Trinity vs Ivy Bridge

Discussion in 'Architecture and Products' started by rpg.314, Jun 29, 2011.

  1. gongo

    Regular

    Joined:
    Jan 26, 2008
    Messages:
    582
    Likes Received:
    12
    updated my kid sister hd4000 build...well Intel metro-fied the GUI aaand it looks ugly..the icons aspect ratio seems squashed...feels good otherwise...gonna have a uber gaming experience with facebook games. :oops:
     
  2. RedVi

    Regular

    Joined:
    Sep 12, 2010
    Messages:
    387
    Likes Received:
    39
    Location:
    Australia
    Of course, but it's the only way I can see of AMD making that big HSA splash you desired rather than a trickle (Kabini first, then Kaveri).

     
  3. hkultala

    Regular

    Joined:
    May 22, 2002
    Messages:
    284
    Likes Received:
    6
    Location:
    Herwood, Tampere, Finland
    It seems we see it differently.

    The biggest "incoming thing in near furture with HSA integration" is unified address space between GPU and CPU side.

    GCN supports unified address space. Kabini will have GCN-based GPU.
     
  4. Paran

    Regular Newcomer

    Joined:
    Sep 15, 2011
    Messages:
    251
    Likes Received:
    14

    http://semiaccurate.com/2013/04/15/putting-intels-new-hd-4000-drivers-to-the-test/
    http://semiaccurate.com/2013/04/29/a-look-at-intels-opencl-performance/

    Looks promising. Their new OpenCL 1.2 driver seems to be pretty good. Despite the much higher theoretically computing power from AMD, Trinity isn't faster in a couple of OpenCL applications. In notebooks HD4000 should be superior in OpenCL than Trinity (assuming that Trinity mobile is much slower than A10-5800K).
     
  5. Andrew Lauritzen

    Moderator Veteran

    Joined:
    May 21, 2004
    Messages:
    2,526
    Likes Received:
    454
    Location:
    British Columbia, Canada
    The "flops" numbers in the CL benchmark don't make sense to me...
     
  6. Paran

    Regular Newcomer

    Joined:
    Sep 15, 2011
    Messages:
    251
    Likes Received:
    14
    Lol the tester don't like to see HD4000 being faster in some benchmarks. @forums

     
  7. nAo

    nAo Nutella Nutellae
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    4,325
    Likes Received:
    93
    Location:
    San Francisco
    First DX11/OpenCL1.2 part from company A outperforming the Nth DX11/OpenCL1.2 generation part with more than twice the flops from company B.. I guess it doesn't compute for some (pun intended).
     
  8. swaaye

    swaaye Entirely Suboptimal
    Legend

    Joined:
    Mar 15, 2003
    Messages:
    8,457
    Likes Received:
    580
    Location:
    WI, USA
  9. lanek

    Veteran

    Joined:
    Mar 7, 2012
    Messages:
    2,469
    Likes Received:
    315
    Location:
    Switzerland
    RatGPu ? who have been paid for call a benchmark with this name ? seriously .... the HD4000 is asbolutely unable to beat any AMD APU GPU, but you have find THE benchmark where it can beat the AMD APU .. well thats promise a lot for the next intel GPU... I think all the industry, reviewers should stop with all their Bullshit benchmarks and start to use only one benchmark: The RATGPU benchmark .
     
    #810 lanek, May 22, 2013
    Last edited by a moderator: May 22, 2013
  10. Paran

    Regular Newcomer

    Joined:
    Sep 15, 2011
    Messages:
    251
    Likes Received:
    14

    Nonsense. OpenCL is superior on HD4000 and gaming performance too if we compare mobile derivate. Every Trinity below A10 is much slower.
     
  11. jimbo75

    Veteran

    Joined:
    Jan 17, 2010
    Messages:
    1,211
    Likes Received:
    0
    You're getting confused between a superior GPU and a superior CPU. Any win Intel has is based on their superior CPU performance at low resolutions. Raise the graphics load high enough and HD 4000 will lose, every single time.
     
  12. mczak

    Veteran

    Joined:
    Oct 24, 2002
    Messages:
    3,015
    Likes Received:
    112
    HD4000 beating Trinity/Richland APU in some OpenCL tests isn't all that surprising really. It is no big secret that their vliw4 architecture wasn't all that well suited for compute (that is, performs very well for some things, but pretty crappy for others), and amd moving to gcn at least was partially motivated by that (heck there are benchmarks floating around showing kabini beating Llano in a couple OpenCL tests, granted that one is even VLIW5 but that is usually performing similarly to VLIW4 in these tests).
    HD4000 doesn't suffer from that, and that it can use cpu l3 cache too might help quite a bit in some benchmarks too.
     
