Sign in with
Sign up | Sign in

Benchmark Results: HD Tune

Plextor M5 Pro 256 GB Review: 19 nm NAND And Marvell's Latest
By

We benchmark with HD Tune Pro over the full span of the drive using a 64 KiB write access pattern. What we're looking for is consistent write performance.

The test involves writing data to every sector of the drive while measuring speed and latency. In the graphs below, the orange line represents write performance and the yellow dots indicate the write access time.

Crucial m4

Samsung 830

Plextor M5S

Plextor M5 Pro

OCZ Vertex 4

Up until now, our benchmarks have left plenty of free space on each drive. It's only when we use HD Tune to write across the full span of our contenders that we see OCZ's Vertex 4 stumble as it switches from SLC to MLC mode (for an explanation of this, check out this page).

Although the 256 GB Vertex 4 is better able to mitigate the impact of the switch-over than the 128 GB model, there is still a performance degradation as free capacity is consumed. The phenomenal write performance we saw when the Vertex 4 was in SLC mode has to be tempered by the amount of space available and the size of the writes within that free space.  

Our M5 Pro doesn't dominate, but its performance is both solid and consistent. However, it allows Samsung's 830 to emerge as the fastest drive in maximum, average, and minimum write speeds.

Next, we measured read performance, filling each drive using an Iometer test file before running HD Tune. We then used a 4 MiB sequential read access pattern across the full span of each contender.

Crucial m4

Samsung 830

Plextor M5S

Plextor M5 Pro

OCZ Vertex 4 

Plextor's M5S does best in average read speed measurements, while Samsung's 830 and the M5 Pro finish neck and neck.

The Vertex 4 can't compete with the M5 Pro. Its read speeds start off a lot slower than the other drives, and then drop further as the benchmark gets to the last quarter of its run.

As far as large block read transfers go, there really isn't a whole lot of variation between the drives, so user experience in that context isn't going to change much from one model to the next. 

Now, let's compare our Anvil's Storage Utility and HD Tune results to each  manufacturer's sequential read specifications.


Manufacturer Specification (MB/s)
ASU (MB/s)
HD Tune (Avg MB/s)
HD Tune (Max MB/s)
Samsung 830
520
506
496
507
Crucial m4
500
479
480
498
Vertex 4
560
418
389
496
Plextor M5S
520
506
503
521
Plextor M5 Pro
540
504
496
504


There is no universally accepted methodology SSD vendors use to rate their products. Most specs prefaced with an "up to," and a number of different metrics can be used to generate that value.

Our ASU and HD Tune results illustrate this nicely. The most notable deviation from manufacturer specs comes from OCZ's Vertex 4. The company states that is uses ATTO to come up with its sequential read speed numbers. We ran ATTO ourselves and, sure enough, on an empty drive using a 256 MiB test file and a 1 MiB access pattern, we achieved 560 MiB/s. But, if we increased the test file size, we were unable to replicate that number.

On the other hand, while Plextor's M5S slightly exceeds its specification in HD Tune, the M5 Pro missed its mark by 36 MiB/s.

Ask a Category Expert

Create a new thread in the Reviews comments forum about this subject

Example: Notebook, Android, SSD hard drive

Display all 23 comments.
This thread is closed for comments
Top Comments
  • 15 Hide
    grantwar , September 28, 2012 10:35 AM
    I'd love to see how this drive fares against the samsung 840 Pro.
  • 13 Hide
    SpadeM , September 28, 2012 10:56 AM
    grantwarI'd love to see how this drive fares against the samsung 840 Pro.


