• What
    is this?

    You've landed on the AMD Portal on AnandTech. This section is sponsored by AMD. It features a collection of all of our independent AMD content, as well as Tweets & News from AMD directly. AMD will also be running a couple of huge giveaways here so check back for those.

    PRESENTED BY

Gaming Performance: HD 6970M Nipping at the Heels of GTX 485M

If you’re looking for a mobile workstation—you know, like the Racer only with the Quadro FX 5000M in place of the consumer oriented HD 6970M or GTX 485M—then gaming performance is largely meaningless. You can refer to the HP EliteBook 8740w results as a rough estimate of where such a GPU lands in the gaming hierarchy, but without Sandy Bridge it might be somewhat slower in a few titles. For those looking to buy the Racer as a gaming notebook, this page is for you. It’s also for anyone wondering how AMD’s latest and greatest mobile GPU compares to NVIDIA’s power monster GTX 485M. After the 485M leapfrogged the previous generation mobile hardware, we were a little concerned that AMD might not be able to keep pace; our concerns it appears are unwarranted.

Battlefield: Bad Company 2

DiRT 2

Left 4 Dead 2

Mafia II

Mass Effect 2

Metro 2033

Stalker: Call of Pripyat

StarCraft II: Wings of Liberty

At our high detail settings (which typically means DX9/DX10 and no anti-aliasing, although we do enable 4xAA in L4D2), the Racer/6970M places third in most titles behind the 480M SLI and 485M. Perhaps more telling is that it boasts a performance increase of 16 to 75% over the GTX 460M, all for a minor increase in price. The lower end of that range is because our 900p High settings don’t put quite as much strain on the GPU, as we’ll see once we kick things up to Ultra quality. In fact, even at 1080p versus 900p, the 6970M is 8 to 30% faster than the 460M, and it’s up to 40% faster than the 5870M.

Looking at the head-to-head against the 485M makes things a bit more interesting. AMD leads by as much as 32% in L4D2, with a 4% lead in SC2 and a 5% lead in ME2. NVIDIA turns the tables in the remaining five games, with leads of 8% in BFBC2, 14% in DiRT 2, 16% in Mafia II, 1% in Metro 2033, and 5% in STALKER. It looks like NVIDIA’s 485M is in general the faster GPU, but the popular Source Engine games can really skew the results. Of course, most of these titles are hitting 60+ FPS at our High settings, so let’s set phasers for maximum power and fire a full spread of photon torpedoes.

Battlefield: Bad Company 2

DiRT 2

Left 4 Dead 2

Mafia II

Mass Effect 2

Metro 2033

STALKER: Call of Pripyat

StarCraft II: Wings of Liberty

Crank the settings up and watch the frame rates drop. Where most of the games were pulling around 60FPS or higher on the 6970M and 485M, at our Ultra settings only two titles are above 60FPS with the 6970M (three for the 485M), and two more fall below 30FPS. Now granted, 4xAA with Very High/Extreme settings in Metro and STALKER is more than most users would expect out of a notebook; turn off AA in STALKER and the frame rate doubles on the 6970M. Still, many desktops are able to handle this sort of resolution/settings combination on midrange ($200) hardware, so when an $1800+ notebook with the fastest GPU you can currently get falls short it’s a bit disappointing.

Comparing the 6970M and 485M once more, we find NVIDIA with a slight lead in five of the eight games, but if we call anything less than a 5% difference a tie there are really only three games where there’s even a moderate difference. NVIDIA is ahead by 8% in STALKER and 14% in DiRT 2; AMD leads by 10% in StarCraft II. Everything else is splitting hairs. The 6970M puts the hurt on the 460M once again, with performance now 35% to 82% higher than the MSI GT680R.

High Detail Average Performance

Ultra Detail Average Performance

As a final look at gaming performance, we averaged all the scores across the tested games, at both our High and our Ultra detail settings. Notebooks where we don’t have scores for all eight games are shown in light blue, so take those figures with a grain of salt. We do have full results for the 460M, 485M, and 6970M, though, so we’ll focus on those areas.

