Sign in with
Sign up | Sign in

Not All USB 3.0 Implementations Are Created Equal

Not All USB 3.0 Implementations Are Created Equal
By

Despite the fact that today's USB 3.0 products center on the same NEC controller, we compared a handful of different USB 3.0 drives and found performance to range from 113 to 173 MB/s, depending on the implementation used. Should you be worried?

You might think that the performance of one USB 3.0-equipped product would be (or at least should be) the same as another. If so, you'd be wrong. We found that five different components, all based on the NEC controller, deliver different performance.

USB 3.0 remains a difficult topic. While the benefits are undisputed—it's up to 10x faster than USB 2.0, not necessarily more expensive in volume quantities, and backward compatible with earlier USB versions—it will still take a while for USB 3.0 to go mainstream. We've already looked at USB 3.0 thumb drives, USB 3.0 enclosures for 2.5” hard drives, various motherboard implementations that support USB 3.0, and a few USB 3.0 external storage solutions. As expected, hard drives (particularly SSDs) can now operate without bottlenecks, and the infrastructure is ready for throughput of 300 to 400 MB/s.

Intel Hesitating

One of the key market forces won’t implement a USB 3.0 controller into its next chipset generation. The successor to the P55 family is code-named Cougar Point (P67), and it supports a slightly modified socket format: LGA 1155, which uses one pin less than today's LGA 1156 and hence is incompatible. Cougar Point will support 14 USB 2.0 ports, but it lacks SuperSpeed USB 3.0 support. 

While this decision is a mystery to many users, it does make economic sense. Intel had problems with its USB 1.1 implementation in the late '90s, and so waited until the ICH4 in 2002 to implement USB 2.0. Since the introduction of USB 3.0 root controllers represents a significant design change and the market is not going to switch from USB 2.0 to 3.0 in only a few months, Intel is playing a conservative game this time around, even though most of us would prefer a more aggressive strategy.

New Controllers Coming

There are a few more companies working on USB 3.0 controller designs, namely ASMedia (Asus), Texas Instruments, and VIA. We believe that all three will hit the market before the end of the year, and we hope that the different implementations will enable more cost competition in this space. Right now, NEC is the only supplier, and without any competition, NEC is keeping the controller cost too high for mainstream adoption. As a result, lower-cost motherboards will not be equipped with USB 3.0 hardware anytime soon.

Options?

Today, you have two options for jumping on the USB 3.0 bandwagon: either get a motherboard that includes a USB 3.0 controller with two ports, or look for an add-on card that utilizes the same NEC PD720200. There are also ExpressCard products for notebooks based on NEC's chip. Keep in mind that all implementations require x1 PCI Express slots, per the PCI Express 2.0 specification. This means that high-speed USB 3.0 solutions can actually be bottlenecked if the controllers/cards run in PCIe 1.1 slots, which are limited to 250 MB/s upstream and downstream. This applies to all Intel platforms in which additional PCIe connectivity is operated through the southbridge and to all older PCI Express systems. While we don’t see bottleneck issues coming into play today, this is something to keep in mind for future products.

Display 29 Comments.
Top Comments
  • 18 Hide
    Anonymous , August 26, 2010 7:10 AM
    Where is the 600MB/s that USB 3.0 can theoretically deliver? I would like to see a test that saturates the bus completely, to find out how much we can get out of it. Like in USB 2.0 which has theoretical bandwidth of 60MB/s and achieves only 30MB/s. Or is this it? The realistic bandwidth of USB 3.0 is only 180MB/s???
Other Comments
  • 0 Hide
    liquidsnake718 , August 26, 2010 6:20 AM
    So Intel is also coming out with another mainstream board in the P67? I cant believe after almost a year they still havent decided to impliment the usb 3.0 controller on their vanilla boards. What about the supposed X68 board? Will it have more than 2 usb3.0 outputs and one possible on the front?
  • 2 Hide
    Anonymous , August 26, 2010 6:40 AM
    Great to see you guys keeping us up to date on the current state of USB 3.0 implementations. I was a little disappointed that you didn't test the Asus U3S6 USB 3.0 & SATA 6Gb/s Add-on card. http://usa.asus.com/product.aspx?P_ID=lGYmelQ8mJvPtYTv I figured since it uses a PCI-E x4 slot it would have the potential to offer excellent performance.
  • 18 Hide
    Anonymous , August 26, 2010 7:10 AM
    Where is the 600MB/s that USB 3.0 can theoretically deliver? I would like to see a test that saturates the bus completely, to find out how much we can get out of it. Like in USB 2.0 which has theoretical bandwidth of 60MB/s and achieves only 30MB/s. Or is this it? The realistic bandwidth of USB 3.0 is only 180MB/s???
  • 8 Hide
    Darkerson , August 26, 2010 7:10 AM
    I would have liked to have seen the Asrock Deluxe 3 or Extreme 3 on here, as they too have on-board USB3 (4 ports if im not mistaken) to see how it stacks up to the others listed, but it was a nice and informative article none-the-less.
  • 1 Hide
    pirateboy , August 26, 2010 7:35 AM
    darkersonI would have liked to have seen the Asrock Deluxe 3 or Extreme 3 on here, as they too has on-board USB3 (4 ports if im not mistaken) to see how it stacks up to the others listed, but it was a nice and informative article none-the-less.


