Sign in with
Sign up | Sign in

Results: Transfer Diagrams

USB Thumb Drive Vs. Hard Drive
By

The transfer diagram of our IBM hard drive illustrates the characteristic step pattern that you will find on a majority of modern hard drives. However, most of the hard drives that are available today deliver similar read and write throughput, which isn’t the case for this older unit. Of course, 12 MB/s maximum read throughput seems pathetic today, as hard drives such as the Samsung Spinpoint F1 max out at roughly 120 MB/s. That’s 10x the performance at 100x the capacity.

Super Talent Pico C 8 GB

The Super Talent Pico C drive sticks to the promises on its manufacturer’s website: the tiny drive actually delivers around 30 MB/s and typically stays well above 25 MB/s. Write performance averages at 12 MB/s and is still faster than what you’d get from the 10 GB IBM DTTA hard drive. But that’s only one part of the story; I/O performance tells another one.

Display all 35 comments.
This thread is closed for comments
  • 0 Hide
    reasonablevoice , September 11, 2008 6:53 AM
    I am probably being a bit of pompous person but...

    "The development of flash memory has not only been quicker, but it has also been almost linear, meaning that capacities have at least doubled every year."

    To my layman's understanding, that doubling you were talking about describes an exponential rise in capacity, not a linear one. The graph even bears this out.
  • 0 Hide
    bonelyfish , September 11, 2008 7:32 AM
    What happened to harddisk after 2008 (see the chart)? What is minimum harddisk/USB capacity mean?
  • 2 Hide
    cangelini , September 11, 2008 8:23 AM
    reasonablevoiceI am probably being a bit of pompous person but..."The development of flash memory has not only been quicker, but it has also been almost linear, meaning that capacities have at least doubled every year."To my layman's understanding, that doubling you were talking about describes an exponential rise in capacity, not a linear one. The graph even bears this out.


    Reasonable, agreed. Based on the data there, it looks more like an exponential increase. Change made.
  • 0 Hide
    eodeo , September 11, 2008 1:44 PM
    very interesting article, thank you for posting it. very refreshing.
  • 2 Hide
    ceo_mr_man , September 11, 2008 2:22 PM
    Tom's is certainly not known for editorial excellence, but at least the articles are generally technically accurate. Frankly, that's fine - if I want flowery prose, I can read the Economist. This article, though, is especially poor. The "Capacity Development" chart, in addition to being a poorly formatted, generic Excel chart, is particularly weak.
    What in the world does it matter what the minimum capacity was of any of these drives? Even if it did matter, the chart is ridiculously inaccurate - 200gb is the minimum ATA size? Right.... how about those ten thousand new shrink wrapped 80gb drives I have in stock right now? Who cares, anyway? What a pointless data series to include!
    Then, the authors make extrapolations into the future which they present as being valid and factual data points, equal in relevance to real data points. When presenting guesses along with real data, the authors should make sure to highlight the points that are extrapolations. They should at least give those guesses asterisks or highlight the background of the chart to indicate that there’s an important difference in the data being presented. And when they extrapolate, even reasonably, about one set of data points, why wouldn't they also give reasonable guesses for the other data series? At a glance, this chart makes it look like flash drives are already almost equal to hard drives in capacity because the authors put their guesses about the future of flash drives on the chart but fail to do the same for traditional disks. If they posted reasonable guesses for hard drives (say, 50% per year, though I'd recommend a more thorough analysis of the long term trend to arrive at a better estimate) then they would have seen that in the same year they project flash to be at 1tb, hard drives will likely be at 5tb - hardly almost equal!
    Next, the authors both seem to have missed a simple error that seriously affects the visual impact of the graph. After flash is projected to hit 128gb next year, they show it at 512gb in 2010. 256gb is passed completely by. So by their own reasoning, the size of flash disks in 2010 is likely to be about half what they show on this chart.

    For an article that's not necessarily timely or relevant, it should at least have been a nice look at where things were ten years ago, where they are today, and where they might be in the future. A look at actual productivity or something like that might have been nice, in addition to a more thoughtful presentation of the data.

    I only criticize Tom’s because I love it, btw. This is a generally great site, but there’s always room for improvement.
  • -5 Hide
    ceo_mr_man , September 11, 2008 2:26 PM
    Tom's is not particularly known for editorial excellence, but at least the articles are generally technically accurate. That's fine - if I want flowery prose, I can read some essays or something. This article, though, is especially poor. The "Capacity Development" chart, in addition to being a poorly formatted, generic Excel chart, is particularly weak.
    What in the world does it matter what the minimum capacity was of any of these drives? Even if it did matter, the chart is ridiculously inaccurate - 200gb is the minimum ATA size? Right.... how about those ten thousand new shrink wrapped 80gb drives I have in stock right now? Who cares, anyway? What a pointless data series to include!
    Then, the authors make extrapolations into the future which they present as being valid and factual data points, equal in relevance to real data points. When presenting guesses along with real data, the authors should make sure to highlight the points that are extrapolations. They should at least give those guesses asterisks or highlight the background of the chart to indicate that there’s an important difference in the data being presented. And when they extrapolate, even reasonably, about one set of data points, why wouldn't they also give reasonable guesses for the other data series? At a glance, this chart makes it look like flash drives are already almost equal to hard drives in capacity because the authors put their guesses about the future of flash drives on the chart but fail to do the same for traditional disks. If they posted reasonable guesses for hard drives (say, 50% per year, though I'd recommend a more thorough analysis of the long term trend to arrive at a better estimate) then they would have seen that in the same year they project flash to be at 1tb, hard drives will likely be at 5tb - hardly almost equal!
    Next, the authors both seem to have missed a simple error that seriously affects the visual impact of the graph. After flash is projected to hit 128gb next year, they show it at 512gb in 2010. 256gb is passed completely by. So by their own reasoning, the size of flash disks in 2010 is likely to be about half what they show on this chart.

