SSD Speeds: Faster Reads or equal?

Discussion in 'MacBook Pro' started by Hexley, Dec 17, 2012.

  1. Hexley macrumors 6502a

    Joined:
    Jun 10, 2009
    #1
    Looking at the benchmarks of the two leading SSDs on today's market, Samsung 840 Pro and the Vector, I notice something odd between them. Samsung had faster reads than writes. While the Vector had roughly equal write/read throughput.

    Write
    467.3MB/s: Samsung 840 Pro
    490.5MB/s: OCZ Vector

    Read
    515.3MB/s: Samsung 840 Pro
    497.0MB/s: OCZ Vector

    Benchmark used is Blackmagicdesign's Disk Speed Test.

    My guess is that this was purposely done. For reasons I do not know and would like to know through your comments below. :)
     
  2. Rizzm macrumors 6502a

    Joined:
    Feb 5, 2012
    #2
    What exactly is your question? They're just different.
     
  3. Hexley thread starter macrumors 6502a

    Joined:
    Jun 10, 2009
    #3
    My question is what should be given more importance. A drive that has equal write/read speeds or a drive that reads faster than it writes?
     
  4. Rizzm macrumors 6502a

    Joined:
    Feb 5, 2012
    #4
    I can't give a researched answer but I would guess it doesn't matter.
     
  5. Hexley thread starter macrumors 6502a

    Joined:
    Jun 10, 2009
    #5
    Me either. :D

    I guess I'll go with power consumption then.
     
  6. Rizzm macrumors 6502a

    Joined:
    Feb 5, 2012
    #6
    Well first I'd get the fastest read AND writes I can find. I just wouldn't compare the ratio of read/write on the same drive to base my purchase. Find one with the fastest reads and writes you can get.

    Price and performance of both read/write were my criteria and I ended up with a 240GB Mushkin Chronos Deluxe. Managed to find one for $140. 8D
     
  7. CJM macrumors 65816

    CJM

    Joined:
    May 7, 2005
    Location:
    U.K.
    #7
    If you really want to know, read Anandtech's articles on flash memory types and why there's a difference in the new Samsung drives.

    As a heads-up, the first one is very technical.
     
  8. Hexley thread starter macrumors 6502a

    Joined:
    Jun 10, 2009
    #8
    Thank you for the link CJM.

    My question is more on which SSD is right for me. Whether a drive that has equal write/read speed is better than a slower write and faster read.

    It appears that a lot of flash memory I've used tend to have better read speeds than write (CF, SDXC, iPods, iPads and iPhones).

    What caught my attention is the power consumption of the 840 Pro. If installing it on my MBPro will lengthen to a significant degree my battery life then I'm all for it.
     
  9. Weaselboy Moderator

    Weaselboy

    Staff Member

    Joined:
    Jan 23, 2005
    Location:
    California
    #9
    The differences you are seeing between these two drives is so small that you will never ever be able to tell the difference using the drive. Nothing about those speed tests would nudge me one way or the other. Just buy whichever one you think will be the most reliable/compatible and is cheapest.
     
  10. kfmfe04 macrumors member

    Joined:
    Jun 19, 2010
    #10
    I agree - I can't tell much of a difference between a read and a write (but I haven't really benchmarked it either). However, one big difference is in the heat generated.

    Reading appears to be be much cooler than prolonged writing.
     
  11. robvas macrumors 68020

    Joined:
    Mar 29, 2009
    Location:
    USA
    #11
    Do not put too much thought into the maximum sequential write/read speeds of a certain drive. It's a pretty meaningless benchmark unless you are copying a huge amount of data around the drive. You very rarely see it in the real world.

    You should be more worried about smaller, random transfers.

    Either way you probably won't be able to tell a difference in the latest SSD's. But BlackMagic isn't really a great tool to compare drives with, other than for bragging rights.
     
  12. dusk007 macrumors 68040

    dusk007

    Joined:
    Dec 5, 2009
    #12
    Speed of an SSD doesn't matter anyway. They are all so close in Speed that it makes virtually no difference. Random reads are already so fast that you need some 16+ core system you really get any use out of it other than benchmark numbers. Sequential reads and writes both generally don't matter because there is just barely any situation where they happen often enough (again other than in benchmarks).
    You will almost always be limited by the source target drive you copy too. Just copying or expanding archives on the same drive needs a fast copy algorithm which is somewhere at about half that speed anyway.
    Today one hardly feels the difference between 250 and 500 MB/s in real world use. If you map it on a scale of 1 to 10 in perceived system speed.
    1 being an 5.4k HDD
    2 a really fast 10k rpm HDD.
    SSDs with 250MB/s are at 8 and 500MB/s at 8.5.
    A RAM drive is at 10.

