Mac Pro with SSD benchamrks

Discussion in 'Mac Pro' started by snapdragonx, Jul 14, 2013.

  1. snapdragonx macrumors regular

    Joined:
    Oct 31, 2012
    #1
    I've been looking at a lot of benchmarks comparing the new iMac (i7, SSD) with the Mac Pro (6-core).

    The iMac seems to perform slightly better in most tasks than the iMac. But the Mac Pro has a much slower mechanical hard drive.

    So, does anyone have a Mac Pro with a decent SSD that could post some Geekbench or other benchmarks? I'd love to see how the Mac Pro does with a decent SSD when compared to the iMac.

    I ask because I'm looking at returning my brand new iMac and getting a Mac Pro, but I don't want to "downgrade" in terms of performance.

    Thanks in advance :)
     
  2. wonderspark macrumors 68030

    wonderspark

    Joined:
    Feb 4, 2010
    Location:
    Oregon
    #2
    I've run Geekbench on both HDD and SSD (Samsung 830) with nearly identical results, but that should be due to Geekbench being based on CPU and RAM speed, not disk speed.
     
  3. wonderspark macrumors 68030

    wonderspark

    Joined:
    Feb 4, 2010
    Location:
    Oregon
    #3
    Here are a couple tests. I ran one just now on 10.8.4, and the other back on 10.6.8 with one of my RAM sticks removed to see if triple channel was faster. (It was, by about 50 points.)
     

    Attached Files:

  4. COrocket macrumors 6502

    Joined:
    Dec 9, 2012
    #4
    Heres a consideration if you are going to get a 2012 Mac Pro. The drive bays are only SATA 3Gb/s, whereas most modern SSD's can saturate a 6 Gb/s connection, which is featured in the new iMacs. So theoretically the iMac equipped with an SSD can acess data twice as fast as the Mac Pro with an SSD in the drive bay. So to get all the performance possible you would have to look at a PCIe based solution, or wait until the new Mac Pros come out later this year.
     
  5. mtasquared macrumors regular

    mtasquared

    Joined:
    May 3, 2012
    #5
    Geekbench 64 bit on my 3.33 GHz mac pro = 15302

    I might have bought an iMac but I need a large display (40 inch LCD) and the ability to expand and extend the system. And little things like buffered RAM. Basically I wanted no limits on what I could do with it.
     
  6. snapdragonx thread starter macrumors regular

    Joined:
    Oct 31, 2012
    #6
    Thanks for the info guys.

    Tough decision. Think I might have to wait for the new Mac Pro. The current model just has too much old technology. I didn't realise it still used SATA 3Gb/s
     
  7. MacVidCards Suspended

    Joined:
    Nov 17, 2008
    Location:
    Hollywood, CA
    #7
    The NEW and IMPROVED Mac Pro has done away with SATA ports.

    You can buy an external TB2 enclosure just the second they materialize. It will add $100 or so dollars to the cost of connecting the drive, but the nMP will be shiny and cylindrical on the plus side.
     
  8. -hh, Jul 15, 2013
    Last edited: Jul 15, 2013

    -hh macrumors 68020

    -hh

    Joined:
    Jul 17, 2001
    Location:
    NJ Highlands, Earth
    #8
    The decision does become a bit harder with the impending release of the 6,1 However, a big part of the question comes down to what do you really need the machine for?

    For example, if you're looking at having moderate horsepower with gobs of storage, you might build up some system-level options which look like this:


    Option A (Cost $TBD)

    iMac ...w/SSD option for fast boot, apps
    TB - Promise Pegasus R4 (4TB RAID0) ... primary access data repository
    USB3 - generic 4TB hard drive(s) ... Time Machine

    Option B (Cost $TBD)

    2012 Mac Pro
    +PCIe based SSD (OWC or Sonnet) ... for fast boot, apps
    Slots 1 & 2: 2 & 2TB HDDs (4 TB RAID0) ... primary access data repository
    Slot 3: 4TB HDD .. Time Machine
    Slot 4: Repurposed OEM drive; Mirror of Boot (scheduled backup by CCC)

    Option C (Cost $TBD)

    2013 Mac Pro .. ships with SSD (size?) ... fast boot, apps
    TB - Promise Pegasus R4 (4TB RAID0) ... primary access data repository
    USB3 - random 4TB hard drive(s) ... Time Machine


    FWIW, I went through this decision process last year, although (obviously!) without the consideration of the 2013 redesign. For my particular use case, the 2012 Mac Pro was a better value proposition than the iMac, despite only having SATA2 and USB2.

