computing speeds?

Discussion in 'Mac Pro' started by LCS/drummer, May 23, 2009.

  1. LCS/drummer macrumors newbie

    Joined:
    Jul 28, 2008
    #1
    Im looking to buy a few macs. a desktop, and a MacBook Pro(hopefully 17")

    I really want a mac pro, and i was planning on getting the 2.93 dual Quads. with 16GB/Ram

    this is probably confusing. especially if your not a computer nerd :p but i tried to convey my question best i could.

    Well i was reading the May issue of MacWorld, i was reading their review of the new Mac Pro, and the review disappointed me. heres why:

    For starters, they compared a 2.66 quad, vs the 2.26 octo setup.(the two bare models) why they didnt compare the best set ups (2.93 quad vs. 2.93 octo) is beyond me.

    They ran some everyday tasks on both machines. they said with photoshop, iTunes, Imovie, compressor, and safari(?) the quad core machine was faster in using these normal apps based on the fact that:

    1. most applications have enough trouble using 4 cores efficiently, let alone 8.
    2. the cores in the octo set up are running slower than the four core setup (duh)

    i would say that if they tested the 2.93 Quad Mac pro vs. the 2.93 Octo Mac Pro, their test results would be a tad different, because their second reason(cores running slower) is no longer applicable based on the fact all the processors have the same clock speeds.

    not to mention, Snow Leopard show make a big step in multi-core efficiency across applications. right?

    ----My Question to all you Mac Pro users is this:

    What would you say about this? anyone here that uses a 2.93 octo setup see slow speeds in "everyday" apps?

    or did MacWorld Magazine do a half-assed attempt to review these machines?

    thanks for the help!
     
  2. 300D macrumors 65816

    300D

    Joined:
    May 2, 2009
    Location:
    Tulsa
    #2
    Probably because they weren't shipping yet at the time or few people will actually buy the top-end models.

    More cores = more system overhead so the quad will still be slightly faster in non-multithreaded apps. The difference would be in fractions of a second though.

    No. Programmers still have to write their software to use it.

    :D Nothing about a 2.93 mac pro (quad or octo) would be considered slow! The hard drive is still the biggest bottleneck in the the entire computer (except the optical drive, but who uses them anymore?). Using an SSD would speed things up even more.

    Yes. They got the first machines they could get their hands on. Since the low-spec models always ship first, thats what they tested.
     
  3. Tesselator macrumors 601

    Tesselator

    Joined:
    Jan 9, 2008
    Location:
    Japan
    #3

    Yeah, LOL! MacWorld.... Hehhehehe... I think MW "technical reviews" can safely and wholly be ignored! I trust the 16-year-old down the street more. :)

    I don't believe SL with allow current applications to make better use of multi-cores. It's illogical to assume that it would make a significant difference. We're still dealing with the same basic variables with only a few minor complications added in. The variables are:

    Bus speed,
    Bus throughput,
    Clock Speed,
    Number of Cores,
    Size of L1 ~ L3 Cache.

    The few minor complications which are adding confusion are:

    Virtual Cores,
    Turbo Boost.

    Understand all seven of those issues (variables) should afford you the ability to make decisions that will trump any rubbish published by the likes of MacWorld - just based on the published numbers from Intel. Understanding these are fairly simple and a good start are the Wikipedia pages on the topic. Understanding virtual core and number of cores can be a little tricky because it requires knowledge of which applications are multi-threaded and/or what parts of them are multi-threaded - and additionally how many applications and BG tasks you will be using at the same time (not loaded, but actually using). This is perhaps also relevant when considering how turbo-boost figures into it.
     
  4. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #4
    To me, this is the hard part. ;) Lots of research required. :D
     
  5. Cynicalone macrumors 68040

    Cynicalone

    Joined:
    Jul 9, 2008
    Location:
    Okie land
    #5
    I use the 2.93 Octo everyday I would hardly call it slow. When purchasing the machine I was thinking long term. I might not be able to use all 8 (or 16 virtual) cores now but what about in a year or two?
     
  6. Tesselator macrumors 601

    Tesselator

    Joined:
    Jan 9, 2008
    Location:
    Japan
    #6
    Yup! :)


    It won't change much if at all. The apps that can be multi-threaded already are. Not all.. not even most.. applications can be tho. This seems to be missed by a majority of forum users.

    To know if the app you want to use is MT or not go and download the demo, install it, and use it like you would normally. Monitor the CPU activity for each and that will tell you.

    In my case I use lots of rendering engines and they all use MT to the fullest extent. LightWave with or without external engines profiles the best and I really enjoy using it! The native renderer scales at between 99.8 and 100% across all physical cores so a render that used to take 8 min. per frame now takes exactly 1 min. on my 8-core. The application itself (editing and set-up) doesn't run any faster on 8 than it does on 4 - or at least not noticeably. Anyway 8 to 1 increase is significant for me. With many frames to render it's often the difference between 1 day and 8 days... or in the extreme the difference between 1 week and 8 weeks. :)
     

Share This Page