    Here you go:
    http://www.anandtech.com/bench/Product/665?vs=646
Other Comments
  • 0 Hide
    Anonymous , September 28, 2012 4:43 AM
    You have the wrong Marvell controller listed for the crucial M4 and the Plextor M5S. This controller is new.
    The old one was the 88SS9174-BKK2.
    http://www.tomshardware.com/reviews/plextor-m3-crucial-m4-octane-performance-pro,3178.html
  • 9 Hide
    mayankleoboy1 , September 28, 2012 5:17 AM
    can we have a benchmark of the time it takes to install a fresh copy of Win7+SP1 on a SSD ? Because thats the first thing a user will do after buying a new drive.
  • 4 Hide
    echondo , September 28, 2012 5:54 AM
    mayankleoboy1can we have a benchmark of the time it takes to install a fresh copy of Win7+SP1 on a SSD ? Because thats the first thing a user will do after buying a new drive.


    I believe we can all assume it will take around 10-15 minutes. My old SATA2 Vertex drive can have Windows installed with all the Windows updates I want in around 20 minutes, it would be less time but I have to install all my drivers first for my motherboard :p 
  • 0 Hide
    fat-chunk , September 28, 2012 10:24 AM
    Long live SSDs!
  • 15 Hide
    grantwar , September 28, 2012 10:35 AM
    I'd love to see how this drive fares against the samsung 840 Pro.
  • 13 Hide
    SpadeM , September 28, 2012 10:56 AM
    grantwarI'd love to see how this drive fares against the samsung 840 Pro.


    Here you go:
    http://www.anandtech.com/bench/Product/665?vs=646
  • 6 Hide
    JeanLuc , September 28, 2012 12:30 PM
    I was on Youtube the other day on the Corsair channel and they were showing the advantages of 'Ram cache' which was lights years faster then SSD's in therms of throughput. Could Toms consider doing an article into Ram cache as I think it would be of interest to people who have 16-32Gb systems (since DDR3 is cheap at the moment) can spare the extra system ram to cache files and software.
  • 6 Hide
    merikafyeah , September 28, 2012 12:35 PM
    SpadeMHere you go: http://www.anandtech.com/bench/Product/665?vs=646

    Wow. The 840 Pro beat the M5 Pro in virtually everything according to that data. The 840 Pro does cost significantly more though.
  • 4 Hide
    merikafyeah , September 28, 2012 12:41 PM
    JeanLucI was on Youtube the other day on the Corsair channel and they were showing the advantages of 'Ram cache' which was lights years faster then SSD's in therms of throughput. Could Toms consider doing an article into Ram cache as I think it would be of interest to people who have 16-32Gb systems (since DDR3 is cheap at the moment) can spare the extra system ram to cache files and software.

    TheSSDReview did something similar:
    http://thessdreview.com/our-reviews/romex-fancycache-review-ssd-performance-at-13gbs-and-765000-iops-in-60-seconds-flat/

    No point in even comparing RAM to SSDs, as even "slow" RAM is faster than even the best SSDs by about the same amount as the best SSDs are faster than floppy disks.
  • 0 Hide
    aicom , September 28, 2012 5:09 PM
    This review would have been impressive if it was published 2 weeks ago. With the 840 Pro out, it simply blows the M5P out of the water. Too bad it wasn't included in the benchmark charts here.
  • 0 Hide
    richard hart , September 28, 2012 5:27 PM
    @ rafale. Thanks for spotting that mistake. Its now been corrected.
  • 0 Hide
    bobdylan11 , September 28, 2012 9:42 PM
    woot, installing mine at this moment
    450 for 512gb
    good luck beating that "stat", samsung
  • 0 Hide
    KenZen2B , September 29, 2012 4:37 AM
    merikafyeah 09/28/2012 2:41 PM Insert quote.
    Report
    -1+ .
    JeanLuc :

    I was on Youtube the other day on the Corsair channel and they were showing the advantages of 'Ram cache' which was lights years faster then SSD's in therms of throughput. Could Toms consider doing an article into Ram cache as I think it would be of interest to people who have 16-32Gb systems (since DDR3 is cheap at the moment) can spare the extra system ram to cache files and software.



    TheSSDReview did something similar:
    http://thessdreview.com/our-review [...] onds-flat/

    No point in even comparing RAM to SSDs, as even "slow" RAM is faster than even the best SSDs by about the same amount as the best SSDs are faster than floppy disks.

