Sign in with
Sign up | Sign in

Graphic Details

AVADirect's W860CU: Mobility Radeon HD 5870 Vs. GeForce GTX 285M
By

Though the Clevo W860CU is a nice desktop replacement notebook and AVADirect expertly configured our samples, the actual technology behind the graphics cards is completely out of the hands of those companies. Specifically, the graphics processors are not at all what buyers might expect them to be.

Desktop vs. Mobility Radeon Graphics
 Desktop Radeon
HD 5870
Desktop Radeon
HD 5770
Mobility Radeon
HD 5870
Transistors2.15 billion1.04 billion1.04 billion
Engine Clock850 MHz850 MHz700 MHz
Shader (ALUs)1,600800800
Texture Units804040
Z/Stencil Units1286464
Compute Performance2.72 TFLOPS1.36 TFLOPS1.12 TFLOPS
DRAM TypeGDDR5-4800GDDR5-4800GDDR5-4000
DRAM Interface256-bits128-bits128-bits
Memory Bandwidth153.6 GB/s76.8 GB/s64.0 GB/s
TDP188W108W50W


In response to last summer’s exposé
, Nvidia told us that its highest-model notebook GPU gets a similar name as its highest-model desktop GPU simply so people would understand that both cards represent the highest model of each market. Representatives of the firm vigorously denied that the naming similarity inferred a similar performance level, though we contend that this is the type of assumption Nvidia hopes buyers will make. In the case linked above, it’s easy to see that the GeForce GTX 285M bears no relation to the GeForce GTX 285, that these use completely different architectures, and that the GeForce GTX 285M architecture is taken directly from the old GeForce 8800 GTS 512MB GPU. Beyond basic underclocking, several updates used in the transformation from GeForce 8800 GTS 512MB to a GeForce GTX 285M make the newer version cheaper to produce and more energy efficient.

Nvidia stood alone in last summer in using an inferior core for its “premium” mobile offering, but it appears that AMD is learning its rival’s tricks. That is to say, while last summer’s Mobility Radeon products were simply underclocked variations of similarly-named desktop models, the company’s latest “premium” mobile model is nothing more than half of its desktop namesake.

Radeon HD 5870: CypressRadeon HD 5870: CypressMobility Radeon HD 5870: JuniperMobility Radeon HD 5870: Juniper

Recognizing these images from our previous Radeon HD 5770 launch article, some readers will undoubtedly think we got our pictures mixed up. The “high end” Mobility Radeon HD 5870 can’t be based on the mid-market desktop Radeon HD 5770, can it?

Desktop vs. Mobile GeForce Graphics
 Desktop GeForce
GTX 285
Desktop GeForce
8800 GTS 512MB
GeForce
GTX 285M
Transistors2.15 billion754 million754 million
Engine Clock648 MHz650 MHz576 MHz
Stream
Processors
240128128
Texture Units806464
ROP Units322424
Compute Performance1.06 TFLOPS624 GFLOPS576 GFLOPS
DRAM TypeGDDR3-2484GDDR3-1940GDDR3-2040
DRAM Interface512-bits256-bits256-bits
Memory Bandwidth159 GB/s64 GB/s65.3 GB/s
TDP183W135W75W


Our Radeon HD 5870, Radeon HD 5770, and Mobility Radeon HD 5870 statistics come directly from the AMD Web site. The Mobility Radeon HD 5870 is not just any Radeon HD 5770, however. Instead, it’s actually an underclocked Radeon HD 5770 and is likely programmed to further enhance its power-savings capabilities.

In other words, the Mobility Radeon HD 5870 is the Radeon HD 5770’s slower, less-energetic younger brother. Yet, shoppers who can track down the mobile module will find it costs three times as much as the desktop part, even though it does not include a cooling fan, heat sink, or display outputs.

The biggest question for us is, which graphics company deserves your ire the most? While we let the readers decide the fates of AMD and Nvidia, our only hope is that their clearly-deceptive naming practices don't adversely affect honest, hard-working builders like AVADirect.

Ask a Category Expert

Create a new thread in the Reviews comments forum about this subject

Example: Notebook, Android, SSD hard drive

Display all 54 comments.
This thread is closed for comments
  • 2 Hide
    DjEaZy , May 13, 2010 6:37 AM
    ... do they have AMD CPU's too?
  • 0 Hide
    anamaniac , May 13, 2010 6:42 AM
    How many partners use Clevo laptops and just rebrand them?
  • 1 Hide
    ta152h , May 13, 2010 6:43 AM
    I'm a little confused why you'd choose an i7 920 to compare with a different platform, but maybe it's because I don't know the mobile platform that well.

    But, if it's like the P55, which it seems to be, there's the added uncertainty of the architecture thrown in.

    Particularly with PCI-E being implemented differently, you might be seeing the inferior implementation of the P55 architecture responsible for a small amount of the relatively poor mobile performance. Since this implementation needs to multiplex the memory bus of the processor, you can run into situations where there is contention.