    The only thing I would care about on modern SSDs is power consumption and reliability. Everything else is too close to matter. Unless you drive some ridiculously fast workstation it makes no difference.

    In general though reads are always much more important than writes. Writes are usually cached in RAM anyway and practically don't matter in real life there is virtually no source for data that would feed data to the SSD to sustain anywhere near the possible speed unless you generate garbage data with a CPU in RAM and write that to the SSD (such as benchmark tools do). Never ever happens in real life, even swapping is too random and the difference is never felt as it is such a background process.
    Reads at least always help a little but most of the time today the CPU limits app launch times and such. The SSDs stopped to matter one generation ago IMO.

    There is virtually no difference that you can measure without a stop watch.
    http://www.computerbase.de/artikel/laufwerke/2012/test-samsung-serie-840-pro-256-gb/7/
    http://www.computerbase.de/artikel/laufwerke/2012/test-samsung-serie-840-pro-256-gb/6/
    Except for unpacking operations the real word results are all over the board and almost random and barely a difference.
    The crucial M4 only has some 250MB/s in writes (which is 1/2 the vertex 4) and is on all of these installation tests almost equally fast.
     
  13. richnyc macrumors regular

    richnyc

    Joined:
    Nov 8, 2012
    Location:
    NYC
    #13
    ^^^ This. Agreed 100%!!!

    I think the rest is just a waste of resources and money... It took me a few weeks to realize when I jumped into upgrading my MBP and read all the reviews I could get my hands on;)
     
  14. Orlandoech macrumors 68040

    Orlandoech

    Joined:
    Jun 2, 2011
    Location:
    Salt Lake City, UT
    #14
    Can you do me a favor and provide BlackMagic Screens of each of these benchmarks, just a screen of the app, not of your user background?

    I have posted an example/. [​IMG]
     
  15. derbothaus macrumors 601

    derbothaus

    Joined:
    Jul 17, 2010
    #15
    After 400MB/s writes matter more. Cheap SSD's have a write disparity vs. the faster high tier. As a boot HD you utilize 4K random writes 60% of the time the drive is active so yes, it matters.
     
  16. d4m1r macrumors regular

    d4m1r

    Joined:
    Oct 1, 2011
    Location:
    Canada
    #16
    If I had to pick between faster read or faster writes, I'd go with the function I do most often...And that is reading in most peoples case.
     
  17. johnnyturbouk macrumors 68000

    johnnyturbouk

    Joined:
    Feb 9, 2011
    Location:
    on the yellow [oled] brick road to tech nirvana.
    #17
    hit the needle on the head!

    i have had 3 generations of ocz ssds without issue - blazing fast performance

    i think i wil go with the samsung ssd next: 256 gb 840pro
     
  18. dusk007 macrumors 68040

    dusk007

    Joined:
    Dec 5, 2009
    #18
    60% maybe true but the throughput of those is really low. Therefore speed doesn't translate into speed. It is like saying with a car you drive most 60% of the time and driving speed matters. If you drive in a 30km/h zone, it won't make any difference whatsoever that you are driving a Bugatti Veyron that can go 400km/h in theory.
    4k random write are really low queue and really low throughput because if it was more the disk cache catches all. Effectively any system drive as a RAM cache for writes so any too slow write will never be felt.
     
  19. tivoboy macrumors 68030

    Joined:
    May 15, 2005
    #19
    read faster

    reads are pretty much always faster than writes
     
  20. derbothaus macrumors 601

    derbothaus

    Joined:
    Jul 17, 2010
    #20
    I can tell the difference on gen 2 SSD vs. gen 3 SSD when bandwidth is capped in sequentials (ie. SATA2 @ 265MB/s) The gen 3 SSD with 85-100MB/s 4K does feel faster than the 25MB/s 4K drive. Feeling 65MB/s vs. 85MB/s may be debatable. My point was to not base SSD speed purely on advertised sequential speeds. I know how confusing that is for many users.
     

Share This Page