    For an article that's not necessarily timely or relevant, it should at least have been a nice look at where things were ten years ago, where they are today, and where they might be in the future. A look at actual productivity or something like that might have been nice, in addition to a more thoughtful presentation of the data.

    I only criticize Tom’s because I love it, btw. This is a generally great site, but there’s always room for improvement.
  • 1 Hide
    knickle , September 11, 2008 2:46 PM
    That's a good catch. If capacities double every year, then 2010 should reach 256GB, not 512. 1TB would be hit in 2012 (in theory).
  • 0 Hide
    eccentric909 , September 11, 2008 3:12 PM
    There are already 128GB SSDs on Newegg today (2008). Doubling every year would mean, 256GB in 2009, 512GB in 2010 and so on.
  • 1 Hide
    eccentric909 , September 11, 2008 3:14 PM
    Quote:
    a regular desk top flash drive at 128 gb is still going for 3000 bucks athte cheapest


    128GB SSD Patriot Flash Drives on Newegg run for $450 before mail-in rebate.
  • -1 Hide
    eklipz330 , September 11, 2008 3:14 PM
    they should have thrown in a velociraptor >.>

    KIDDING, I like this article, shows how far we've progressed in the technical field. YAY FOR HOO-MANS!
  • 2 Hide
    fritzair , September 11, 2008 3:56 PM
    I understand the relative merits of publishing a report such as this. "Tom" points to a rather swift increase and CHANGE in storage media. One poster said he didn't think the solid state device would ever supplant the spinning disc. What "tom" may be hinting to is that day is coming and it may not be so far away. As USB 3.0 is distributed more widely will we see a resulting skip upward in the functionality of new USB drives? Will Intel ever allow for a RAID configuration of multiple large thumb drives?
    I could imagine a USB 3.0 compliant solid state storage device on a mother board (similar to EPROM)with entire OS(s) as the startup partition.
    Imagine the savings in electrical costs to an organization with Flash drives as the primary storage medium.
  • 0 Hide
    elpresidente2075 , September 11, 2008 4:14 PM
    I wonder how much that hard disk cost when it was new, compared to the same for that flash drive...
  • 0 Hide
    Anonymous , September 11, 2008 4:37 PM
    First, Good Article!

    Saddly enough, I have some 60GB ATA drives running in a server now (dual 500Mhz machine...the 2 ATA's were an upgrade from the old SCSI array of a number of 8GB drives) that stores 2 Oracle databases. Technology that is cheap and could give a speed boost to such old monstrosities is always a plus :)  Finding out pursuing it would be a waste of my time is also a benefit.

    Something else that I have wondered about and thought Tom's might be willing to try... a caching controller with a standard HDD and a SSDD in RAID 1. Best of both worlds or worst waste of money ever?
  • 1 Hide
    gwolfman , September 11, 2008 4:37 PM
    This bring about an interesting question if someone could please answer:

    According to this article, it's better to run a virtual machine from a 1.8" USB HDD (are there smaller USB storage devices that use a rotating platter?) than a "high performance" (ocz rally is pretty good, right) usb flash drive?
  • 0 Hide
    JonathanDeane , September 11, 2008 4:38 PM
    elpresidente2075I wonder how much that hard disk cost when it was new, compared to the same for that flash drive...


    Hahahaha so true... back then a decent PC was 3-5 thousand dollars, not even a top of the line machine. Now you can get a PC thats decent for less then 1 thousand and 2 thousand is getting near top of the line with out getting into extreme territory which will always be more expensive.
  • 1 Hide
    knickle , September 11, 2008 4:42 PM
    Eccentric909There are already 128GB SSDs on Newegg today (2008). Doubling every year would mean, 256GB in 2009, 512GB in 2010 and so on.

    Of course SSD has a higher capacity (and is much faster too). But the article and the chart have nothing to do with SSD. It to pertains specifically to USB drives (Flash Sticks, thumb drives, etc.), not SSD. Therefore it's wrong.
  • 1 Hide
    Anonymous , September 11, 2008 4:46 PM
    Given the nature of growth (exponential), a semi-log plot would have been a better representation...you could have gone back in time farther than you did in your graph, without the ridiculous distortion at far left =)
  • 1 Hide
    snarfies1 , September 11, 2008 7:05 PM
    New technology outperforms 10-year-old technology! Film at 11!
  • 2 Hide
    Anonymous , September 11, 2008 8:11 PM
    The author remarks that MLC flash reads are terrible, but I believe that he failed to account for moderately sophisticated drivers such as that in the 80 gig Intel MLC flash drive reviewed just a day or two ago, which showed competitive write rates.
  • 0 Hide
    knickle , September 11, 2008 9:37 PM
    You're comparing apples to oranges. The thumb drive is going to be a single chip, where an SSD drive that you are refering to has 10 channels of 20 x 4GB chips.

    Taken from the article you are refering to:
    "The current generation of 50 nm NAND flash stores 32 Gbit (4 GB) per chip. If you now use 20 of them (two per channel times 10 channels) you’ll reach exactly 80 GB. If you distribute reads and write across 10 channels it is obvious that performance will scale beautifully."

    The article is attempting to compair something on your keychain with a hard drive from 10 years ago. It is not to compare a 10 year old HD with the new SSDs. The SSDs will blow it out of the water, and then some.
Display more comments