Barefeats benchmarks are in.

Discussion in 'iMac' started by StephenCampbell, Oct 3, 2013.

  1. StephenCampbell macrumors 65816

    Joined:
    Sep 21, 2009
    #1
    http://barefeats.com/haswel2.html

    So glad I got the i7.

    Do you suppose that the 2013 Fusion Drive would perform relatively to the 2013 Flash the way the 2012 Fusion performs relatively to the 2012 Flash? They don't have a 2013 Fusion in there.
     
  2. Nirurin macrumors member

    Joined:
    Oct 3, 2013
    #2
    http://barefeats.com/haswel1.html

    Thought i'd put up the link to the GPU tests here too.
     
  3. WilliamG macrumors 604

    Joined:
    Mar 29, 2008
    Location:
    Seattle
    #3
    Well, there you go. No point getting the high-end 2013 model if you can pick up a high-end 2012 for a good price.

    Still can't believe they only tested Diablo 3 as an ACTUAL game… lol!

    And yes, the new SSD speeds are utterly, utterly pointless. I honestly can't see a single real-world usage scenario for them unless you spend your days reading off MASSIVE files from your SSD, or spending your days copying MASSIVE files to RAID Thunderbolt SSD storage. And hey, if you spend your days doing that over wifi, then I guess wireless ac might be nice, but… really?

    This was a minor upgrade, at best. I expect more changes next year, Apple!

    And for people wondering why Apple didn't make a big deal of it at some keynote, at least you now know why!
     
  4. jayselle macrumors member

    jayselle

    Joined:
    Feb 12, 2006
    #4
    Umm, no. This will have a substantial impact. Applications will open quicker. Saves will happen faster. Photos will render quicker. Etc.
     
  5. Erphern macrumors 6502

    Joined:
    Mar 6, 2013
    Location:
    UK
    #5
    Gosh. If storage and network I/O doesn't impress you, I'm really not sure what you're looking for in a computer.
     
  6. Bear macrumors G3

    Joined:
    Jul 23, 2002
    Location:
    Sol III - Terra
    #6
    Yes, the 2013 Fusion Dirve performance will be improved over the 2012 iMac because of the SSD portion being PCIe based.

    As for the i7 - it all depends on what you use your computer for.
     
  7. WilliamG macrumors 604

    Joined:
    Mar 29, 2008
    Location:
    Seattle
    #7
    No, they won't. Over a "regular" SSD (I'm talking the 300MB-400MB/s), there will zero tangible benefit. Benchmarks, quite frankly, inflate real-world random I/O results astronomically.

    I leave it to Apple to release things. Then I decide if I'm impressed. :p
     
  8. bp1000 macrumors 65816

    Joined:
    Jul 7, 2011
    #8
    Good to see a third party review and benchmark with i7 and SSD.

    What we all expected really tho? i7's have always been 15-20% faster under benchmark conditions, because as we know, they are hyper-threaded.

    I think most people know if they need an i7, they usually are the type who do a lot of video editing or 3d rendering. i7 multi-core chips and this type of media editing go hand in hand.

    Only some apps make use of multi-core technology and only very few max out all 4 cores.


    Most apps are still single core, which sucks for me sometimes, i wish some programs i used had multi-core support.

    Even games that are multi-core supported dont saturate all 4 cores anyway so they would never benefit from i7. Only apps that saturate all 4 cores like serious data processing (video/3d) would benefit from the i7 core threading.
     
  9. Serban Suspended

    Joined:
    Jan 8, 2013
    #9
    So the i5 3.4 is better than last year i5 3.2, and the 780M is at least 10% improved over the 680MX, and PCIe is better as well and i guess the wifi SO overall the new iMac is worth it..but not really if you come from a late 2012 27" iMac
     
  10. AaronM5670 macrumors 6502

    AaronM5670

    Joined:
    Apr 19, 2012
    Location:
    London
    #10
    So glad I got the i7 and the GTX 775M- perfect for what I need - the differences seem to be marginal though.
     
  11. Serban Suspended

    Joined:
    Jan 8, 2013
    #11
    but still...i think the i5 3.4Ghz is the fastes quad core i5
     
  12. Bear macrumors G3

    Joined:
    Jul 23, 2002
    Location:
    Sol III - Terra
    #12
    That pretty much sums it up.

    Yes it is.
     
  13. WilliamG macrumors 604

    Joined:
    Mar 29, 2008
    Location:
    Seattle
    #13
    Absolutely. If you have no iMac at all, the 2013 is unquestionably better. It's just disappointing in that a year it's not really much of an improvement.
     
  14. Nismo73 macrumors 6502a

    Joined:
    Jan 4, 2013
    #14
    So now the best bang for your buck is the refurb $1999 i7 680, even when you take into account an education discount on late '13 iMacs. (it used to be $2199)
     
  15. bp1000 macrumors 65816

    Joined:
    Jul 7, 2011
    #15
    Don't forget these are benchmark tests maxing out all four cores. Very few programs and scenarios do this.

    Most programs are still single core and the single core tests are favourable.

    The SSD tests are more interesting. More of a measure of how fast the system will be.

    Another hidden gem is that Haswell whilst being faster still seems to have reduced the power consumption. With mavericks this might reduce still. It is currently clocking at 10% the watts the 27" imac consumed just a few years ago.

    Until more programs make use of multi core processing and multi threading, SSD and ram trump CPU power in most use cases apart from benchmarks, exception being something like video processing where you need all three.

    Keeping in mind most programs use just 1 core, having a faster single core computation and twice the SSD speed is a serious improvement.
     
