also @ TechSpot: Intel says Haswell will improve battery life by 50 percent

Nvidia GeForce GTX 480 SLI vs. ATI Radeon HD 5870 Crossfire

By

On June 23, 2010, 7:10 AM

Recently we revisited the battle between the high-end Nvidia GeForce GTX 400 and ATI Radeon HD 5000 graphics cards series using updated drivers and a new testing method that saw us abolish all time demos. This comparison was limited to single GPU testing at what we consider to be the mainstream screen resolution for high-end PC gaming.

But for those amongst you that like to take things to the next level, gaming at 2560x1600 with 30-inch displays and beyond. For extreme users that do not necessarily care about value, power consumption or even heat... who offers the best gaming solution? This question leads us to a new showdown between ATI and Nvidia, only this time using a pair of GeForce GTX 480 graphics cards in SLI versus a pair of Radeon HD 5870 using Crossfire technology.


For many of us the prospect of spending $400 on a Radeon HD 5870 graphics card seems a little crazy, let alone the $500 Nvidia is asking for the GeForce GTX 480. So it goes without saying that those willing and able to purchase two of these mighty graphics cards are in the minority. Still, such configurations do exist and those of you looking to go down that path will be interested to check our findings.

Read the complete review.

No tags on this story

User Comments: 110

Got something to say? Post a comment
  1. It doesn't make sense to have a physics card over an additional GTX460.

    Plus, you'll need some extra room for that physics card if you already have two GTX460's, and the return's aren't very high.

  2. As a percentage Crossfire scales just fine-and in most cases if you're comparing to the GTX 480 for instance- CFX will scale better than SLI as a percentage. Mainly because the GTX is starting from a relatively higher fps as a single card. The opposite is also true for the reverse (a single HD 5850/5870 posting higher single card numbers, ~10% and 22-27% respectively, than a GTX460 for example). You can add into the mix benchmarking anomolies such as capped framerates and/or system component limitations (bottlenecks captain), in-game IQ settings and whether any settings are forced at driver level, and whether the game favours SLI, Crossfire or any GPU scaling

    The only metric probably worth taking into account is probably fps, either as an average, if the benchmarking sample is large enough (and ideally it should cover the resolutions commensurate with the cards likely pairing, a range of games widely played including RTS, FPS, RPG, Sim's etc.) or a aggregate fps total

    I don't doubt you're getting close to 100% scaling in some games with CFX- especially if one of the games you're benching is Metro 2033 (IQ and screen res dependant of course)

    One last Q though: How would a dedicated PhysX card compare with a GTX460 doing the job instead.

    If the PhysX component is intensive (the aforementioned Metro 2033 or Dark Void, Mirror's Edge, Batman:AA etc) then offloading PhysX to a dedicated PPU will raise your overall framerates (esp. the minimums). Any reasonable 256Mb frame buffer/ 1Gb memory nvidia card of the previous generation (i.e. 8800GT/9800GT/GTX/GTS250) would suffice as a good PPU.

  3. Thanks once again DBZ.

  4. One point I think many readers may not understand is that crossfire mostly does not work at all. The benchmarks are extremely misleading in this regard. They make both Crossfire and SLI out to be completely transparent performance enhancements. They are not -- or at least Crossfire is not.

    I recently purchased another 5830 for use in a crossfire configuration. It was not until after installing and much digging for data that I discovered that crossfire does not work in windowed mode -- full screen only mode (which I never use). Not only that, in full screen mode it has serious problems: it's SLOWER than non-crossfire windowed mode for the games I play, it has very annoying graphic artifacts on the screen, it doesn't support all the resolutions and modes of non-crossfire, and it may require 3rd party tools to enable crossfire mode.

    All this combines into a "do not use multiple graphics cards, ever" for me. It's not a backhanded endorsement of SLI. On the contrary, if this is what AMD thinks provides decent competition with SLI, then SLI is likely equally flawed. No doubt there must be some use for NxGFX cards, but from my experience with crossfire, it must be an extremely narrow use restricted to a very few mass-market games that specifically support multiple-gpu systems.

