. Same CPU except that theirs is at 3.8Ghz. Mobo are both X48 chipsets so the performance difference caused by the motherboard should be maybe 1 or 2%. They used high frequency DDR2 while you guys used high frequency DDR3. And the OS is Vista Ultimate, with them using 64 bit with SP1. Hard drive is different too, with you guys using a Raptor X while they used a Hitachi Desktar.

Sorry if I'm sounding negative, but it's just that the performance difference is really big.

[Apoptosis]

Well, for starters the extra 600Mhz is going to mean more for ATI than NVIDIA as they are more CPU bound. Secondly, in COD4 and in Crysis I run FRAPS at the toughest part in the real game I could find. Just like I used to in Bioshock and Laura Croft Tomb Raider.

In Crysis version 1.21, I used Very High settings at the toughest part with FRAPS, so the numbers are actually are good in my book. I could run the in-game benchmark if you want an unrealistic view of the game play using a time demo running faster than real-time if you like. Would be easier on me! I e-mailed the guys over at Anandtech to see what settings they used and how they benchmarked it.

[DX]

The bigger they are (Nvidia) the harder they fall. We should have seen this coming when Nvidia started their campaign of FUD against Crossfire and other ATI tech http://forums.legitreviews.com/about14617.html while at the same time hiding the fact that they have slowed down on their technical advances just like 3DFX did towards the end of their reign as graphics king. Now ATI looks like it has taken the crown. I have to wonder just how far Nvidia will fall now that many of their chips are being called "defective" and are looking at massive recalls http://www.theinquirer.net/gb/inquirer/ ... reportedly. It isn't looking good for them for the next year at least. Is this the beginning of the end for Nvidia? Will they follow 3DFX into oblivion?

[gwolfman]

Nice! Now if ATi can get the load temps down and power consumption I'd be sold. Hopefully Palit's redesign will help a lot! I sure hope so.

[skier]

Great Review Nate! However, I have a questions, which I asked before on another graphic card review.

Why is the 4870 X2 the slowest out of all the graphic cards tested for CoD4? According to Hardware Canucks, the 4870 X2 absolutely destroys the competition. At 2560X1600 on highest detail, 4X AA and 16X AF, they got average frame rates of 122.5 and a minimum of 54, while you guys got only 92FPS at 19X12. That's a huge difference of frames.

For Crysis, AnandTech managed 40FPS at 19X12, Hardware Canucks managed average of 36FPS and minimum of 21.6FPS at 19X10 with High Detail, 2X AA, and 64 bit while you guys got only 28FPS at 19X12. Now AnandTech doesn't state what quality they used, but it's probably something like high and above right? That we can only guess....

The test systems, such as CPU, RAM, etc have an impact on the GPU performance, but is the performance difference that big? I compared the test systems between you guys and Hardware Canucks, and the systems are basically the same. Same CPU except that theirs is at 3.8Ghz. Mobo are both X48 chipsets so the performance difference caused by the motherboard should be maybe 1 or 2%. They used high frequency DDR2 while you guys used high frequency DDR3. And the OS is Vista Ultimate, with them using 64 bit with SP1. Hard drive is different too, with you guys using a Raptor X while they used a Hitachi Desktar.

Sorry if I'm sounding negative, but it's just that the performance difference is really big.

as a rule every review will vary slightly, with differences in hardware, software, temps, and different actual benchmarks, you'll never get a replicable result, so they are averaged, maybe nate(and the LR crew) averages more results, maybe less, the main point to hardware reviews with performance being the major key, is: how did it do comparatively to the other cards tested, and in most reviews, the 4870X2 is the most powerful single-card solution in the market as of now.with an exception of maybe one or games, in certain points in the game, but you ALSO have to keep in mind that these aren't exactly the greatest drivers for the 4870X2, so performance will only increase over time

[Apoptosis]

For Crysis, AnandTech managed 40FPS at 19X12, Hardware Canucks managed average of 36FPS and minimum of 21.6FPS at 19X10 with High Detail, 2X AA, and 64 bit while you guys got only 28FPS at 19X12. Now AnandTech doesn't state what quality they used, but it's probably something like high and above right? That we can only guess....