    In looking at this topic further over the past month, my general read on the 2013 is that the 'Thunderbolt Tax' is roughly $150 per port used...and given the headaches that I've had with USB3 hard drives with lousy firmware which allow them to fall asleep but which can't be re-awoken without a literal reboot, I can't consider USB3 to be a particularly viable professional-level storage medium option: I pretty much only use it to make clone/burns of my backups which are to be taken off-site.


    -hh
     
  9. snapdragonx thread starter macrumors regular

    Joined:
    Oct 31, 2012
    #9
    My reason for not wanting to stick with the iMac is the sub-par display. For what I paid for it, I don't think the yellow-tinge at the bottom of the screen is acceptable. From what I've read, most of them have the same issue.

    I've used a USB 3 external RAID for the last year without issue, so moving to the new Mac Pro wouldn't be an issue for me in terms of storage.

    The 2012 Mac Pro is tempting, but upgrading it with an SSD, USB 3 and a decent GPU would just cost far too much. It's already more expensive than the iMac when BTO with a 6-core CPU.
     
  10. subsonix macrumors 68040

    Joined:
    Feb 2, 2008
    #10
    And replaced it with PCIe storage rated at 1250 MB/s.
     
  11. -hh macrumors 68020

    -hh

    Joined:
    Jul 17, 2001
    Location:
    NJ Highlands, Earth
    #11
    ...with a new form factor that Apple claims is rated at up to 1250.

    In the meantime, there's at least two vendors (Sonnet, OWC) who have been selling several Standard PCIe Slot SSD solutions which have been perfrmance benchmarked to as high as 950/800 MB/s.

    This existing I/O level is high enough to demonstrate that the system's performance bottleneck will move on the new Mac Pro to the Thunderbolt-based external storage, based on benchmarking by Digital LLoyd (incompressible data test of ~800 max on a Pegasus J4 4xRAID0 SSD).

    The bottom line is that while there's certainly gains, the majority of it was already available last year for any Mac Pro user who simply looked beyond what Apple was offering as the OEM as options within a highly expandable design.

    YMMV, but I'll mark this one down as "Pedantically true, but far less profound than is being suggested by the current Marketing'.


    -hh
     
  12. subsonix macrumors 68040

    Joined:
    Feb 2, 2008
    #12
    None of them are actually PCIe disks, they are mSATA disks (just SATA) in RAID 0. You can get the same performance from regular SSDs in RAID 0 over SATA III.

    Show me a bootable PCIe disk for the Mac Pro. All you have mentioned so far are regular SATA disks.

    Pedantically? You don't think it's worth mentioning the difference from SATA II to a new PCIe disk? In a thread about SSDs in Mac Pro, the new disk happens to be a huge improvement over SATA II.

    If you have looked at other PCIe disks it's pretty obvious that there is not reason to doubt the rated speed, at all.
     
  13. -hh macrumors 68020

    -hh

    Joined:
    Jul 17, 2001
    Location:
    NJ Highlands, Earth
    #13
    It doesn't matter that none of them are actually PCIe disks, because from a capabilities standpoint, how a particular technology works isn't what's important. What's important is what is the capability that is provided.


    My PCIe based SSD system boots my 2012 Mac Pro. Sure, we happen to know that it is because its SATA inside, but from a capabilty requirements standpoint, I do not need to care how.

    Not necessarily, and what this is really illustrating is just how easy it is for people to get distracted.

    To dictate a specific engineering solution to meet a capability requirement is an artificial constraint and a flaw in requirements generation.


    -hh
     
  14. subsonix macrumors 68040

    Joined:
    Feb 2, 2008
    #14
    Of course it does, because from a capabilities standpoint, those examples are capped by the limitations of SATA.

    They use 2 SATA disks in RAID 0 to overcome this instead of using 1 disk with a faster interface.


    If you are going to continue to call those disks PCIe based disk then a regular SATA III card magically makes your disks, "PCIe disks".


    You have this completely backwards my friend.

    PCIe is the future for storage, even sata-io have recognized this.
     
  15. -hh macrumors 68020

    -hh

    Joined:
    Jul 17, 2001
    Location:
    NJ Highlands, Earth
    #15
    No, because by your logic, no automobile should ever be sold with an internal combusion engine, because technology has shown us that the only way to ever generate more than 10,000 shaft horsepower in <ton is with a Turbine Engine.

    The technical weeds that you're going into does allow us to have insight into why X is better than Y, but from a pragmatic standpoint, when either X or Y satisfy the capability requirement, then they both satisfy the requirement, and the differentiation will be based on some other metric of significance (such as cost, future capability growth potential, size, power consumption, etc, etc).


