Help: Placed in new SSD, very slow

Discussion in 'Mac mini' started by kas23, Jan 20, 2012.

  1. kas23 macrumors 603

    kas23

    Joined:
    Oct 28, 2007
    #1
    I recently bought a 2011 Mac Mini 2.5GHz 4 GB ram with stock 500 GB HDD. It is running the latest version of Lion, with only 2 applications so far: CCC and Blackmagic Disk Speed Test. I booted and ran fine before putting in the SSD, which is an OWC 120 GB Mercury Electra 6G SSD (just purchased, new). I decided to place the SSD into Mini in addition to the stock HDD. I opened up Mini, swapped out SATA connector and moved HDD to slot furthest from fan (using SATA connector from ifixit mini dual drive kit. SSD was placed in slot closest to fan, using stock SATA connector. Installation was ineventful, nothing broke, everything went back together nicely.

    I booted up using stock HDD and was able to see the SSD in there. I formatted it and then cloned the HDD to the SSD using CCC. I was able to boot using SSD, boot time being roughly equal to stock HDD actually. For now, I have left the original bootable copy of Lion on the HDD, but start-up is switched to SSD in System Preferences.

    To test speed of new SSD, using Blackmagic, I get read speeds of 190 MB/s and write speeds of 140 MB/s. This is a far cry from the OWC advertised speeds and some people on here have achieved much faster speeds. Do you think this is an accurate reading? (I tried using AJA Kona, but I can never get this to work for me, on any of my Macs. I press Start and it the button flashes Stop, but then goes right back to Start.) Is there something else that I am missing or is this a defective SSD? Could the Lion boot copy on the stock HDD be interfering in some way?
     
  2. paulrbeers macrumors 68040

    Joined:
    Dec 17, 2009
    #2
    You can try this:

    http://support.apple.com/kb/HT3964

    But you also have to understand, different benchmarking tools will use different criteria for their benchmarks. I don't know enough about Blackmagic, but, for example, a benchmark that is heavily based around incompressible data, the Sandforce 2200 drives do not do well in (which is what OWC drives are based on). It also depends on if it is random reads and writes, as opposed to consecutive reads/writes will also have an adverse affect on the overall performance. You also may not be getting a SATA3 connection (something may have gotten crimped when installing?).

    Here's the real question at hand, is it considerably faster than your mechanical hard drive? Do you feel it made a significant difference? Then move along and be happy. People get so caught up in benchmarks, i.e "why does this guy score this when I only got this", it just drives me nuts. I never benchmark my systems. I am confident in my man-hood. If everything is flying, then there isn't a reason to get caught up in the "oh god, I'm not getting X"....

    Frankly, your drive is probably out working you anyway
     
  3. FrankHahn macrumors 6502a

    Joined:
    May 17, 2011
  4. paulrbeers macrumors 68040

    Joined:
    Dec 17, 2009
    #4
    Just pointing out that Intel based Macs do not have PRAM, instead it is NVRAM.
     
  5. jamesryanbell macrumors 68020

    jamesryanbell

    Joined:
    Mar 17, 2009
    #5
    Search for my thread about the OWC SSD I put in my 11" MBA. Same thing with Black Magic. OWC tells me "they don't test well with programs like that".
     
  6. kas23 thread starter macrumors 603

    kas23

    Joined:
    Oct 28, 2007
    #6
    Will do.

    I have heard blackmagic not being accurate with OWC SSD's too. Woudl you recommend any other benchmarking software to test disk speed? I have downloaded xbench, but for some reason this is not working. I have not been able to get AJA Kona to work with either my 2011 MacBook Air running Lion or my new Mini. Does anyone else know why?

    As for my SATA connection, it could be crimped because it does have to bend a couple of times to fit onto the logic board, but wouldn't the drive fail to be recognized then? My Mini not only recognizes the new SSD, but it can boot off of it fine (I suppose).

    I completely agree with your viewpoint. If it works, it works. But I don't know if it is working to its full potential and it is quite disappointing when they advertise a particular speed and you get less than half of it. Plus, I have seen benchmarks from other users where the OWC SSD has been very fast.

    Nevertheless, would you recommend resetting it. Could there be any adverse consequences from resetting the NVRAM?

    Have you used any alternative software (besides AJA Kona - which I can seem to get working with Lion) to test disk speeds?
     
  7. xSavagedMonkey macrumors newbie

    Joined:
    Jan 20, 2012
    #7
    I just installed a OCZ Agility 3, according to OCZ the drive can reach speeds of 500mbps read/write on SATA 3.

