Mobile Gaming Showdown

The 4GB of Kingston DDR3 coupled with the second-fastest quad-core in Intel's mobile lineup should ensure an extremely capable showing in our regular benchmarks. What we're really going to be interested in is seeing who winds up with the fastest mobile graphics part, barring the value-adds from vendors. Here's a refresh of the configurations of our review units:

AVADirect Clevo W860CU Testbed
Processor Intel Core i7 820QM
(4x1.73GHz, 45nm, 8MB L3, Turbo to 3GHz, 45W)
Memory 2x2GB DDR3-1333 (Max 2x4GB)
Graphics ATI Mobility Radeon HD 5870 1GB GDDR5
(800 Stream Processors, 750MHz/4GHz Core/RAM clocks)

(128 CUDA Cores, 576MHz/1.5GHz/2GHz Core/Shader/RAM clocks)
Display 15.6" LED Glossy 16:9 1080p (1920x1080)
Hard Drive(s) Corsair 128GB Nova Series SSD
Optical Drive Blu-ray and DVD+/-RW Drive
Battery 3-Cell, 12V, 42Wh battery
Operating System Windows 7 Home Premium 64-bit
Pricing $2477.12 (Radeon HD 5870)
$2555.60 (GTX 285M) as configured from AVADirect

These are gaming notebooks, so let's cut straight to the meat and run some gaming tests. For gaming-level hardware, we're going to need settings that really push these GPUs. Mercifully, the W860CU comes with a beautiful 1080p screen that allows us to see just how far these solutions can go. At the top of the line, the GeForce GTX 285M and Mobility Radeon HD 5870 should both produce fluid gameplay at the W860CU's native resolution. We'll also run 1600x900 results to see how dropping down to HD+ affects performance.

We used the built-in benchmarks for every game except Mass Effect 2, which we measured using FRAPS by playing through a section of the game. For games that support DirectX 11, we also ran benchmarks using it on the Mobility 5870 to see if there was a performance benefit or hit.

Batman: Arkham Asylum

Battlefield: Bad Company 2

Crysis: Warhead

DiRT 2

Empire: Total War

Far Cry 2

Left 4 Dead 2

Mass Effect 2

Stalker: Call of Pripyat

In every case except for DiRT 2, the Mobility Radeon HD 5870 meets or beats the performance of the GeForce GTX 285M, and in DiRT 2 the 5870 remains perfectly playable even when DirectX 11 is enabled.

Though we average the scores of the four benchmarks in S.T.A.L.K.E.R., it does bear mentioning that while the GTX 285M produced higher numbers in the first three, the 5870 consistently hit 30fps in the Sunshafts benchmark, a feat the 285M could not pull off.

Unfortunately, the news isn't as good as we'd like. Crysis: Warhead still remains the outlier; at “Gamer” settings and 1080p, neither GPU can make it over the 30fps hurdle. A difference between notebook and desktop hardware is expected, but Cypress-based desktop cards can handle Crysis: Warhead at Enthusiast settings at 1080p with performance to spare. These notebook chips are still sitting on the low-end of the last generation's top of the line.

Likewise, one can't help but be underwhelmed by the performance of the Mobility Radeon HD 5870. The performance of NVIDIA's mobile graphics parts has been a sitting target for a long time now, and while the victory is still technically decisive, it's really not the supreme triumph it ought to have been. G92 has been around for a long time now, with NVIDIA resting on its laurels. Anyone hoping AMD's new chip would shock NVIDIA out of its complacency has to be disappointed by these results. Needless to say, GTX 480M shouldn't have too much trouble reclaiming the mobile crown (albeit with higher power requirements that might make it difficult to fit into a 15.6" chassis).

AVADirect Clevo W860CU Overview Speedy Application Performance as Well
Comments Locked


View All Comments

  • rscoot - Friday, June 4, 2010 - link

    I cannot help but to think that the reviewer in this article is either new to enthusiast computing in general, or just plain ignorant of certain facts.

    First of all, I can think of no point in time where the desktop part and laptop part had the same number of pipelines/shaders and were called by the same model number. The mobility 9700pro, for example was a 4 pipeline part, the desktop part was 8 pipelines. The 6800 and 7800 mobile parts that you extol the virtues of are the same way. It is simply unrealistic to expect ATi and nVidia to cram full sized desktop parts into a laptop chassis.