    There's more than one way to skin a cat...the objective is still the same: that the cat gets skinned. We can come up with ways to do it with a knife, with an axe, with two knives...etc.

    Once the engineering approaches are detailed which each satisfy the capability requirement, we can then assess the trade-offs that each one brings to decide what is the best approach.


    From the simple user's perspective, all he knows is that he stuck them into an existing PCIe slot. As far as he's concerned, the rest can be magic.

    Oh, I wish that I was wrong and not so cynical. Unfortunately, I have 30 years of professional experience in product requirements definition & development and far far far too many experiences where my customers have tried to also dictate the engineering solutions that they want.

    Now please note that it doesn't mean that the engineering solutions that they specify are automatically wrong - - it is just that they're out of their lane: they're responsible for identifying the requirements.

    I can still one program from 1985 where this problem of 'dictate the solution' was so bad that we ended up with seven (7!) top management design mandates which from an engineering standpoint were impossible because they were mutually exclusive.

    It will only be the future if it is the optimal technological solution based on specific capability requirements. Otherwise, you're into the "Technology for Technology's Sake" paradigm, which is extremely rare in commercial business: T4TS is usually only found in DARPA and a few other high risk / high payoff organizations...


    -hh
     
  16. subsonix, Jul 16, 2013
    Last edited: Jul 16, 2013

    subsonix macrumors 68040

    Joined:
    Feb 2, 2008
    #16
    Well, I agree from a pragmatic standpoint it doesn't matter if 950MB/s is all you want. A raided solution may be exactly what you asked for. This topic came up because I mentioned that the new Mac Pro have added something superior to SATA II from a performance standpoint.

    The problem with generalizing you standpoint is that the raided solution doesn't scale well at all.


    Again we need to look at how this got brought up into the thread. I'm not saying that the RAID solutions are not capable. But they are limited by the speeds of SATA III, if you want to go beyond the rated 950MB/s you need to add yet another disk.



    Again, this is quite a different topic. I'm not saying that these solutions aren't fine for what they do, but it would be idiotic for Apple to use such a solution for a new design. It works well and is cost effective in the current Mac Pro, and that's fine.


    I also happen to have this experience. You may claim that the requirements are set to high, that's one thing. But to engineer a solution to meet the requirements is exactly how it's suppose to work.


    EDIT: Reading this again, I don't get why you think this is even remotely related. Are you suggesting that Apple's customers somehow proposed the engineering solution for their SSD?


    We have already established that SATA III is a bottleneck for storage, many require faster storage, hence why you got yourself the PCIe based solution. Going beyond SATA III, SATA-IO have already specified a new standard built on PCIe.
     
  17. freejazz-man macrumors regular

    Joined:
    May 12, 2010
    #17
    geez - talk about pedantic
     
  18. -hh macrumors 68020

    -hh

    Joined:
    Jul 17, 2001
    Location:
    NJ Highlands, Earth
    #18
    Not quite. We all would love to have 'faster', but what we don't necessarily want is to pay extra for a 'faster' in one part of a system when it isn't tangibly beneficial, because of performance bottlenecks elsewhere within the system that result in no net change.

    To this end, since there's no TB media that's capable of exceeding 950+MB/s performance, we have an upper pratical limit on how fast is fast enough .. or more concisely, how fast is worth paying for today (cost as an independent variable).

    Sure, within the 3-5 year lifecycle window of a 2013 MP we will want to consider how likely it is for faster TB devices to come about, but considering that the way that we get to 950 today is with a 4 x SSD RAID0, the technology path forward isn't necessarily obvious (or cheap): the capability may become available, but history shows us that it will invariably be at a price only afforded by those who need to be on the bleeding edge.


    And TB also goes to "add another disk", implimented through the binding of its controller into a larger logical unit --> and the reason why is because that's where the performance bottleneck is.

    FWIW, between these two, we should be acknowledging that there's a big difference between "State of the Art" and "State of the Shelf".


    And at that level I agree. In fact, I've even explicitly suggested 'SSD Blades' as an engineering design consideration in 'Future Mac Pro' discussions from a year or two ago...and I believe that I even suggested stealing them from the existing Apple laptop parts bin. However, even this is merely offering one engineering solution approach to a capability requirement for improving system responsiveness through higher storage I/O bandwidth...and another "skin the cat" engineering approach could have been to provision the MP with an extra 64GB of RAM and having the firmware create a RAM drive and load the entire OS there during boot-up.