    I doubt it's significant, but I'm curious why you wouldn't want to make a comparison with a more similar desktop platform. Was it because you couldn't get an unlocked Lynnfield to get the clock speeds for the processors the same in Turbo mode?
  • 9 Hide
    Anonymous , May 13, 2010 6:48 AM
    This doesn't make much sense to me... if a 5870M chip = roughly a 5770 desktop chip and a 285M = roughly an 8800gts.. why is it not completely spanking it? we all know 5770>8800.. by a rather large margin! what could be the cause of this?
  • 2 Hide
    anamaniac , May 13, 2010 7:04 AM
    Tom's, you should show your power usage results to AMD and ask for an explanation, on why a lower rated part is using more power.
    Granted, with a 45W CPU and 50W GPU, 30 mins is expected on a 40W battery if fully stressed.
  • -7 Hide
    gti88 , May 13, 2010 7:37 AM
    It's not "mobile gaming" at all.
    So, is there any reason to own such notebuook?
  • 1 Hide
    jkeopka , May 13, 2010 7:41 AM
    I liked this article because I found it so darn relevant... I actually have this same Clevo Laptop, with the 5870 and 8 gigs of RAM.

    The GTX 285M was a $50 premium over the 5870, and I am glad I chose to stick to the 5870. It is kind of strange one would pay more to have less performance. I guess thats what fanboyism are all about?
  • 3 Hide
    jkeopka , May 13, 2010 7:41 AM
    anamaniacHow many partners use Clevo laptops and just rebrand them?

    Lots. Mine is a Sager 8690... which is a rebranded Clevo W860CU...

    I have seen this model at other sites as well.
  • -2 Hide
    falchard , May 13, 2010 7:44 AM
    Looks more like a bottleneck then anything conclusive. The results in nearly all the tests were close, yet 1 of them should have been clearly ahead.
    I think an ASUS JH73-A1 verse this would have been more interesting as its a bit cheaper for better parts.
  • 1 Hide
    Crashman , May 13, 2010 8:42 AM
    TA152HI'm a little confused why you'd choose an i7 920 to compare with a different platform
    Same speeds in Turbo mode, which is used during games, the primary focus being gaming performance.
    TA152HBut, if it's like the P55, which it seems to be, there's the added uncertainty of the architecture thrown in.
    That's true, but neither graphics solution provided the performance needed to highlight the mobile processor's on-die PCIe controller's performance advantage.
    TA152HI'm curious why you wouldn't want to make a comparison with a more similar desktop platform. Was it because you couldn't get an unlocked Lynnfield to get the clock speeds for the processors the same in Turbo mode?
    Exactly. Besides, Tom's Hardware has already seen that clock-for-clock, Lynfield games at least as well as Bloomfield when a single card is used. If nothing else, the comparison favors the mobile solution's lower power consumption.
  • 4 Hide
    Crashman , May 13, 2010 8:44 AM
    demomanisbestThis doesn't make much sense to me... if a 5870M chip = roughly a 5770 desktop chip and a 285M = roughly an 8800gts.. why is it not completely spanking it? we all know 5770>8800.. by a rather large margin! what could be the cause of this?

    Drivers maybe?
  • 0 Hide
    liquidsnake718 , May 13, 2010 8:45 AM
    anamaniacTom's, you should show your power usage results to AMD and ask for an explanation, on why a lower rated part is using more power.Granted, with a 45W CPU and 50W GPU, 30 mins is expected on a 40W battery if fully stressed.


    I also agree with this poster. It would be useful to learn the average battery time these units would work under various loads such as surfing, gaming, movies, and idle. I am also aware of the different settings but you can easily make a benchmark for this. This would be great to compare with the early core i7 laptops like the Eurocom and Cyberpower laptops that lasted only 45minutes to 1hour on average loads.

    Also heat is a factor for many buyers, especially buyers of a core i7 laptop. Sound is not much of a factor anymore with SSD's but one can still read and appreciate a chart for the db of the gpu and cpu fans but this is minor.

    Maybe you can create a core i7 gaming laptop comparison and compare gen1 vs gen 2 core i7 laptops as well which would probably please some serious buyers(not me) in the mobile gaming arena.
  • -2 Hide
    killerclick , May 13, 2010 9:21 AM
    "Gaming notebooks" are for suckers with too much money. Putting up with slower, ridiculously expensive hardware and smaller screens in order to be able to play in school or... where? Where do you "gaming notebook" guys play your games anyway?
  • 1 Hide
    dragonsqrrl , May 13, 2010 9:52 AM
    demomanisbestThis doesn't make much sense to me... if a 5870M chip = roughly a 5770 desktop chip and a 285M = roughly an 8800gts.. why is it not completely spanking it? we all know 5770>8800.. by a rather large margin! what could be the cause of this?

    Actually the HD5770 vs. 8800gts 512 isn't as much of a blow out as you might think. The HD5770 performs roughly on par with the HD4870, and on average these two cards perform just a few percentage points faster then the GTS 250, a card that also uses the g92 GPU and is basically an overclocked 8800gts 512. In fact the only real differences between the GTS 250 and the 8800gts 512 is the manufacturing process/die size and clock speeds, the GTS 250 is 55nm while the 8800gts 512 is 65nm. So in in this way the GTX 285M actually shares more similarities with the GTS 250 then the 8800, although the GTX 285M is severely under-clocked by comparison.