    Secondly, to act as if a 25% increase in performance over the previously fastest part is something that is trivial and not good enough makes the reviewer look like a nVidia apologist or again, just plain ignorant. Power use in GPU's in general (until the radeon 5800 series) has increased at an insane rate, and manufacturers have about hit the wall in regards to how much power they can push through a piece of silicon and not have it burst into flames. In the constrained environment that the notebook form factor presents, these power requirements become even more important. The 480M part just exacerbates the situation with its 100W TDP. There is no way you can fit almost 200 watts of power use system wide into a chassis that is even somewhat portable. This part is going to be relegated to the huge 17/18" mobile workstation chassis. Considering how much they cut down Fermi to fit even in that bloated TDP envelope, I do not believe that it will be a great deal faster than the 5870, certainly not 25% faster, which was blasted as trivial by this very reviewer. So will your opinion be the same about the 480M, especially considering its using at least 33% more power than the 5870?

    The methodology and presentation of the data in this review could also use some work. Why lump the 1600x900 and 1080p results on the same graph when you have data for almost all the laptops involved at both resolutions? It just makes the graph look cluttered and confusing and makes it difficult to figure out how fast each laptop is at a given resolution/detail setting.

    So please, clean up your reviews, do a little bit of research on the subject that you're reviewing and next time you won't come off as so amateurish.
  • JarredWalton - Friday, June 4, 2010 - link

    Dustin and Vivek are also doing laptop reviews, along with me. We had to settle on a base resolution, and going forward we'll pull 1080p and other scores out of the charts and stick with 1600x900 on high-end notebooks (along with the native LCD resolution for the reviewed laptop). The reason is that it's a resolution virtually everything supports, so like other reviews we'll show an apples-to-apples at 1600x900 along with performance at native. The W870CU we tested only has 1600x900, and I have another laptop that's a 1680x1050 display (gasp! 16:10 is still around!?), so hopefully it will make more sense as we get more notebooks into our revised list.

    As far as the "only 25%" you need to correct that to read: at absolute best it might be 25% faster. Actually, in the benchmarks we ran the largest margin of victory for the HD 5870 is 18% (Left 4 Dead 2 with 4xAA). Battlefield: Bad Company 2 also has a lead of 16% (with DX11 2xAA at 1080p vs. DX10 2xAA at 1080p) and Far Cry 2 is a ~15% lead. Beyond those two games, most of the other titles are a wash. Batman and Crysis are a 3-9% lead for 5870; DiRT 2 actually favors the 285M by up to 22% (in DX9 mode -- turn on DX11 and performance on the 5870 drops even more). Empire: Total War favors NVIDIA by 9%. Mass Effect 2 has 5870 ahead by 10% at HD+ and behind by 9% at 1080p--go figure. Finally, Stalker: Call of Pripyat is a tie (within 2%), though the DX11 modes improve quality without a noticeable impact on performance.

    So yes, the performance lead of ATI's part over the aging GTX 285M is underwhelming. It's still the better part, but it's not significantly faster and only (potentially) uses a bit less power. If you're after performance in a notebook, I'd have to say the more interesting match will be the new GTX 480M, simply because it's not two year old technology from NVIDIA. It will be power hungry and hot for sure, but anything capable of running games generally falls into that category. And unlike you, I expect the 480M will be more like an AVERAGE of 35% faster than the 5870, so it will have a linear increase in performance and power draw. (ATI could likely match it if that's the case... just raise clocks on the 5870 part and increase power to 100W.)

    We'll find out some time this month exactly how the two really compare. Given we already have 285M leading 5870 in a few titles, and on the desktop the 480 has outpaced the 5870 by 15-25%, I expect 480M will take an easy lead in every title. As for the size and power, we're already very much aware of that, and we say as much in the conclusion.
  • Dustin Sklavos - Friday, June 4, 2010 - link

    "First of all, I can think of no point in time where the desktop part and laptop part had the same number of pipelines/shaders and were called by the same model number. The mobility 9700pro, for example was a 4 pipeline part, the desktop part was 8 pipelines. The 6800 and 7800 mobile parts that you extol the virtues of are the same way. It is simply unrealistic to expect ATi and nVidia to cram full sized desktop parts into a laptop chassis."

    Seriously? I'll limit myself to the top end:
    Mobility Radeon X800 XT, Mobility Radeon X1800 XT, Mobility Radeon HD 3850/3870, Mobility Radeon HD 4850/4870, GeForce Go 7800 GTX, GeForce Go 7900 GTX, GeForce Go 7950 GTX

    Actually up until this generation ATI in particular has had a surprisingly good run of matching mobile model numbers with the chip powering them, with very few exceptions. Nvidia was also doing extremely well up through the 8600M GT.