    Now my Mac mini is saying that it's on SATA 3

    [​IMG]

    And here's the results I get in Blackmagic speed test.

    [​IMG]

    I'm beginning to think the program isn't accurate for this specific drive.
     
  8. Tyr., Jan 20, 2012
    Last edited: Jan 20, 2012

    Tyr. macrumors member

    Joined:
    Aug 1, 2007
    Location:
    Antwerp, Belgium
    #8
    That makes sense since classic tools will test speed with a sequential write/read. That's OK for classic HDD's because the data is written sequentially on the platter but IIRC SSD's get a lot of their speed from splitting the data between the different flash memory chips on the device. There's a reason sites like Anandtech use their own benchmarking tools to gauge SSD performance.

    Why not do some real life tests ?

    Generate a 4Gb file and see how long it takes.

    Code:
    time mkfile 4g ~/Downloads/test
    Then duplicate it and see how long that takes.

    Code:
    time cp -p ~/Downloads/test ~/Downloads/test.2
    That'll give you an indication of wether or not the synthetic benchmark was a long way off or not.
     
  9. xSavagedMonkey macrumors newbie

    Joined:
    Jan 20, 2012
    #9
    I thought I'd give it a try,

    Code:
    Liams-Mac-mini:~ Liam$ time mkfile 4g ~/Downloads/test
    
    real	0m16.856s
    user	0m0.899s
    sys	0m15.384s
    Liams-Mac-mini:~ Liam$ time cp -p ~/Downloads/test ~/Downloads/test.2
    
    real	0m16.797s
    user	0m0.010s
    sys	0m3.648s
    Liams-Mac-mini:~ Liam$ 
    
    
    Is this good, bad?
     
  10. Tyr., Jan 20, 2012
    Last edited: Jan 20, 2012

    Tyr. macrumors member

    Joined:
    Aug 1, 2007
    Location:
    Antwerp, Belgium
    #10
    The first one was the slowest and in that one you wrote 4096 megabytes of randomly generated data to disk in 16.856 seconds or at a rate of almost 243 megabytes per second for a pure sequential write, that's already a lot higher than your previously tested 184 MB/s.

    Looks like that's in line with Anandtechs' sequential write test for random data. So a good result I'd say.
     
  11. xSavagedMonkey macrumors newbie

    Joined:
    Jan 20, 2012
    #11
    Thanks, according to OCZ on SATA 3 it's 500mbps/480mbps any idea why it's half that? Do yo think putting my SSD on data connector 1 instead of 2 will help reach it's true speeds?
     
  12. Tyr. macrumors member

    Joined:
    Aug 1, 2007
    Location:
    Antwerp, Belgium
    #12
    Those are theoretical values. In real life there's overhead everywhere: SATA commands, filesystem limitations, bus speeds, firmware on disk controllers, etc. Using a different connector should make no difference, they both have the same bandwidth allotted to them.
     
  13. xSavagedMonkey macrumors newbie

    Joined:
    Jan 20, 2012
    #13
    Cheers, I'm seeing huge difference in speeds with boots times, loading apps etc so I'm happy :) Thanks for your help.
     
  14. iRCL macrumors 6502

    Joined:
    Nov 2, 2011
    #14
    Be ready for the real fun when the OWC SSD starts running extremely slowly after a good amount of use because of TRIM issues as detailed in many many many threads on these forums (including many posts about awful experiences contacting OWC about those problems.. 'working as expected')

    And yes all the SSDs particularly anything using Sandforce technology straight up lies to consumers by presenting them with data rates in the absolute best case of compression etc.

    I think it's hilarious how everyone hates on black magic and leaves nasty app feedback on the apple app store for it giving "bad" results - realize it's giving you the real results.....

    Based on everything I have seen here I would never use a non apple SSD in an apple product - and I am a serious aftermarket kind of guy in general
     
  15. Dhelsdon macrumors 65816

    Dhelsdon

    Joined:
    Feb 5, 2010
    Location:
    Canadian Eh!
    #15
    Here's mine with a 128GB OCZ Vertex Turbo

    Code:
    Devons-MacBook-Pro-2:~ devoniscool$ time mkdir 4g ~/Downloads/test
    
    real	0m0.004s
    user	0m0.000s
    sys	0m0.002s
    Devons-MacBook-Pro-2:~ devoniscool$ time cp -p ~/Downloads/test ~Downloads/test.2
    cp: /Users/devoniscool/Downloads/test is a directory (not copied).
    
    real	0m0.005s
    user	0m0.001s
    sys	0m0.002s
    and from BlackMagic I get average 150mbps write and about 200mbps read
    mind you that is just me lowballing the averages.
     