  16. jayselle macrumors member

    jayselle

    Joined:
    Feb 12, 2006
    #16
    Clearly you've never experienced photo/video editing on a single SSD compared to two SSDs in raid 0. There is a very tangible difference.
     
  17. WilliamG, Oct 3, 2013
    Last edited: Oct 3, 2013

    WilliamG macrumors 604

    Joined:
    Mar 29, 2008
    Location:
    Seattle
    #17
    Actually, I spend my life photo editing RAWs from my 5D MKIII. :) And I went from my first SSD, an OCZ Vertex (1st generation) to a Samsung 830, and noticed no tangible difference in my workflow. The processor made the larger difference (going from my 2009 iMac i7 to the 2012 iMac i7).

    Going from a hard drive to ANY SSD is the best thing you can do. Once you hit SSD, whatever... I'd like to point out that I have RAID 0 SSDs in my i7 Sony Vaio Z2 (yes, 2x 64GB in RAID 0) from mid-2011, which hits almost 1GB/s read speeds, and it didn't improve my photo-editing speeds at any point, - not even when exporting RAWs etc etc. And who has room for their huge library of RAWs on small SSDs, anyway? :D

    Video-editing is absolutely another matter since it's quite disk intensive, but they have to be some pretty big files for you to notice much of a difference.

    I really don't see a big difference for 99.9% of what most people do. But we can agree to disagree. :)
     
  18. laserbeam273 macrumors 6502

    Joined:
    Sep 7, 2010
    Location:
    Australia
    #18
    The real value is in small random read/write speed (IOPS). The less time the CPU has to wait for the SSD, the better. It affects basically everything.

    Some things like loading games (GBs of random textures to load) will benefit immensely from it.

    But anyway, looks like single core speed isn't much quicker than my 2012 MBPr (GeekBench 3: 3000 vs 3960), so won't be jumping on this wagon.
     
  19. cirus macrumors 6502a

    Joined:
    Mar 15, 2011
    #19
    Applications won't open any quicker. Saves will not happen quicker. Photo rendering is pretty much CPU (and RAM) limited.

    SSD's massively show increases in benchmarks. Real world.....not so much.

    Article on difference between sata 2 and 3.
    http://www.tomshardware.com/reviews/ssd-upgrade-sata-3gbps,3469-13.html

    Benchmarks- Massive difference.

    [​IMG]

    Real World.

    Whooping 9% faster than sata 3 booting up, 22% faster shutting down.

    [​IMG]

    [​IMG]

    Booting Applications.

    [​IMG]

    Basically no difference between a good drive on sata 3 and sata 2. PCIe based drives won't be any faster.

    Loading a game for instance bottlenecks are elsewhere in the system.

    [​IMG]

    The one thing which will change is sequential transfers. But then you probably are not transferring GB files on a daily basis and the difference between 3 and 5 seconds for a transfer is minimal.

    [​IMG]

    And sequential tests generally tend to come down to controllers in real world situations.

    Good consistency and IOPS performance > sequential speeds.
     
  20. RCAFBrat macrumors 6502

    Joined:
    Jul 10, 2013
    Location:
    Montreal, QC
    #20
    I thought I had read that Mavericks will make use of available threads; will it allow programs that currently do not do multi core processing faster?
     
  21. bp1000 macrumors 65816

    Joined:
    Jul 7, 2011
    #21
    afaik programs have to specifically support multi core platforms. Their code has to enable the ability to use multiple cores. However, some simple dont need to use more than 1 core so people dont bother.

    I'm not sure about mavericks but i dont think an OS can manipulate the computation instructions of a third party program to make use of multi core processing, although i could be wrong!
     
  22. joema2 macrumors 65816

    joema2

    Joined:
    Sep 3, 2013
    #22
    Programs don't use cores, they spawn threads. The threads are then automatically dispatched to the available cores by the OS thread scheduler.

    If you use the Mac OS X Activity Monitor and look under the "threads" column, you'll see most processes have many threads. Each thread can run independently (and simultaneously) on an available core. A quick look at my wife's iMac shows Lightroom using 24 threads, iTunes 18 threads, and FireFox 38 threads.

    It is obvious a typical system has many hundred or thousands of threads. However most of these threads are not in a runnable state but are waiting on some event -- user input, system call return, signal from another thread, etc.

    If the activities of the threads within a process do not interact, a programmer can spawn these and harness multi-core hardware with little additional work. Imagine an office manager who with a pool of secretaries, who directs each one to write a letter to a client. Each one can work independently and get done faster, even if one finishes first.

    If threads interact or share data, the programmer must synchronize them to avoid overwrites and race conditions. This takes a little more work. Imagine an office manager with a pool of secretaries who directs each one write one paragraph of a letter to a client. At some point they must all synchronize and combine their individual work.

    A competent programmer will study their code and determine which execution paths are CPU intensive. He will then spawn multiple threads to handle these, assuming the availability of multi-core hardware. This does take a little additional work, to synchronize the results of the threads.

    Since multi-core hardware has been available for *many* years, a decent developer will write their code accordingly. Unfortunately there are many developers not in this category, in spite of aids such as Apple's Grand Central Dispatch to assist them: http://en.wikipedia.org/wiki/Grand_Central_Dispatch

    Even if an individual process is not properly multi-threaded, there are many processes on a typical system. Within each process there are many threads. On a quad core system, all it takes is four runnable threads across all these processes to harness the available CPU capacity. So even without well-written multi-threaded code, a multi-core or hyperthreaded CPU can be beneficial.
     
  23. Michaelgtrusa macrumors 604

    Michaelgtrusa

    Joined:
    Oct 13, 2008
    Location:
    Everywhere And Nowhere

Share This Page