    If ATI was able to reduce the TDP of the RV770 to cram it into a couple laptops, and Cypress has very similar thermal characteristics, is it not reasonable to suggest they might have been able to do it? Given the rarity of the Mobility 4850/4870 one can speculate that binning the parts that could run at such low voltage was difficult, so for this generation they just opted for Juniper instead of Cypress.

    ALSO consider that Nvidia skipped trying to cram the GT200 into notebooks, but FERMI can be cut down and fit in?
  • sheltem - Friday, June 4, 2010 - link

    This laptop is wayyyy tooo freakin' heavy for a 15.6". The Envy 17 weighs 7.51 pounds and is way cheaper. The only downside is the slightly weaker graphics card: Mobility 5850.
  • Nurn - Friday, June 4, 2010 - link

    I would have thought that your conclusion would end with "In summary, the ATI video card is both faster and less expensive than the NVidia part, making it the overall winner and a no-brainer when selecting options for the AVADirect Clevo W860CU".
  • Setsunayaki - Friday, June 4, 2010 - link

    I love to be critical so here it goes.

    This laptop fails to define itself by being too close to a desktop. So close that it loses abilities laptops are known for, while simultaneously being an "overpriced" mobile desktop.

    Gaming Benchmarks at 1600x900 when it has a 1920x1080 screen is a real no-no, along with a battery life of 55 - 75 minutes idle is awful. This means a game will cause this laptop to lose all power in 30 - 50 minutes...

    Due to the security nature of dealing with a lot of open Wi-Fi out there, I prefer running Linux with Nvidia, because Nvidia drivers on Linux are better than the ATI counterparts. Virtualization also becomes far more important on laptops, specially if you are in a team who has one laptop assigned to them off-site.

    A rule of thumb...If you are going to spend $2000 - $3000 on a laptop, the battery better be good enough to run for at least 4 - 5 hours under your normal work load. This is the longest time the average worker tends to spend off site on a laptop. Three hours tends to be average time a worker spends off site with a laptop [college courses, meetings and exchanges/presentations come to mind.]
  • Death666Angel - Saturday, June 5, 2010 - link

    Sorry, have to disagree with your points against this. As stated multiple times, this is a desktop replacement gamer rig. Not a workstation where you need x-hours of battery life, which the like of Dell and HP offer through strap-on batteries and DVD-drive batteries. He explained this in the review.
    He also tested the laptops with 1080 native resolution, so no complaint there. (Maybe you just missed it since it is in the same graph and kinda confusing.) Going for "the lowest common denominator" is a smart thing imo, because it allows for laptop v laptop comparisons much better.
    As this is a gaming rig, Linux support is really not needed, so I fail to see the need to criticise that as well.

    The last thing you mentioned is valid, only a very specialized or rich crowed will go for a high end laptop for gaming, since it still compares miserably against a real desktop gaming rig. But he has stated so in the review as well.
  • ezinner - Saturday, June 5, 2010 - link

    I want quad core performance and a decent gpu with 1080p for under $1,500. At the $2,000 or higher mark, you in boutique territory like the Alienware.
    If I could afford this beast, I would be using it for audio and video production.
  • bennyg - Saturday, June 5, 2010 - link

    Too many gaming lappys are sold, proving it's a valid market segment.

    I don't need 4-5hr battery life. I have two regular workplaces + regular expo work where AC isn't an issue. I also like to not waste lunchtime at the pub next door where I get an hour of work done over some proper food. Then there's home (and sometimes work when really bored ;p) where I want a decent GPU but it doesn't need to be a 300W monster to provide decent graphics quality. I don't want the hassle of having to sync data across multiple PCs. My G51J suits me perfectly - jack of all trades.

    They do 1600x900 benchmarks because you can run a 1080p screen at 16x9 but not the other way round. If they had only benchies @ native res there'd be a hundred comments about the different res being a variable! However I still think it is a useful metric because most will be gaming @ native res where possible.

    One criticism I have is that this comparison is being done on hardware that's been around for months. Same with other notebook reviews (e.g. G51J). Plenty of other sites manage to get reviews like this done very early on - is the problem that Anandtech is too honest for the likes of most marketing departments ??

    (PS Anandtech - the "reply" button on the comments section isn't doing anything for me - it looks like it's just a hyperlink to the page title ending with a #)
  • JohnNyceis - Sunday, June 6, 2010 - link

    Hi, please let me know if the inline commenting is working better for you now.

Log in

Don't have an account? Sign up now