First, it’s interesting to note that the 485M and 6970M end up tied once we average our test suite frame rates. At High, the very good performance of AMD in L4D2 obscures the fact that NVIDIA usually leads, but at Ultra things are so close that calling it a draw seems reasonable. If you place more value on, say, STALKER you might go the NVIDIA route; StarCraft II addicts might prefer AMD. Ultimately, we’ll give NVIDIA the slight edge in subjective performance (and we still need to look at power and battery life), but cost definitely looks to be in AMD’s favor. As for the 460M, depending on which notebook you want to compare to and what detail settings, AMD ends up 45 to 54% faster at High detail (and still 19% faster running 1080p High), and the margin grows to 63 to 70% faster at Ultra detail.

Mobile Sandy Bridge: Why You Don’t Need a Desktop Battery Life and Power: For Better and for Worse
POST A COMMENT

47 Comments

View All Comments

  • JarredWalton - Thursday, March 17, 2011 - link

    Well, the whole notebook is an ES right now, so I can't really make a judgement call one way or the other on the 6970M. If it has an ES GPU, all Eurocom needs to say is, "We sent you a pre-release notebook." (And they already said that before shipping, so we're covered on that front.) This version I received also includes the Cougar Point bug, and if were ordering it right now that would make me pretty angry. Hopefully everything will be sorted out when the product shows up on their actual site, but I wouldn't be shocked if some fishy stuff happened on occasion.

    I'd also think AMD should have something to say about them shipping ES 6970M cards to customers, at least if they're not supposed to be available yet. What notebooks have shipped with the card to actual customers as opposed to reviewers? To my knowledge, every Clevo Sandy Bridge system is still affected by the Cougar Point bug and should be recalled, so my guess is only the X7200/Phantom 2.0 would be going out to customers.
    Reply
  • Windkull - Thursday, March 17, 2011 - link

    Jarred,

    They shipped some ENG samples on X7200s saying that they were not:

    http://forum.notebookreview.com/sager-clevo/556994...

    That said, I think consensus seemed to be that the ENG samples ran a bit too hot but might actually have more cores unlocked.

    Also, Sandy Bridge Clevos have been getting delivered for about a week now, I think they started shipping 10 days ago or so.
    Reply
  • JarredWalton - Thursday, March 17, 2011 - link

    See update note in the conclusion; thanks for the info! Reply
  • jah1subs - Thursday, March 17, 2011 - link

    I see all of this information about so many laptop/notebooks. But how do I know whether or not it has a matte display? My eyes have a problem with glare. Reply
  • JarredWalton - Thursday, March 17, 2011 - link

    Most companies will say either "glossy" or "glare type" on the LCD, or "matte" or "anti-glare". In general, though, assume everything is glossy unless otherwise noted... or buy a business laptop, which are usually anti-glare (ThinkPad, Latitude, ProBook, etc.) Reply
  • Ikefu - Thursday, March 17, 2011 - link

    So the new Mac book pros use switchable graphics from their AMD gfx cards to the build in Sandy Bridge gpu when unplugged. Why didn't they do that here or at least give you the option?

    Is there some technical limitation preventing this with a 6970M?
    Reply
  • mczak - Thursday, March 17, 2011 - link

    that's for sure. With ~10% less clocks and 25% less shader units than the GTX 485M, it ends up just slightly faster than GTX 480M.
    Though 6950M vs. 470M might be interesting - I'd expect the former to be faster but possibly not that much. It'll also depend on the exact clocks probably as those aren't really fixed for the mobile parts.
    Reply
  • Beenthere - Thursday, March 17, 2011 - link

    Won't buy any product with InHell inside. Reply
  • 5150Joker - Thursday, March 17, 2011 - link

    Jarred,

    Did you leave Physx enabled for the nvidia cards when running Vantage? It would explain why there was such a huge overall score gap.
    Reply
  • JarredWalton - Friday, March 18, 2011 - link

    Oh... hmmm, yeah, I probably did. I don't know that I've ever shut off PhysX for 3DMark Vantage. Reply

Log in

Don't have an account? Sign up now