Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie

Marvell 9128 SATA 3 controller benchmarks.

Options
  • 12-11-2011 5:21pm
    #1
    Registered Users Posts: 3,532 ✭✭✭


    Hi all.

    I will be benchmarking a new SSD this week on this controller and Intel's own SATA2 ports on a P55 system.

    There seems to be mixed opinions on this controller in terms of performance. There are a lot of posts on the web saying that it performs worse than Intel's SATA2 ports in some benchmarks, however a lot of these are out of date. There are some that say they have no problems with the Marvell controller, with little to back them up. There is broad agreement that using the 9128 for RAID results in bandwidth limitations.

    So to satisfy my own curiosity and hopefully some of yours, I will perform a number of benchmarks using up to date BIOS and drivers - using both the Marvell driver and MSAHCI driver. Just to see where the differences are, in terms of random and sequential speeds. (Obviously the sequential read/writes on Intel's ports will be limited.)

    I plan on running some tests on AS SSD and ATTO, if you have any suggestion on what else you'd might want to see, let me know.

    Edit: Just a note on TRIM, I don't know if I will bother testing it or not. Rumours indicate that with the latest BIOS/drivers the marvel controller will pass trim. However the garbage collection on new drives are deemed to be good enough in comparison. Having said that, I will think I will start with the Intel ports, let the drive rest (so trim can pass) then secure erase the drive so that I will have a fresh drive for testing on the marvel ports.
    Any problems with this methodology or testing, please let me know.


Comments

  • Registered Users Posts: 4,983 ✭✭✭Tea_Bag


    subbing to the thread..


  • Registered Users Posts: 3,532 ✭✭✭Unregistered.


    Picked up SSD today! Crucial M4 128GB. Looks like it's shipped with the latest firmware (009).


  • Registered Users Posts: 4,983 ✭✭✭Tea_Bag


    when you're done benchmarking ill PM you my address cause reviewers never keep their hardware :P


  • Registered Users Posts: 3,532 ✭✭✭Unregistered.


    So here's some quick benches on the chipset's SATA2 ports. MSAHCI driver vs Intel's RTS driver. Performed on a fresh install of windows 7, SSD as a secondary disk with nothing on it. First runs of AS-SSD and ATTO. Restart + some idle time between driver change.

    MS:
    23u6as8.png344yupw.pngdiskmarkms1.png



    Intel:
    245eh6b.pnga5di00.pngdiskmarkintl1.png

    Intel is clearly on top. Tomorrow will be the big one, when i test the marvel controller.

    Edit: Added Crystal Disk Mark bench. Mixed results with CDM, the MS driver providing higher sequential speeds and random speeds at larger block sizes.


  • Registered Users Posts: 3,532 ✭✭✭Unregistered.


    Tea_Bag wrote: »
    when you're done benchmarking ill PM you my address cause reviewers never keep their hardware :P

    Nice try, but I'm not reviewing ;)


  • Advertisement
  • Registered Users Posts: 3,532 ✭✭✭Unregistered.


    Results with drive connected to the Marvell SATA3 port.

    MS driver:

    sata3asssdms1.pngsata3attoms1.pngsata3diskmarkms1.png


    Marvell driver:

    sata3asssdmarv1.png
    sata3attomarv.pngsata3diskmarkmarv1.png


    Interesting results...
    • Sequential Reads are much better on the SATA3 port - as expected
    • AS-SSD shows intel to have perfectly good sequential write speed and CrystalDiskMArk shows it to be the worst out of them all by a fair bit.
    • random writes at small block sizes are better on the SATA3 port, but reads are much slower.
    • The marvel driver shows desperate 4K-64thread read speed.
    • MS and Marvel are pertty much on par except for the the above and 4K at large queue dept pn CDM - wonder why AS-SSD and CDM don't corroborate?


Advertisement