    When you think about it in this way, the fact that a significantly under-clocked GTS 250 is capable of performing similarly to a significantly under-clocked HD5770 really isn't much of a surprise. However the Mobility HD5870 does have one considerable advantage over GTX 285M, the inclusion of DX11 support. But considering the generational gap between these two cards I don't think this is a huge disadvantage for Nvidia.

    The bigger issue in my opinion is the performance of the Mobility HD5870 and its real world power consumption in comparison to its published TDP. ATI's new flagship mobile GPU is able to perform marginally better then the GTX 285M. This is against a mobile card that's well over a year old and soon to be replaced by the GTX 480M. I understand the rational behind ATI and Nvidia using their midrange GPU's in their high end mobile offerings (POWER CONSUMPTION), but I honestly think ATI should have bumped up the clock speed a bit more for the sake of next gen performance competitiveness. And I find it strange that two cards with significantly different TDP's would consume similar amounts of power... there's definitely something strange going on in that regard, especially considering the supposed 50W TDP of the Mobility HD5870.
  • 2 Hide
    masterjaw , May 13, 2010 9:57 AM
    ^ In other places aside from home, maybe. Like LAN parties, perhaps.

    Gaming notebooks are already gaining a quite niche of market. But still, as observed before, the performance gap is enormous compared to desktop PCs.

    I would be interesting if you compare these to ASUS gaming notebooks as well.
  • 3 Hide
    sakanagai , May 13, 2010 10:14 AM
    The fourth page has the tables in the wrong places. The Nvidia analysis follows the ATI chart, while the analysis on the ATI parts follows the Nvidia chart.
  • 4 Hide
    drowned , May 13, 2010 10:15 AM
    Max PC mag tested the 285M SLI version of this and found it gave about 1:10 battery life when playing a dvd with all power saving stuff enabled at 50% brightness. Honestly though, this isn't a notebook...it's a desktop that folds in half.
  • 1 Hide
    dragonsqrrl , May 13, 2010 10:32 AM
    Nice article. I wouldn't have a problem with using midrange desktop GPU's in high-end mobile cards if it weren't for these ridiculous prices. $400... are you kidding me? I understand that given the power consumption of current gen high-end desktop cards, there's simply no way ATI or Nvidia could transfer a Cypress or gf100 into a notebook and deliver reasonable performance at a reasonable TDP. It's more efficient at the moment to use mid-range GPU's instead. So I don't really think this trend was born out of an attempt to purposely deceive (ignorant?) customers, so much as it was born out of necessity.

    The pricing however is a completely different story, and I think a mid-range GPU should be priced like a mid-range GPU, no matter how it's re-branded. Charging the same price as a desktop HD5870 for a GPU with half the transistor count and less then half the performance is the true deception.
  • 3 Hide
    juliom , May 13, 2010 11:17 AM
    dragonsqrrlActually the HD5770 vs. 8800gts 512 isn't as much of a blow out as you might think. The HD5770 performs roughly on par with the HD4870, and on average these two cards perform just a few percentage points faster then the GTS 250, a card that also uses the g92 GPU and is basically an overclocked 8800gts 512. In fact the only real differences between the GTS 250 and the 8800gts 512 is the manufacturing process/die size and clock speeds, the GTS 250 is 55nm while the 8800gts 512 is 65nm. So in in this way the GTX 285M actually shares more similarities with the GTS 250 then the 8800, although the GTX 285M is severely under-clocked by comparison. When you think about it in this way, the fact that a significantly under-clocked GTS 250 is capable of performing similarly to a significantly under-clocked HD5770 really isn't much of a surprise. However the Mobility HD5870 does have one considerable advantage over GTX 285M, the inclusion of DX11 support. But considering the generational gap between these two cards I don't think this is a huge disadvantage for Nvidia.The bigger issue in my opinion is the performance of the Mobility HD5870 and its real world power consumption in comparison to its published TDP. ATI's new flagship mobile GPU is able to perform marginally better then the GTX 285M. This is against a mobile card that's well over a year old and soon to be replaced by the GTX 480M. I understand the rational behind ATI and Nvidia using their midrange GPU's in their high end mobile offerings (POWER CONSUMPTION), but I honestly think ATI should have bumped up the clock speed a bit more for the sake of next gen performance competitiveness. And I find it strange that two cards with significantly different TDP's would consume similar amounts of power... there's definitely something strange going on in that regard, especially considering the supposed 50W TDP of the Mobility HD5870.


    You're mistaken. The 4850 performs roughly on par with the GTS 250, not the 4870 which is way faster. The 4870 performs as a GTX 260 with 216 cores and the 4890 like a GTX 275.
  • 0 Hide
    Anonymous , May 13, 2010 1:21 PM
    It's not about whether AMD followed Nvidia's naming scheme, it's about taking the beating on sales. So AMD is supposed to keep their honest naming approach for the sake of us few enthusiasts, while Nvidia makes the bank on non-enthusiasts who don't check the facts and don't read the reviews? I fully support AMD in mimicking Nvidia's practices in this case, they were left no choice.
Display more comments