2 Cores Vs. 4 Cores

Discussion in 'Buying Tips and Advice' started by robeddie, Feb 18, 2016.

  1. robeddie macrumors 68000

    robeddie

    Joined:
    Jul 21, 2003
    Location:
    Atlanta
    #1
    Ok, I have what is perhaps a really dumb question. I have a 2011 iMac with a 4-core 2.7ghz i5 processor. The latest 13 inch macbook pro has a 2.7ghz i5 2-core processor. Despite the fact one has twice the cores as the other, their geekbench scores are almost identical.

    Obviously the newer iteration of the i5 is more efficient, etc. and... I know that a lot more than processor speed has to do with real world performance, but ... just isolating the processor, is there any reason to think one or the other has some kind of advantage over the other, given the fact that one has twice as many cores as they other ... despite the fact that their geekbench scores are identical?

    My guess, of course is that there is zero difference in how fast these systems will handle processor related tasks, but I thought it was worth asking the question.
     
  2. aidanpendragon macrumors 6502a

    Joined:
    Jul 26, 2005
    #2
    Is it a Geekbench single-core or multi-core score? I'd think a 4-core chip would handle multithreaded apps better, all else being equal (clock speed, i5)...even if the MBP's processor is a few generations "better."
     
  3. kohlson macrumors 6502a

    Joined:
    Apr 23, 2010
    #3
    This is why there are benchmarks -- cut through all the labels and hype to enable a more fair comparison of the performance of similar tasks on different systems.
     
  4. glenthompson macrumors 68000

    glenthompson

    Joined:
    Apr 27, 2011
    Location:
    Virginia
    #4
    It all depends on the software and tasks you're performing. Do your apple use multiple cores? If not then there is no benefit. I have an older Mac Pro that is faster in multi-core tasks than my newer MBP but slower in single core - older 8 core vs. faster 4 core. None of it really matters that much since little of my work is CPU bound. My SSD makes a much bigger difference in performance than the processor.
     
  5. jerwin macrumors 65816

    Joined:
    Jun 13, 2015
    #5
    Mac Book Pro Early 2105 13: 2.7 Ghz Core i5 5257U
    IMac Mid 2011: 2.7 Ghz Core i5 2400

    Perhaps you are underestimating the value of hyperthreading, which is used on mobile i5 chips. That plus the 15 percent improvement clock for clock from Sandy Bridge to Broadwell might well get you there.
     
  6. robeddie, Feb 19, 2016
    Last edited: Feb 19, 2016

    robeddie thread starter macrumors 68000

    robeddie

    Joined:
    Jul 21, 2003
    Location:
    Atlanta
    #6
    Ok, I didn't realize the i5 in the 2015 macbook pro 13 inch had hyperthreading.

    So then that raises my second question: Isn't it better to have 4 'real' cores instead of 2 and then 2 'simulated' cores (Hyperthreading)? Even if the geekbench scores between the two machines is equal, my sense is that the machines with the 4 'real' cores would have an advantage. I think I heard John Siracusa on the ATP podcast say that while hyperthreadin can 'trick' the system into thinking there's double the cores, that in some instances other bottlenecks actually limit the benefit.

    At any rate, replying to another glenthomson above, I realize SSD's and other improvements make a huge difference in real world performance. No debate there. I'm a HUGE fan of SSD's and put one in my 2011 iMac and all my macbook pros. I just wanted to narrow the discussion between the computers processor performance.
     
  7. maflynn Moderator

    maflynn

    Staff Member

    Joined:
    May 3, 2009
    Location:
    Boston
    #7
    Current geekbench benchmarks break out the numbers by single core and multi-core. The former will be identical if you're testing a similar cpu.

    I think benchmarks give you a ballpark measurement of how fast a computer may be, but I don't take them too seriously.
     
  8. TheArcher macrumors newbie

    Joined:
    Jan 16, 2014
    #8
    Yes, robeddie, this is a dumb question. You are trying to compare a desktop processor with a mobile processor. They are build on entirely different architectures separated by four years. It's like trying to compare apples to oranges. While apples and oranges are both fruit, sweet, have flesh, seeds, and skin, their makeup, i.e. DNA, are very different.

    So, here is what I 'think' you are trying to compare ... iMac i5 2.7 21.5 2011 / MacBook Pro i5 2.7 13 E

    You can make a side-by-side comparison here ... http://www.everymac.com/ultimate-mac-comparison-chart/?compare=all-intel-macs

    And, you can compare the processors side-by-side here ... http://ark.intel.com/compare/52211,84985

    That being said, the i5-5257U cpu in the MacBook Pro is much more efficient than the i5-2500S in the iMac even though it has two cores. The i5-5257U is doing much more with less, by far.
     