    No, I wasn't talking about requirements that were too high: I was talking about requirements which were inappropriate because they dictated a specific engineering solution and because these inappropriate dictations conflicted. I can't disclose the specific product, so please endulge in this automotive analogy:

    VP#1: put in a 3L gas engine .. its the best
    VP#2: put in a 5L diesel engine .. its the best
    VP#3: put in a 4L V6 engine .. its the best
    ...(4 thru 6 - ditto)
    VP#7: put in a 2L turbo engine .. its the best

    Since we can only install one motor, just how do you think we were going to satisfy all seven requirements?

    And if you're thinking that this is "stupid" requirements, you're absolutely correct ... and my point is that it still happens anyway.


    EDIT: Reading this again, I don't get why you think this is even remotely related. Are you suggesting that Apple's customers somehow proposed the engineering solution for their SSD?[/quote]

    No. It is common knowledge that the biggest bang for the buck today is for solid state media ... and even though we know that that's ultimately going to be the engineering solution, that is not our capability requirement.

    As the saying goes in the land of Performance Specifications, they could make it out of Bubblegum and if it works, I have no reason to complain.

    Where we do get into trouble with requirements generation and vetting is when a savvy customer says "My requirement is no less than 900MB/s because I know you can accomplish this with a PCIe SSD." The reason why this requirement is flawed is because the customer hasn't really analyzed the "why" behind what he says he needs...at best, it is a derivative requirement.

    To use another automotive analogy, it would be like a race car driver saying, "You must give me 800HP because I know you can" rather than saying, "I need to turn laps at faster than X in order to win the race". The reason why the latter is the better capability requirement is because one can meet the 800HP requirement without improving lap times (example: heavier engine hurts handling, making the corners slower).


    True, but it doesn't provide any tangible benefit for non solid state spinning media ... ie, HDDs, as they're bottlenecked down at ~200 per interface node. While we can agree that solid state is the future, it is simply still too cost-prohibitive for use cases where a lot of storage capacity is required. Where we currently stand ("State of the Shelf") is that the lower end of the storage capacity capability demand is presently going to solid state, as well as some of the extremely high fringe bleeding edge. The classical "Desktop Power User" is going to be the last to transition, which is a key demographic for the Mac Pro. My personal assessment would be that we're still easily 3+ years out from such a system going to pure solid state, which puts it at the end of the likely useful lifecycle of a 2013 machine.


    -hh
     
  19. subsonix, Jul 16, 2013
    Last edited: Jul 16, 2013

    subsonix macrumors 68040

    Joined:
    Feb 2, 2008
    #19
    First of all, you or I don't know if we are going to pay extra for it. Secondly you said that you have added one of those card based SSDs yourself, so how come you now say that it isn't tangibly beneficial.

    TB is a completely different topic to what we are discussing. Apple is currently saying "up to" 20Gb/s data transfer speed. We have to wait and see where the real world limit of TB2 is, but as I said, TB is a completely different topic.

    There are other PCIe based SSDs on the market, and as I have said SATA is moving to PCIe, it's already established. Here are some other PCIe SSDs currently on the market:

    http://www.fusionio.com/products/iofx/
    http://www.intel.com/content/www/us/en/solid-state-drives/solid-state-drives-910-series.html
    http://ocz.com/consumer/revodrive-3-x2-pcie-ssd




    So why are we having this discussion, do you have something to prove? As I have said repeatedly SATA-IO have already established PCIe for SATA. There are dedicated DRAM drives on the (extreme) enterprise market but they typically have super-caps to backup in case of power failures, it's not a viable solution to use regular ram for the OS at all, because a power failure could leave the system in an unrecoverable state with data loss as a result.

    http://www.ddrdrive.com/

    They aren't inappropriate because they are on the same page as SATA-IO and very much in line with how things are developing at the moment.

    There is nothing wrong with the SSDs being PCIe based at all, building a solution based on SATA III and RAID would be very odd.

    And this is obviously completely unrelated to Apple, and the new Mac Pro.



    This is a different point, you keep changing subject. We where discussing the merits of PCIe vs RAIDed SATA III on a PCIe card.
     
  20. Dr. Stealth macrumors 6502

    Joined:
    Sep 14, 2004
    Location:
    SoCal
    #20

    The current MP is not dead by a long shot. My 2010 Geekbenches @ 27,448 and is running dual 4GB GTX 680s. Try that with the new MP when it comes out months from now.

    http://browser.primatelabs.com/geekbench2/1434763
     
  21. snapdragonx thread starter macrumors regular

    Joined:
    Oct 31, 2012
    #21
    Impressive, but is it possible to upgrade the current Mac Pro with USB 3.0, Thunderbolt and SATA 6gb/s?
     