  16. iRCL macrumors 6502

    Joined:
    Nov 2, 2011
    #16
    You do realize that nothing was copied because you had to use cp -R since you're using a directory. So all you've shown is how long it takes the system to generate an error message
     
  17. Dhelsdon macrumors 65816

    Dhelsdon

    Joined:
    Feb 5, 2010
    Location:
    Canadian Eh!
    #17
    Oh, I didn't know that. I am a noob with terminal, so the first part wouldn't have created a file at all then is that correct?
     
  18. melissajenna macrumors newbie

    Joined:
    Jan 20, 2011
    Location:
    Shell Beach, CA
    #18
    Hey there, MJ from iFixit here. Will you please email your issue to support@ifixit.com? I'm thinking we might be able to help you out. :)
     
  19. jamesryanbell macrumors 68020

    jamesryanbell

    Joined:
    Mar 17, 2009
    #19
    Yes, there's a version of AJA that works with Lion if my memory serves me correctly. But yes, I've used at least two different programs, and I get substantially lower numbers than anything on OWC's site. I don't hate the drive or anything (I mainly got it for the space), but it's just kind of lame to claim crazy fast numbers and then when I ask them about why such low numbers, they say only one program will give you those (which you have to purchase). Not too cool in my opinion. ANY benchmarking software should yield reproducible numbers with something as simple as read and write speeds of an SSD.
     
  20. kas23 thread starter macrumors 603

    kas23

    Joined:
    Oct 28, 2007
    #20
    Done. You guys are awesome.

    I'm thinking I may have found out my problem. More testing...
     
  21. cardude280 macrumors member

    Joined:
    Feb 26, 2006
    Location:
    West Virginia
    #21
    You do know that trimenabler works on sandforce drives to enable trim in SL and lion right?

    And while not straight up lies they are showing something different, BlackMagic shows incompressable sequencial read and write speed for video editing, while most of what people work with are small compressable text/.dylib type files
     
  22. kas23 thread starter macrumors 603

    kas23

    Joined:
    Oct 28, 2007
    #22
    I keep reading that trim cannot be enabled for 3rd party SSDs in Lion. Is this just a bunch of smoke and mirrors? A way of scaring people into buying factory installed SSDs? I swear, there are some extremely convincing people out there that will tell you the world will end if you install a 3rd party SSD into a Mac.
     
  23. shortcut3d macrumors 65816

    Joined:
    Aug 24, 2011
    #23
    You DO NOT want to enable trim on Sandforce drives in OSX. Just rely on the over provisioning and garbage collection. Let the drive idle for sometime. Also, the slow speeds could be caused by spotlight indexing on a fresh install.

    Also, do not benchmark too much because the controller will throttle drive back for GC, maintenance and reliability. OCZ has a reset tool for their drives, not sure if OWC has one.
     
  24. kas23 thread starter macrumors 603

    kas23

    Joined:
    Oct 28, 2007
    #24
    I found the reason for part of my problem. Hopefully someone can learn from my mistake. One thing I didn't mention earlier was that my Mini was running VERY hot. Now, I've seen plenty of posts on here and on the Apple forums that this is normal. But, this was fry-an-egg on it hot. I could hear the fan running, but I really had to listen.

    I decided to run a hardware test on it (held "D" down at boot) and I received an error:

    4MOT/1/40000002: Exhaust - 0

    A quick search told me that my fan was on the fritz. A powered down and popped the Mini open to find the fan not attached to the logic board. I must've not seated it properly. I seated it and ran the hardware test again. Passed. And no more superhot Mini.

    As for the SSD disk speeds. Still the same. I ran the terminal commands early in this thread and got speeds on the 300s.
     
  25. spyderx macrumors member

    Joined:
    Mar 20, 2011
    #25
    These tools need to be taken in context and compared to others with similar configurations. You can't expect a given benchmark tool to show you theoretical maximums of a given drive or interface.

    What are they good for?
    --Compare before and after performance of an install
    --compare drive a to drive b on a given system
    --compare to others using same software

    The #'s are the #'s and I wouldn't take them for being perfect. They are good for A to B comparison, that's about it.

    As an example: My cheap, 64GB Sandforce based SSD running a 3gbs connection on my mac mini 5.1 gets a 75mb/s write / 180mb/sec read rating in Black Magic. that's a heck of a lot faster than the 500gb drive the machine came with.
     

Share This Page