  9. Samuelsan2001 macrumors 603

    Joined:
    Oct 24, 2013
    #9

    There are a lot of differences between the systems, faster Ram, ultrafast SSD, better components on the logic board etc etc, all this adds up to a tiny laptop that will match your desktop simple as that.
     
  10. Efrem macrumors member

    Joined:
    Jul 30, 2009
    #10
    This is like comparing the running speeds of a horse and a mouse based on counting steps per minute. It ignores stride length, i.e., how much work the processor can accomplish per clock tick. More recent architectures get more done per clock tick for two reasons: (a) Increased transistor count allows for more parallel activity and thus reduces the number of clock ticks needed to carry out a task, and (b) Intel's designers learn from experience.

    Aside from that, the value of multiple cores/threads depends on your workload: (a) Is it processor-limited in the first place? If it is, (b1) Do you have enough things going on to take advantage of parallelism? Or (b2), are your main apps designed for multi-threading?
     
  11. jerwin macrumors 65816

    Joined:
    Jun 13, 2015
    #11
    geekbench test results:

    MacBook Pro 2015 Intel Core i5-5257U 64 bit Seems to be right around 3200 single, 7000 multicore
    iMac 27-inch 2011 Intel Core i5 64 bit Intel Core i5-2500S Seems to be right about 3025 single, 8000 multicore

    I haven't done any statistics-- just looked at a portion of the scores to weed out the people who conducted the tests under load, and since 64 bit code is ubiquitous and more efficient that 32 bit code, I focused on the 64 bit side of things-- thus throwing out all the freeloaders who didn't pay for a license.

    Basically, a 2.7 Ghz mobile Broadwell core is faster than a 2.7 Ghz desktop Sandy Bridge core. Hyperthreading makes up for a lot on the multicore side, but it's not enough.

    Anyway I'm done parsing Geekbench. It reminds me that for only 250 bucks more, I could have raised my macs score from 12,000 to 16,000.
     
  12. robeddie thread starter macrumors 68000

    robeddie

    Joined:
    Jul 21, 2003
    Location:
    Atlanta
    #12
    All good answers and info to my question. Thank you all. As far as comparing Apples to Oranges, it may be, but if I buy the macbook pro, I'll be wanting to do many of the same kinds of things on both. So in this case, comparing 'apples to oranges' is exactly what I want to do :)
     
  13. jerwin macrumors 65816

    Joined:
    Jun 13, 2015
    #13
    If you need more CPU power than your imac can provide, a macbook pro will be a poor substitute. If you're held back by weight, disk speed, or some other factor, a macbook pro may suffice. You could look into a "mobile i7", if cpu power is still a concern. If that's not good enough, the 15 inch macbook pros use the quad core i7, though at considerable extra expense.
     
  14. robeddie thread starter macrumors 68000

    robeddie

    Joined:
    Jul 21, 2003
    Location:
    Atlanta
    #14
    Ah no. (It seems some people want to change the narrow focus of what I'm asking here). To clarify: I'm fine with the cpu power of my iMac. And ... while the macbook pro gets the same geekbench score, it does that with only 2 cores and the help of hyperthreading.
    So I was wondering if geekbench was telling the whole story of processor performance or whether there might be a 'catch' so to speak. For instance, can I count on hyperthreading always giving me that ideal level of performance? Or will some tasks reveal 'bottlenecks' and limit hyperthreadings beneift, thereby making the macbook pro seem slower than the iMac in certain scenarios?
     
  15. jerwin, Feb 20, 2016
    Last edited: Feb 20, 2016

    jerwin macrumors 65816

    Joined:
    Jun 13, 2015
    #15
    http://arstechnica.com/features/2002/10/hyperthreading

    Hyper-threading's strength is that it allows the scheduling logic maximum flexibility to fill execution slots, thereby making more efficient use of available execution resources by keeping the execution core busier. If you compare the SMP diagram with the hyper-threading diagram, you can see that the same amount of work gets done in both systems, but the hyper-threaded system uses a fraction of the resources and has a fraction of the waste of the SMP system; note the scarcity of empty execution slots in the hyper-threaded machine versus the SMP machine.

    (left-- smp, right, hyperthreaded)

    [​IMG] [​IMG]


    An old school computer enthusiast might say that the hyperthreaded CPU has fewer wait states. I believe that a 2 core hypethreaded CPU can approach the performance of a 4 core SMP cpu, but unless you're concerned with heat, power consumption or cost, 4 real cores will beat 2 cores + 2 virtual cores (of the same design)​
     
  16. robeddie thread starter macrumors 68000

    robeddie

    Joined:
    Jul 21, 2003
    Location:
    Atlanta
    #16
    Ok. So that does answer my question. All other things being equal, even if the geekbench scores are the same, the system with 4 REAL cores (and no hyperthreading) is preferable to one with 2 cores + hyperthreading.
     