  22. mtasquared macrumors regular

    mtasquared

    Joined:
    May 3, 2012
    #22
    I think you can get cards for usb3 and sata 6. Definitely not thunderbolt.
     
  23. Dr. Stealth macrumors 6502

    Joined:
    Sep 14, 2004
    Location:
    SoCal
    #23
    USB 3.0 - Yes

    SATA 6gb/s - Yes

    Thunderbolt is expensive Snake Oil.....
     
  24. -hh macrumors 68020

    -hh

    Joined:
    Jul 17, 2001
    Location:
    NJ Highlands, Earth
    #24
    Really? Normalized per Gigabyte, SSDs are more expensive than Hard Drives...are they not?

    No, not what I said, and not a contradiction:

    What I said was that an upgrade investment is not value-added when it does not result in an improvement on a system capability requirement.

    For my specific case, the expenditure resulted in a specific and tangible growth in system capability, which means that it added value (eg, beneficial). FYI, if it was a 'good value' is just a cost:benefit analysis and requires cost data too.


    Not really: while this paragraph was talking about the technology of the SSD being PCIe based, individual subsystems do merit being assessed at the system level...which includes the performance of data storage I/O via TB, since there's not much evidence of capacity for it locally (ie, internally).

    The first step is to determine the benefit. Specifically, if having a boot drive that's 20% faster than its data storage results in a tangible capability improvement to the system. FYI, it is tempting for us to say "take the +20% because it is 'Free'", but that's a fallacy because TANSTAAFL always applies, somewhere, somehow.


    Which merely demonstrates that an enabling technology is maturing.

    What it does not say or prove is that it must be adopted because it is the only appropriate technology.


    I've said that {unpolluted} capability requirements are what is valid for driving to specific engineering solutions. You've disagreed.


    I'm sorry you didn't understand the analogy. Here's another: you're driving a car tonight Eastbound to go to dinner and approaching an intersection: your Wife is in the front seat and telling you to turn North and your Mom is in the back seat, insisting that you turn South. Please enlighten us with just what you consequently do which will keep both happy. Good luck!


    Except that the facts of the matter are that SATA+RAID had been a contemporary solution because it was the best value. Some of that has been moving to "Fusion" drives in some appliation areas and to SSDs in others.

    What's going to happen with PCIe SSD is that it will be treated the same as any other new competing technology: it will be baselined against the old and its pros/cons determined, weighted and 'best value' versus the customer requirements will be determined.

    For a related contemporary example, consider Apple's Fusion Drive, or other similar SSD+HDD hybrids: they all seek to find a 'best value' sweet spot between speed and storage capacity by employing these two technologies.


    I'd love to be an optimist who can agree with you. Unfortunately, this has been a chronic problem in products requirements generation and Apple isn't magically immune from it happening there too...just look at their "thinner" trend.

    Specific to the new Mac Pro, wait and see; personally I suspect that its engineering design has an increased degree of reliance on future capability growth in microelectronics density improvements, which is an increase in risks and in all likelihood, costs.


    I've been discussing capabilities the entire time.


    -hh
     
  25. subsonix macrumors 68040

    Joined:
    Feb 2, 2008
    #25
    Yes, but not normalized on MB/s. Most people here get RAIDs to achieve faster throughput. Will the new Mac Pro cost more than the current one, we will have to wait and see.


    And for the new Mac Pro the upgrade is from spinning SATA II disks to PCIe solid state.


    Yes, really. Go back and see how I entered the discussion, you just want to turn this into your favorite topic.

    Going from SATA II to a 1250 MB/s SSD results in way more than 20% increase in performance.


    It's only true if you want Apple to stand still and not progress and adapt new emerging standards.


    That's not what you said in post #13.

    I specifically said that you could argue that the requirements are set too high (basically the new Mac Pro is too good), which is what you are saying now as well.


    Yes… And this doesn't relate to the new Mac Pro, at all.



    Yep, if you are talking about regular drives in a RAID. That does not mean that it's an intelligent choice for a new PCIe based SSD, since the interface is much faster than SATA to begin with, RAID and multiple drives isn't necessary.



    Yes, I have noticed this, and hence my remark. Look how I entered the discussion, this was just a comment to fill in the blanks so to speak. I know you like to turn this into your favorite topic: "the new Mac Pro is inferior". I know where you stand, and I'm not interested in discussing it.
     

Share This Page