I just shot DW over at Anandtech and he replied with this:

everything on High except shaders which were Very High.

All of our settings were on Very High:



This could be some of the difference.

[geokilla]

All right. Thanks for the clear up =)

Are you going to by any chance but the 4870 X2 into folding?

[jnanster]

Great Review, told me everything I need to know.
And thanks for the explanation to geokilla, always wondered why every reviewer get different scores for games.
Thanks for making it tough on cards, that’s were it counts.

My only comment is that in terms of real world, Crysis is the only game in your review (and everyone else’s) with serious framerate issues. In CoD4, the 4870x2 gets the lowest (91.87) and the highest (249.37) framerate scores. That “lowest” is a non-issue as it is super playable and you’d be an idiot to spend a half-a-grand just to get 249.37. On two of the games, the 4870x2 was completely useless in Crossfire.

I think the bottom line is that this is a great card for the money, but in the only game were it really needed to excel in crossfire, in Crysis, it flopped. It’s probably dumb to spend buckets of money to play a single game, but that single game is, certainly, the future direction of hyper-realistic graphics that we all want.
That’s what it’s all about right, immersion. Whether those graphics are built on platforms like CryENGINE 2 or not, we need graphic cards that kick butt, either singly or in pares or threes, and not graphic cards that get their butts kicked. The 4870x2 has issues in Crossfire.

[gwolfman]

Great Review, told me everything I need to know.
And thanks for the explanation to geokilla, always wondered why every reviewer get different scores for games.
Thanks for making it tough on cards, that’s were it counts.

My only comment is that in terms of real world, Crysis is the only game in your review (and everyone else’s) with serious framerate issues. In CoD4, the 4870x2 gets the lowest (91.87) and the highest (249.37) framerate scores. That “lowest” is a non-issue as it is super playable and you’d be an idiot to spend a half-a-grand just to get 249.37. On two of the games, the 4870x2 was completely useless in Crossfire.

I think the bottom line is that this is a great card for the money, but in the only game were it really needed to excel in crossfire, in Crysis, it flopped. It’s probably dumb to spend buckets of money to play a single game, but that single game is, certainly, the future direction of hyper-realistic graphics that we all want.
That’s what it’s all about right, immersion. Whether those graphics are built on platforms like CryENGINE 2 or not, we need graphic cards that kick butt, either singly or in pares or threes, and not graphic cards that get their butts kicked. The 4870x2 has issues in Crossfire.

Interesting take. I agree with some parts, especially since frames at 100 compared to 200+ don't make a difference. But people also spend big money for bragging rights and ultra high rez. Plus it shows that'll be a little more future proof than the last card, though spending the extra money isn't worth it (to some people, but then again it is to others). It's a unique product for a unique group of people, and they'll love it. Everyone else just benefits from the competition and lower prices. Yea!

[SergioRoadster]

I tought there was going to be released also a hd4870x2 with 1GB RAM... I have a hd4870 512MB, so because of the memory difference it isn't worth making a crossfirex with one of those 2GB hd4870x2, right?

[Skippman]

775W of current draw!?! JEBUS!

Although there is a certain cool factor when your mates are over and you show them your new rig.

Mate: "What happened to the lights?"
You: "Oh, that? They dim for a second every time I turn on my PC."
Mate: "WTF!?!"

I'm planning to upgrade my PC sometime this winter. I love my 1900XT Crossfire setup and am debating on getting this new setup but I fear my case may not handle it. I'd need a 1000W PSU and I'm not sure the cards would clear my harddrive tray. Also, I'm sure the cards would max out my CPU and I'm not to keen on COMPLETELY rebuilding just yet. Maybe I'll have to break down and just do it though.