  17. Gav2k, Feb 21, 2016
    Last edited: Feb 21, 2016

    Gav2k macrumors G3

    Gav2k

    Joined:
    Jul 24, 2009
    #17
    It's not just hyper threading that's helping the results it's the architecture change.

    Let me be clear hyper threading cannot make up for a while physical core. If those chips where of the same architecture then there would be a comparible difference!
     
  18. jerwin macrumors 65816

    Joined:
    Jun 13, 2015
    #18
    Intel claims that it's good for a 30 percent increase in performance. It can also decrease performance in certain cases.
     
  19. robeddie thread starter macrumors 68000

    robeddie

    Joined:
    Jul 21, 2003
    Location:
    Atlanta
    #19
    Understood. But the architechture change only accounts for some of the reason the 2-core score is equal to the 4-core score, with the rest of the difference being made up for by the addition of hyperthreading. So if the poster directly above is correct that the hyperthreading can actually 'decrease' performance in certain cases, then indeed the 4-core (non hyperthreading) iMac is still better off than the 2-core (hyperthreading) 2015 13-inch macbook pro, despite their geekbench scores being virtually equal.
     
  20. jerwin macrumors 65816

    Joined:
    Jun 13, 2015
    #20
    Maybe you should consider the benchmarks broken down into individual programs, rather than looking at a single homogenized score:

    MacBook Pro (13-inch Retina Early 2015) vs iMac (27-inch Mid 2011)

    The Macbook Pro is disproportionately stronger on AES, Lua single core, N-Body Single Core, and has faster memory. The iMac is much faster on a great many multicore benchmarks including DGEMM Multicore, SGEMM Multicore, JPEG Decompress Multicore and so on.
    The difference on AES can probably be chalked up to new cryptographic optimizations in the haswell microarchitecture. Not sure about Lua single core, and N-body.

    Here's a comparison of a 2011 IMac with a 2015 Skylake iMac.


    Generally across the board improvements of 50 percent. Not all that compelling compared to the days when an upgrade could be expected to produce many times the performance of the old machine, but at least there's no tradeoff.

     
  21. Gav2k macrumors G3

    Gav2k

    Joined:
    Jul 24, 2009
    #21
    It really depends on the task thrown at the system. Everyday tasks will always see a benefit. It's server based workloads where hyper threading can be a hinderance.
     
  22. jerwin, Feb 22, 2016
    Last edited: Feb 22, 2016

    jerwin macrumors 65816

    Joined:
    Jun 13, 2015
    #22
    Suppose you have a single threaded application-- it's really critical that this single thread run as fast as possible. You also have a number of less critical background applications fetching mail performing system housekeeping, scanning for viruses, etc.

    With hyperthreading disabled, this primary application can monopolize one core. The other core, or the other three cores can be shared by the non critical applications.

    With hyperthreading enabled, there a possibility that this primary application won't be able monopolize a core-- instead, it will share some of the core's processing units with these other threads.

    The other issue is that a lot of this received wisdom has been generated by two groups. Gamers and High performance computing. Gamers value any excuse to save money by buying an i5 instead of i7, and plowing the savings into a video card. Mac users, by contrast, have very few options, so the choice to go with a i5 isn't as compelling-- Apple's high end video cards aren't. HPC users are typically balancing the computational advantages of running more threads and the computational costs associated with thread to thread communications...

    In any case, geekbench may place a premium on AES that you don't share. I suppose that whole disk encryption might be more practical and necessary with a macbook pro, but it's this subbenchmark that produces the illusion of parity.

    I think n-body relies on a "fused multiply add" instruction that may postdate sandy bridge. (see AVX2)
     
  23. Umbrarian macrumors member

    Joined:
    Oct 12, 2012
    Location:
    Texas
    #23

    Maybe maybe not.

    If your cpu is fully pegged HT is not going to do much, as it is near 100% efficiency
     
  24. jerwin macrumors 65816

    Joined:
    Jun 13, 2015
    #24
    Counting no-ops, yes. But hyperthreading turns those no-ops into useful work.
     
  25. kaiju_wars macrumors member

    kaiju_wars

    Joined:
    Feb 19, 2016
    Location:
    United States
    #25
    The quickest answer is... what are you doing with your computer?
    If your tasks involve a lot of CPU cores, then a quad-core will benefit you more. If you don't need to utilize more cores, a dual-core CPU will be fine.
     

Share This Page