Iris vs. HD 5000

Discussion in 'MacBook Pro' started by ShaneBunting, Oct 22, 2013.

  1. ShaneBunting Suspended

    Joined:
    Aug 10, 2009
    Location:
    United Kingdom
    #1
    Hello,

    INTEL IRIS (as found on new 13" Pro's)

    vs.

    INTEL HD 5000 (as found in current Air's)


    Which is better, and why?

    I'm counting on graphics geeks to give me an insight to some of the technical specifications behind them and for other potential buyers of the new Pro to comment too.

    Thanks! :)
     
  2. PDFierro macrumors 68040

    Joined:
    Sep 8, 2009
    #2
    Iris is better. It's 5100 compared to the 5000 in the Air.
     
  3. leman macrumors 604

    Joined:
    Oct 14, 2008
    #3
    The difference is clock speeds. All this information is readily available on Wikipedia btw.
     
  4. ShaneBunting thread starter Suspended

    Joined:
    Aug 10, 2009
    Location:
    United Kingdom
    #4
    Thanks

    And I know, but I don't understand the technical terms with graphics. The people on here are good at explaining it in a lot of threads!
     
  5. PDFierro macrumors 68040

    Joined:
    Sep 8, 2009
    #5
  6. astalavistathes macrumors member

    Joined:
    Oct 15, 2013
    #6
    The article compares the i7-4558U. My question is: 1 of the 3 available processors for the 13" model are i5-4288U and the other is i5-4258U and they both have the Iris 5100. Do they have the same performance in terms of graphics with the i7-4558U or less? In other words, are the stats shown in the article the same for all these 3 variants?
     
  7. ND40oz macrumors newbie

    Joined:
    Aug 13, 2011
    #7
    The i5-4258U has a max GPU speed of 1.1 GHz, the i5-4288U has a max GPU speed of 1.2 GHz, so it should be a bit faster in terms of GPU performance. The i7-4558U has the same GPU speeds as the i5-4288U, so don't jump to all the way to the i7 if you're just looking for GPU performance. You'd only move to the i7 if you want the 4MB of cache or the slightly faster clock speeds.

    Comparison: http://ark.intel.com/compare/75990,75991,75992
     
  8. astalavistathes macrumors member

    Joined:
    Oct 15, 2013
    #8
    Fanastic! Thanks!!
     
  9. dusk007 macrumors 68040

    dusk007

    Joined:
    Dec 5, 2009
    #9
    The main difference is TDP.
    The HD 5000 shows up in 15W chips.
    The HD 5100 aka Iris shows up in 28W chips.

    The GPUs both are equally big with the same number of EUs and they can clock quite high if necessary (1.1Ghz vs 1.2Ghz isn't much).
    At 1000Mhz the 40EUs suck about 22.5W (notebookcheck showed that running furmark). So while the 15W chip can in theory clock to 1100Mhz it will almost never manage. Intel allows their chips to go above TDP if they are cool enough for short burst but if you play a game it will settle at much lower performance.
    Usually around 900Mhz for Iris (according to notebookcheck). Games aren't quite as power consuming as furmark.
    Hd 5000 with only 15W will settle at much lower clocks.
    The 28W just means that the GPU can clock higher and also that the CPU doesn't have to give up so much speed. It is 28W total. What ever the GPU needs the CPU cannot have. So they have to fight for TDP. You can imagine that with 15W the GPU cannot run very high clocks if it wants to still provide enough CPU speed to handle everything else.
     
  10. Breedlove macrumors regular

    Joined:
    Oct 23, 2006
    #10
    Iris 5100 vs. HD 5000

    Now that the new rMBPs have been released, does anyone have any benchmarks comparing the two?

    I've already read all of the conjectures made in this forum as to the performance differences and as helpful they might have been two weeks ago, they don't compare to benchmark results and in actual performance tests.

    Any help?
     
  11. Raibyn macrumors regular

    Joined:
    Sep 24, 2013
    Location:
    South Carolina
    #12
    Here are some results comparing the Intel 5000 in an 11" MacBook Air with two new rMBP 13 inchers with the Intel Iris 5100 graphics chip.
     

