Power Consumption

The m4 has pretty good idle and sequential power characteristics. It's only under high queue depth random writes that the m4 looks a little power hungry, and that's only because the transfer rates we're talking about are so very high.

Idle Power—Idle at Desktop

Load Power—128KB Sequential Write

Load Power—4KB Random Write, QD=32

AnandTech Storage Bench 2010 Final Words
POST A COMMENT

103 Comments

View All Comments

  • 7Enigma - Friday, April 01, 2011 - link

    No. You do not. The TRIM command will either run at a set interval or you can do it manually depending on the drive maker (INTEL for instance has a Toolbox that I use once a week or so), and this will return the drive to near-new performance. It's only OS' and drives that do not support TRIM that a complete format (or another method such as copying a large amount of data to the drive and erasing) would be required for this. Reply
  • eamon - Sunday, April 03, 2011 - link

    You don't even need that - windows will issue the trim command automatically after a file delete if a block becomes available. You don't need any special driver or special tool; normal use will trigger trims on win7. Reply
  • danbi - Wednesday, April 06, 2011 - link

    The typical "OS drive" is "write rarely, read many times". There might be occasional writes to the OS drive, in form of log files, new drivers etc. You rarely, if ever delete files from the "OS drive". Perhaps only when you uninstall software.

    How would TRIM (supporting OS) ever help in such scenario?

    One thing I miss form the "TRIM performance" is what happens when you overwrite already existing files -- pretty common task, by the way.

    Another thing, performance with COW file systems, such as ZFS. These file systems will never overwrite data (except metadata) and most of the time writes will be "sequential".

    By the way, another missing performance metric is what each drive considers "sequential write". How much KB data in one I/O operations sequential? Is this common with all drives?
    Reply
  • faster - Thursday, March 31, 2011 - link

    The egg has the RevoDrive 240 GB for $570. The 3rd gen Intel drive is $614 at 250 GB. The new M4 drive is $599?. Price competition in this price point would warrant a comparison in performance. I would like to see the X2 skew the chart to illustrate performance per dollar invested. I understand that the X2 is a PCIE self contained RAID card, but it is a bootable card making it a hard drive competitor. No matter how you feel about it, it would be an interesting comparison from an economic viewpoint. Reply
  • flexcore - Friday, April 01, 2011 - link

    Good review! The problem is there are so many different versions of each drive. They change performance characteristics as the size of the drive changes. (Not even getting into OCZ's recent crimes against consumers)

    I like what MilwaukeeMike brought up about different types of users and how different drives seem to be suited better for different usage patterns. As a AMD user I would also appreciate a review on how these drives perform on AMD platform vs Intel. Then we have TRIM. This is becoming more and more of the normal operational mode, but not always, and what about in RAID configuration.

    WOW, these are not the only questions that I continually have been hearing asked on forums around the web. This is a lot of information and work but, you are a leader in SSD reviews. I applaud your efforts ANAND and want to thank you for doing the work you do. I think you are heading in the right direction with your own storage benchmarks, but thats only part of what consumers need. More real world usage to be able discern the actual advances from each new generatin of drive is important also. I look forward to reading more of your insights into where we are and where we are heading with SSD technology.
    Reply
  • 7Enigma - Friday, April 01, 2011 - link

    What we need is some understanding of what is really needed by a SSD. We have all these different benchmarks that are combining IMO way too many tasks at once that a human just cannot perform. Yes it's great to tease out a pretty graph, but what I think most of us want to really see is what is the real-world difference.

    Things like the Gaming benchmark where there is practically NO difference between current and last-gen drives (hell even the G2 is pretty close) are very important to me so I don't get the upgrade bug for something I really have no major benefit from. Vantage as well.

    Anand's said it for a while but there is a huge difference between SSD and mechanical drive, once you upgrade to SSD, however, the improvements for most human workloads (ie non benchmarks) is slim.

    I'd love a mini article that would track 4-5 different people's usage models: gamer, web surfer/facebook/iTunes average joe, encoder, and whatever other major "type" of person that frequents Anandtech (it should be biased towards the readership IMO), and then create a benchmark based on that.

    We don't need 24 programs opening simultaneously while downloading a torrent and doing a virus scan. What we do need to see is if there is any tangible benefit to 6Gbps SATA over 3 in normal daily use and G1 to Vertex3. If not the numbers are great and if buying a SSD for the first time or building a system OK, but the need to upgrade from an existing SSD is just not there.

    Great article btw!
    Reply
  • X-Nemesis - Friday, April 01, 2011 - link

    This is what I think as well...it does not seem like there is really any real world benefit to upgrade an older generation SSD to the new 6Gbps offerings. The only reason to upgrade would be size at a much cheaper price point. Reply
  • Nicolas Pillot - Friday, April 01, 2011 - link

    I'm quite new to the field, and looking at the charts i have a plainly simple (but stupid ?) question : how is it that the read speed are lower than the write speed ?! Reply
  • 7Enigma - Friday, April 01, 2011 - link

    Make sure you are comparing apples to apples. Sequential read/write will be much higher then random read/write so if you are comparing a random read to a sequential write you will see a discrepancy. Reply
  • iwodo - Friday, April 01, 2011 - link

    Just thought Anand you should compare all the 40 - 80GB Range SSD. Since it is very likely that we buy the lower capacity drive then the expensive but also slightly faster SSD. Reply

Log in

Don't have an account? Sign up now