  13. Paran

    Regular Newcomer

    Joined:
    Sep 15, 2011
    Messages:
    251
    Likes Received:
    14
    No we are talking about the GPU. Try Luxmark GPU, you see AMD losing there. Try any decent HD4000 and compare it with a mobile A4-A8 APU. It will lose in many games. He says HD4000 is unable to beat any AMD APU which is big nonsense. Same for ULV comparisons where AMD never had an advantage.

    http://www.notebookcheck.com/fileadmin/Notebooks/Sonstiges/Games/Metro_Last_Light/tabelle.jpg

    There you have Full HD. Top notch Trinity is unable to beat a decent HD4000. A4-A8 isn't faster than HD4000 in the notebook segment. Most people are not aware of this.
     
  14. jimbo75

    Veteran

    Joined:
    Jan 17, 2010
    Messages:
    1,211
    Likes Received:
    0
    So explain to me why the HD 4000 in the i5 3360M loses heavily to the HD 4000 i7 3610QM. If it's just down the the graphics, why is the faster (1.2 GHz) i5 3350M losing so badly (by 60%) to the slower 1.1 GHz 3510QM? You've just proved my point, it is purely down to massive extra cpu performance.
     
  15. Blazkowicz

    Legend Veteran

    Joined:
    Dec 24, 2004
    Messages:
    5,607
    Likes Received:
    256
    But that i3 can do massively faster framerates when coupled to an external GPU.

    Maybe the i3 laptop has ddr3 1333 memory so that the OEM can save $1 on each laptop sold. This cripples your 3D performance, coupled with less L3 if that plays a role for graphics.

    Likewise what memory is that A10-4600M using, ddr3 1600 or 1866?
    With both AMD and Intel selling APUs, you now have to pay attention to your main memory bandwith.
     
  16. Andrew Lauritzen

    Moderator Veteran

    Joined:
    May 21, 2004
    Messages:
    2,526
    Likes Received:
    454
    Location:
    British Columbia, Canada
    I'm sure it's partially CPU, but it also has a nearly 50% higher TDP. The other important aspects of the system aren't even listed... i5's have less last level cache (GPU shares this) and sometimes less external memory bandwidth as well I believe.

    In any case, comparing 100W A10's to tiny Ivy Bridge GPUs and drawing architecture conclusions is a bit silly. Ivy GPU really only becomes TDP limited around 17W or so, so that's where you can make interesting architectural comparisons. With Haswell, there are bigger SKUs so it can be compared a bit further up the scale.
     
  17. jimbo75

    Veteran

    Joined:
    Jan 17, 2010
    Messages:
    1,211
    Likes Received:
    0
    The i5 3360M is a 35W chip while the i7 3610QM is 45W.

    I'm sure Metro is an outlier because it's extremely well threaded and we're seeing a huge difference between 8 and 4 threads, but that also means the graphics on the 35W A10 4600M are being held back by the weak cpu.
     
  18. jimbo75

    Veteran

    Joined:
    Jan 17, 2010
    Messages:
    1,211
    Likes Received:
    0
    Look at the power consumption of the HD 4000 when gaming btw -

    http://www.tomshardware.com/reviews/kabini-a4-5000-review,3518-13.html

    [​IMG]

    This is clearly blowing well past 17W during intense graphics load and it's not being reported in most of the tech press. I really hope Anand has the balls to show us how efficient Haswell's iGPU is while he's telling us how fast it is.
     
  19. Alexko

    Veteran Subscriber

    Joined:
    Aug 31, 2009
    Messages:
    4,496
    Likes Received:
    910
    I wonder what's going on, here. Ivy is rated at 17W, Kabini at 15W, so that's a 2W difference. Kabini also includes the southbridge, so let's say 4W.

    But there's a 14W difference here. It could be the Intel laptop's fault (power hungry display/motherboard/HDD/something) or perhaps Kabini's turbo isn't very aggressive.

    Edit: can't be the display, they turned it off and used an external one.
     
Loading...

Share This Page

  • About Us

    Beyond3D has been around for over a decade and prides itself on being the best place on the web for in-depth, technically-driven discussion and analysis of 3D graphics hardware. If you love pixels and transistors, you've come to the right place!

    Beyond3D is proudly published by GPU Tools Ltd.
Loading...