    -=-=-

    IMHO, I'd love it if the reviewers would simply ignore Crossfire/SLI or at least put VERY large caveats on it when showing comparisons to single-card solutions. They're not comparable on basic functionality. It's very misleading to compare something that works with something that doesn't.

  5. One point I think many readers may not understand is that crossfire mostly does not work at all. The benchmarks are extremely misleading in this regard. They make both Crossfire and SLI out to be completely transparent performance enhancements. They are not -- or at least Crossfire is not.

    I recently purchased another 5830 for use in a crossfire configuration. It was not until after installing and much digging for data that I discovered that crossfire does not work in windowed mode -- full screen only mode (which I never use). Not only that, in full screen mode it has serious problems: it's SLOWER than non-crossfire windowed mode for the games I play, it has very annoying graphic artifacts on the screen, it doesn't support all the resolutions and modes of non-crossfire, and it may require 3rd party tools to enable crossfire mode.

    All this combines into a "do not use multiple graphics cards, ever" for me. It's not a backhanded endorsement of SLI. On the contrary, if this is what AMD thinks provides decent competition with SLI, then SLI is likely equally flawed. No doubt there must be some use for NxGFX cards, but from my experience with crossfire, it must be an extremely narrow use restricted to a very few mass-market games that specifically support multiple-gpu systems.

    -=-=-

    IMHO, I'd love it if the reviewers would simply ignore Crossfire/SLI or at least put VERY large caveats on it when showing comparisons to single-card solutions. They're not comparable on basic functionality. It's very misleading to compare something that works with something that doesn't.

    Possibly the most solipsistic, anecdotal post to date. There is obviously not a great demand for using CF in widowed mode. I'm sure that every single review, benchmark and article on the planet is wrong, or they are just getting extraordinarily lucky as it works for them.

    , but from my experience with crossfire, it must be an extremely narrow use restricted to a very few mass-market games tha

    Why don't you provide a list of these games that do not support CF/SLI?

    If you can find them, they are probably pre 2007, and if they do not support it...they don't need it in the first place.

  6. I'm with Red on this one - almost all the games I've played support CrossfireX just fine - and if you need a little reassurance then enable the logo to appear when an application uses CX. If you're playing windowed then you probably didn't need the performance boost of CX or SLI anyway - just get a decent single GPU card.

    The only disappointment (due to lack of proper CX support) that I've had recently was FFXIV - it apparently does just fine with SLI but crap with CX. The CX logo will appear just fine and the second GPU will run at a fairly high level of usage, but your performance will actually go down with CX enabled. I wonder if this is the title you're talking about...

    Edit: I also thought it was fairly well known that SLI and Crossfire don't function properly on windowed apps.

  7. They didnt make the game for people that can spend 600 dollars every 6 months to upgrade hardware. They made it for the people who can afford 15 usd a month to play. Papa.

  8. Thank you guest,why did you take soo long to come on here and tell me that,i just bought myself a second 4770 for a performance boost and now i find out that i wasted money on the card.

    I thought the extra fps i'm getting was from the second card.

    Why did so many people waste money on a 4890 if a single 4770 could get higher fps?

  9. What I need to know it the bench marks for ATI Radeon HD 5870 x2GB over the GeForce GTX 480

    I can't find that anywere and anyone help me.

  10. Tom's Hardware GPU gaming charts.

    For most gaming benchmarks the 2Gb and 1Gb cards offer pretty much identical performance. The 2Gb card, in general terms, only makes sense if you plan on a multi-monitor setup.....which of course doesn't then make it a direct comparison with the GTX 480

Recently commented stories

Post a new comment

Social Login & Guest Posting TechSpot Members
Login here or sign up for free,
it takes about a minute.
Get complete access to the TechSpot community. Join thousands of technology enthusiasts that contribute and share knowledge in our forum. Get a private inbox, upload your own photo gallery and more.
TechSpot on:

Subscribe to TechSpot

Get free exclusive content, learn about new features and breaking tech news.