Are my hard drives slow?

Discussion in 'Mac Pro' started by drnebulous, Aug 28, 2014.

  1. drnebulous, Aug 28, 2014
    Last edited: Aug 28, 2014

    drnebulous macrumors regular

    drnebulous

    Joined:
    Apr 27, 2014
    Location:
    Salford, UK
    #1
    I have 6 hard drives in my mac pro and I'm sure some of them are too slow.


    Western Digital Caviar Black 500gb = 90mb/s write + 111mb/s read
    Western Digital Caviar Black 500gb = 60mb/s write + 75mb/s read
    Western Digital Caviar Black 1tb = 80mb/s write + 90mb/s read
    Western Digital SE Enterprise 1tb =172mb/s write + 175mb/s read
    Samsung 160GB = 50mb/s write + 51mb/s read
    Samsung 840 Pro 128gb = 300mb/s write + 380mb/s read

    The caviar black hd's seem very slow in comparison to the SE. Is there something wrong? The Samsung 160gb is quite old, so that is sort of understandable (maybe?), but these caviar black hd's are only 3 months old. This is quite worrying really as I paid hundreds out for all these hard drives.

    I performed all the ratings with BM Disk Speed Test at a 5GB stress.


    How would I upgrade the firmware for these?
     
  2. goMac macrumors 603

    Joined:
    Apr 15, 2004
    #2
    What are you using to put six drives in the machine? I'm assuming maybe you're using the optical bays? What drives are in which bay?
     
  3. cebseb macrumors regular

    Joined:
    Feb 21, 2011
    #3
    That's very curious. Approximately how much space is being used in all the drives? I would have actually put the two black drives in a raid 0.
     
  4. drnebulous thread starter macrumors regular

    drnebulous

    Joined:
    Apr 27, 2014
    Location:
    Salford, UK
    #4

    The Samsung 160GB is in the optical bay slot next to the cd drive, the Samsung SSD is on a SATA III PCI card, and the rest are in the standard hard drive slots.

    There is only 150gb to 250gb used on each hard drive (except for the two samsungs and the 1tb WD which has 450GB used). I can't put them in RAID 0 as I need one to write and the other to read for different purposes. I'm a sound engineer so I have everything spread between different hard drive to enable simultaneous read/write.
     
  5. cebseb macrumors regular

    Joined:
    Feb 21, 2011
    #5
    OK, I still have no answer for you. The info you gave me just made me even more confused.

    The speeds are acceptable, but the difference in the two blacks is disconcerting. Are these numbers reproducible?
     
  6. westrock2000, Aug 29, 2014
    Last edited: Aug 29, 2014

    westrock2000 macrumors 6502

    Joined:
    Oct 18, 2013
    #6
    Just a couple suggestions.

    The 160GB drive is slower mostly because it uses platters that have less data density, meaning not as many bits are packed into the same area. So the platter has to physically move further along to retrieve the same amount of information as a newer drive with more data packed into the same area.

    While the 500GB drives are smaller then the 1TB you have, it depends on if you have the newer version (WD5003AZEX) which is a gimped 1TB platter or the older version (WD5002AALX) which uses a 500GB platter. So you would have to know the actual makeup of how many platters and how big each platter is to understand how fast it should be. But "generally" a bigger drive will be faster then a smaller drive, especially if they are separated by a couple years or more, mostly due to the newer drive having denser platters.

    Now the other thing that affects speed on a platter drive is the location on the platter where the test is done. A platter is a spinning circle, and you have to remember that the further out you move on a circle, the more distance you will cover on the circle for a given "speed" (RPM) of the platter.

    So if your drives already have some data on them, when you go to use your benchmark, it is going to write and read to an unused area of the drive. Platter drives always start filling from the outside edge and move in. This means if the drive already had a considerable amount of data on it, then the benchmark will occur further in on the platter, which means you will get considerably less speed. From outer edge to inner edge, you can see a hard drives bandwidth drop in half. There's nothing that can be done about this, and it's one of the strengths of an SSD.

    I would think that the raw speed would still be plenty for reading and writing song files. The access speed of the platter drives would be more a problem for seeking through the song (if it always reads from disk). For music...assuming not HUGE master files, I would always recommend using an SSD, again if space permits. For video, unfortunately due to the raw capacity needed you are almost always stuck using platter drives.

    But again, it's up to you and your application. If you do not need anymore speed, do not worry about. And if you are having stuttering problems, that is more related to the access times which nothing will fix on a platter. RAID 0 will not affect access times. Only a faster spindle speed (10K or 15K RPM) or moving to SSD. If you are having problems where you are making small edits to the song file, that is not related to the transfer speeds you benchmark is showing. That is related to your hard drives ability to go in and change "random" areas of data on the platter. This is called the "4K Random" benchmark and is the Achilles heal of all platter drives because they HATE moving from one part of the platter to the other. Spindle speed can help a tiny bit, but the only true cure for random access speeds is moving to SSD. The absolute fastest platter drive in the world, is slower in 4K Random benchmarks then the cheapest, crappiest SSD's from 4 years ago.

    4K Random is really the meat and potatoes of a hard drives ability. You have two extremes on a hard drives speed. Sequential and Random. One will be the fastest a drive can function at and one will be the slowest. Real world usage is much closer to the Random, then it is to the Sequential, unless you use the drive specifically to transfer large files such as video. Even pictures and compressed songs that are 5-10MB in size do not fully reflect sequential because the drive writes them too fast and has to move around. To see sequential you really have to be reading or writing a couple 100MB or more.
     
  7. Macsonic macrumors 65816

    Macsonic

    Joined:
    Sep 6, 2009
    Location:
    Earth
    #7
    I know we always use Black Magic to test out HDs but you can try doing a real world test to find out. One time I used Black Magic to test out a Macally Ext HD and a G Tech ext HD. Both Black Magic tests showed different speed test as the Macally seemed slower. So I did a hands on real world test and with a use of a stopwatch, compared the speed in transferring the same files. The real world test showed similar speeds in file transfers for both MacAlly and GTech.
     
  8. shaunp macrumors 65816

    Joined:
    Nov 5, 2010
    #8
    As you are getting different speeds on the two Caviar blacks I would look at how they are connected. Some of the drives would be using the onboard SATA via the drive cages, how are the others connected?

    Also what type of speed are you needing for your intended Apps? Benchmarks are only relevant to prove whether or not the drives are providing the performance you expect, but you'd know that from the responsiveness of the app anyway.

    Personally I don't see the point in buying HDD's except for bulk storage (office files, mp3's) or for apps that stream data rather than requiring random access. I can't help thinking you would have been better off not getting the two 500GB Caviar's and getting a single SSD instead. Yes it's a bit more expensive, but you wouldn't be wondering whether it's going to be quick enough.
     

Share This Page