    Attached Files:

  12. Atomic Walrus macrumors 6502a

    Joined:
    Sep 24, 2012
    #13
    This is going to be a unique situation because a benchmark like Unigine Heaven has no CPU component beyond handling the rendering prep work. A proper game would have physics, AI, networking, ballistics, and so on which is where the TDP limitation would come in.

    Still, the Air continues to impress me. On single-purpose tests (CPU or GPU) the 13" rMBP really doesn't seem to have much, if any edge over it.
     
  13. Wuiffi macrumors 6502a

    Wuiffi

    Joined:
    Oct 6, 2011
    #14
    if you compare the MB Airs Intel 5000 with the Macbook Pro retina there is one more thing you have to remember: the 2.6 und 2.8 processors have a slightly higher clocked Iris chip (100 MHz). It's not much but should make around 9% difference.
     
  14. leman macrumors 604

    Joined:
    Oct 14, 2008
    #15
    Now please do the same but without AA and in fullscreen mode. Your test is severely bandwidth limited (which is the weak point of both models). Plus, windowed mode introduces additional overhead.
     
  15. Raibyn macrumors regular

    Joined:
    Sep 24, 2013
    Location:
    South Carolina
    #16
    I will post those results when I can. I have run several benchmarks with different settings and the rMBP 13 really doesn't perform that much better than the MBA in any of the tests I have run.
     
  16. Atomic Walrus macrumors 6502a

    Joined:
    Sep 24, 2012
    #17
    When you're going to actually see a difference is actually playing games. Not only will the CPU need to do some actual work (more or less depending on the game), but it will throttle over time to manage the temperatures. Benchmarks usually don't run long enough to show the throttling effect.
     
  17. Eidorian macrumors Penryn

    Eidorian

    Joined:
    Mar 23, 2005
    Location:
    Indianapolis
    #18
    I don't expect there to be much of a difference without the eDRAM. The 5100 is just higher clocked and without as much of a thermal cap.
     
  18. Raibyn macrumors regular

    Joined:
    Sep 24, 2013
    Location:
    South Carolina
    #19
    Unigine Heaven runs for quite a while, but I see your point. I will also do some tests with a game or two.
     
  19. Atomic Walrus macrumors 6502a

    Joined:
    Sep 24, 2012
    #20
    Well it runs for something like 5 minutes, that's not long enough to see serious thermal throttling. What you could do is let it run on demo mode for like 20 minutes before you start the benchmark on each machine, that would more or less simulate 30 minutes of a game.

    Actually will be interesting to see the results. If the cooling is better than what's expected for that TDP the chip will have no problem drawing significantly more than 15W of power.
     
  20. rMBP2013 macrumors member

    Joined:
    Oct 22, 2013
    Location:
    Sydney
    #21
    Thanks for posting this.

    I'm surprised at the HUGE difference between the 2.4 GHZ and 2.6 GHZ i5 13" retinas.

    Why the significant variance between max frame rate?
     
  21. Atomic Walrus macrumors 6502a

    Joined:
    Sep 24, 2012
    #22
    To me that looks like a fluke, something you couldn't repeat in other tests. The actual average FPS difference between the two was so small you'd likely never notice it, and in general only average and minimum frame rates matter.

    If the OP wants to repeat the test to see if that discrepancy comes up again it would be worth checking.
     
  22. AlecMyrddyn macrumors 6502

    Joined:
    Dec 5, 2008
    Location:
    Southern Maine
    #23
    A guess?

    The 2.6 i5 (and 2.8 i7) both run with a maximum GPU clock rate of 1200 MHz, the 2.4 is limited to 1100 MHz.

    The MBA's 1.7 i7 is also 1100, while the 1.3 i5 is 1000 MHz.
     
  23. Raibyn, Oct 30, 2013
    Last edited: Oct 30, 2013

    Raibyn macrumors regular

    Joined:
    Sep 24, 2013
    Location:
    South Carolina
    #24
    Here are the NovaBench scores:
     

    Attached Files:

  24. OverpricedJunk macrumors newbie

    Joined:
    Oct 15, 2013
    #25
    Uhh... is this showing the 5000 as having the best graphic capabilities?
    This is incorrect, right?
     

Share This Page