Compute & Video Transcoding Performance

AMD's hope with Fusion is that the GPU transistors will regularly be used for more than just 3D gaming. Today we're still hunting for good uses for GPU compute, but we do have a few benchmarks that we can use to illustrate Llano's GPU compute prowess.

Our compute benchmark comes from Civilization V, which uses DirectCompute to decompress textures on the fly. Civ V includes a sub-benchmark that exclusively tests the speed of their texture decompression algorithm by repeatedly decompressing the textures required for one of the game’s leader scenes. Note that this is a DX11 benchmark and obviously can't run on Intel's Sandy Bridge GPU.

Compute: Civilization V

The Radeon HD 6550D performs as expected - it offers performance close to the 5570 and above that of the 6450. Memory bandwidth constraints are less of an issue here apparently as the 6550D delivers 89% of the performance of the 5570.

One area where we can compare the Radeon HD 6550D and Intel's HD Graphics 3000 is in video transcoding performance. For that we turn to ArcSoft's MediaConverter 7 and measure performance transcoding a 15Mbps H.264 encoded 1080p Quantum of Solace rip to a 4Mbps 720p iPhone 4 compatible file:

Video Transcoding Comparison
  AMD A8-3850 Intel Core i3-2105
Frames Transcoded per Second 63.7 fps 165.2 fps

While the A8-3850 can exploit its GPU cores for transcoding, Intel's Quick Sync continues to be the faster method of transcoding video as we found in our original Sandy Bridge review.

Performance in Older Games Power Consumption & Final Words
POST A COMMENT

99 Comments

View All Comments

  • L. - Thursday, June 30, 2011 - link

    Don't ask too much of a first rushed-in prototype ;)

    Let's see what it does with a decent memory system, dozer modules and a few other updates, I think we might play Metro 2033 on those ;)
    Reply
  • zondas30 - Thursday, June 30, 2011 - link

    Hello everyone, i just registered but this site is realy ny favorite for review and i am browsing it alot.
    now on the subject, what i realy think is that these apu's might not be very dominating in theyr performance but they are realy good for gaming and that asymmetric CrossFire thingy is realy one hell of a thing, for people like me that dont have alot of $ to spend on new pc and loves gaming it should be wery good selection, only thing that interests me is that if i could do that asymmetric CrossFire with my old ati radeon hd 3870, i still love it and it is still powerfull for me on 1440x900 resolution. so as i was saying, it might not be dominating in its porformance with programs but it is fantastic thing for gaming.
    Reply
  • ET - Thursday, June 30, 2011 - link

    I'm interested in how desktop Llano compares in performance to laptop Llano, but found it impossible to compare. The difference in resolutions is excusable (though as has been commented, the resolutions in this article are very old and should be replaced), but then you have different sets of games, and when there's an overlap they are ordered differently, and are run with different options, be it DX9 vs. DX10 or low vs. medium, ... It's a real pain. Reply
  • Arnulf - Thursday, June 30, 2011 - link

    First, thanks for your review, I have been eagerly looking forward to it for months now !

    A thought has occured to me when I got to Conclusion section of your article: how does A8-3850 + Asymmetric Crossfire fare compared to similarly priced Athlon/Phenom II + single discrete GPU - whatever RAM APU uses up ?

    Say A8-3850 = $135, HD6670 (the strongest card for Asymmetric Crossfire) = $84 on Newegg, how do these compare to identically priced Phenom II 955BE (= $115) + HD6750 (= $105), assuming one can get FM1 at the same price as AM3 with identical features ? In other words: is Asymmetric Crossfire worth the hassle of upgrading your platform, assuming ideal circumstances (they are going to fix DX9 at some point, I'm sure) ?
    Reply
  • Seikent - Thursday, June 30, 2011 - link

    As you can see in the review, at least for now, the asymmetric crossfire doesn't scale pretty well. That's the cost of being asymmetric. And it doesn't work on DX9.

    So you should be way much better with the 955 + hd6750 combo on the cpu and gpu side.

    In general, Llano doesn't make much sense with a discrete video card because you're already paying for 'integrated' graphics.
    Reply
  • silverblue - Thursday, June 30, 2011 - link

    What's wrong with the scores for the A8 at 1024x768? Reply
  • duploxxx - Thursday, June 30, 2011 - link

    Great review, solid baseline for initial review, something we can always expect from Anand. Few of your review members dough miss quite a bit training to get to a certain level..... you are to kind to them when you review it before posting :)

    Hope to see a decent review on OC and potential of this chip also combined OC + CF

    one remark, since when does a general users need cpu horse power like SNB 2100, they don't, they would be more then fine with Pentium alike SNB and for the record while benchmarks might show quite a bit of performance difference in single threaded benchmarking apps, common use for browsing, office etc it has 0 added value with this kind of CPU performance.... in atom alike platforms it is more then noticable. Put any unknown user behind either platform compared here for general tasks and they will be fine with both, once you startup any game worth mentioning (even SIMS) they will notice the difference what to choose.
    Reply
  • seapeople - Saturday, July 02, 2011 - link

    But in "common use for browsing, office, etc" then quad core multithreaded performance is almost completely useless. I don't know why people keep bringing up common use office applications as a win for Llano. Reply
  • cacca - Thursday, June 30, 2011 - link

    Please do it.
    Put a big NO on direct x10/11 for Intel and let us know the real performance of llano.

    Is starting to be a joke this website

    "... if gaming is going to be the most intensive thing you do on your notebook... "

    I' would like to know if you think that game are NOT the most stressing application for a desktop used by general public, leaving outside workstation for rendering and video encoding.

    In thi review you never put under stress the 2 different solutions, so we could not see the real shorcomings of the 2 platforms.

    no AA is a joke.
    Reply
  • L. - Thursday, June 30, 2011 - link

    This.
    +1
    rep.

    Whatever floats your boat, but AT needs to make sponsor money too you know, it's not like they could say "i3-2100 is useless don't buy it".

    Besides, the only way to compare two cpu's for gaming today is the following : which one enables the fastest GPU operation and general game score for a given budget (i.e. core i3-2100 + hd 6950 > core i5 + hd6750).

    As you mentioned, video encoding and rendering should NOT be a relevant part of a review, most people don't do it at all (yes, I download x264, but there's like 15 different people tops encoding these and it serves millions of users).

    And rendering... why not complex HPC-type stuff ? Database benchmarks (even that is MORE relevant than rendering) ? SuperPI ... now that's relevant for a user :) - at least if he participates in LN2 OC Competitions, like everyone and their dog.
    Reply

Log in

Don't have an account? Sign up now