    I have read the article and have these thoughts on the subject.
    1) Why have mobo manufactures not placed an extra four memory slots on their mobo to handle this concept ?
    2) Why have RAM manufactures not teamed up with mobo manufactures to supply them with 256 GB memory sticks ?
    3) Has the industry, video and others, (movies, ads, special effects, cartoons, simulations, etc.) moved to this type of strategy to increase their productivity ?
  • 1 Hide
    luciferano , September 29, 2012 6:14 AM
    KenZen2Bmerikafyeah 09/28/2012 2:41 PM Insert quote. Report -1+ .JeanLuc :I was on Youtube the other day on the Corsair channel and they were showing the advantages of 'Ram cache' which was lights years faster then SSD's in therms of throughput. Could Toms consider doing an article into Ram cache as I think it would be of interest to people who have 16-32Gb systems (since DDR3 is cheap at the moment) can spare the extra system ram to cache files and software. TheSSDReview did something similar:http://thessdreview.com/our-review [...] onds-flat/No point in even comparing RAM to SSDs, as even "slow" RAM is faster than even the best SSDs by about the same amount as the best SSDs are faster than floppy disks.I have read the article and have these thoughts on the subject.1) Why have mobo manufactures not placed an extra four memory slots on their mobo to handle this concept ?2) Why have RAM manufactures not teamed up with mobo manufactures to supply them with 256 GB memory sticks ?3) Has the industry, video and others, (movies, ads, special effects, cartoons, simulations, etc.) moved to this type of strategy to increase their productivity ?


    1) mobo manufacturers would have nothing to connect these RAM slots too. They need DDR3 controllers and they'd need more of them, so where are those going to come from? The only thing that I can think of is some sort of integrated PCIe device and it would make the motherboard quite expensive.

    2) 256GB memory sticks would cost thousands of dollars. Even 16GB memory sticks and especially 32GB memory sticks can already be extremely expensive and difficult to make. We can't just make chips that have more memory capacity in a given size than current memory procces nodes can work with, so any such memory module with 256GB of RAM would be huge, to say the least.

    3) It's too expensive.

    Basically, cost is the main inhibitor for all three of your questions.
  • 0 Hide
    merikafyeah , September 29, 2012 12:11 PM
    Maybe it's for the best that they don't release the Limited Edition Shuriken model worldwide.
    Some people might be tempted to throw their SSDs like tech ninjas.
  • 0 Hide
    hrath , September 29, 2012 3:22 PM
    wow that looks epic :) 
  • 0 Hide
    g00b , September 30, 2012 7:52 PM
    I think the first page meant Flash instead of DDR.

    "... reviewed previously uses 25 nm synchronous NAND from Micron and Marvell's 88SS9174-BLD2 controller, while the M5 Pro employs 19 nm Toggle-mode --DDR-- from Toshiba and a more modern Marvell 88SS9187-BLD2 processor."
  • 2 Hide
    g00b , September 30, 2012 7:54 PM
    Ok. I'm partly an idiot :) .

    Toggle-Mode Double Data Rate NAND Flash
  • 0 Hide
    Anonymous , October 2, 2012 3:30 PM
    But how does the M5Pro compare with the Corsair Neutron GTX ?
  • 2 Hide
    Onihikage , October 2, 2012 7:19 PM
    Quote:
    I have read the article and have these thoughts on the subject.
    1) Why have mobo manufactures not placed an extra four memory slots on their mobo to handle this concept ?
    2) Why have RAM manufactures not teamed up with mobo manufactures to supply them with 256 GB memory sticks ?
    3) Has the industry, video and others, (movies, ads, special effects, cartoons, simulations, etc.) moved to this type of strategy to increase their productivity ?


    Luciferano has already commented, but I thought I'd make input on the first point. There are mobos with an extra four slots, enabling up to 64GB of RAM (such as the ASUS P9X79 PRO) but these are uncommon and expensive. Higher RAM capacities under today's technology will almost always require you to be using server hardware; the PC form factor just won't cut it.
Display more comments