    Yep, an on-board usb3 shootout article (oldskool) would be nice
  • -4 Hide
    Fetal , August 26, 2010 7:54 AM
    Intel is doing right! Strategic decisions fare much better than 'Vengeful' Decisions. (plying Worms Reloaded these days :p )
  • 5 Hide
    nitrium , August 26, 2010 8:13 AM
    Shame there isn't a USB 2.0 bar in the graphs to use as a reference.
  • 2 Hide
    Mante , August 26, 2010 8:32 AM
    Asus U3S6 USB 3.0 & SATA 6Gb/s Add-on card. http://usa.asus.com/product.aspx?P_ID=lGYmelQ8mJvPtYTv
    id like to see it in the chars.
    Nice review.
  • 2 Hide
    nevertell , August 26, 2010 8:43 AM
    I would have loved to see UD4, since that is the one I was planning on buying.
  • -8 Hide
    vic20 , August 26, 2010 9:06 AM
    I'm not too thrilled at the thought of a bunch of comapnies starting to produce USB 3 chips. I'd rather just have one even if it costs more. At least there would be some consistancy and we would not have to sit down and research what drive or enclosure to buy.

    USB 2 drives seem to average 25MBs and it's most likely due to all the crappy cheap bridge chips everyone makes. Whats the point of a standard if nobody enforces it? Having the USB2 logo on a box only means it works, not that it works up to spec.

    Just look at whats happened to wifi. I never see speeds over 54Mbit even on N because of all the different manufacturers and incompatiblities. You can't choose who makes the wifi chip in your laptop (Atheros, Intel, Realtek, Raylink, etc etc) so buying an Xtreme N router for example doesn't do any good at all. Not too mention the drop-offs, disconnects, and range problems you run into by mixing brands of wireless chipsets.
  • -5 Hide
    nforce4max , August 26, 2010 12:30 PM
    vic20I'm not too thrilled at the thought of a bunch of comapnies starting to produce USB 3 chips. I'd rather just have one even if it costs more. At least there would be some consistancy and we would not have to sit down and research what drive or enclosure to buy.USB 2 drives seem to average 25MBs and it's most likely due to all the crappy cheap bridge chips everyone makes. Whats the point of a standard if nobody enforces it? Having the USB2 logo on a box only means it works, not that it works up to spec.Just look at whats happened to wifi. I never see speeds over 54Mbit even on N because of all the different manufacturers and incompatiblities. You can't choose who makes the wifi chip in your laptop (Atheros, Intel, Realtek, Raylink, etc etc) so buying an Xtreme N router for example doesn't do any good at all. Not too mention the drop-offs, disconnects, and range problems you run into by mixing brands of wireless chipsets.


    I agree. +1 to balance out the prick....

  • 4 Hide
    tokenz , August 26, 2010 1:18 PM
    I am not to worried about usb3.0 right now as I have esata. But I would still like to see it implemented. The real kicker is another socket for intel. Hey intel. I am going to go with amd this round. That way I wont have to change a $250 motherboard every two years. I am still using an abit ab9 pro with a 775 quad core.
  • 1 Hide
    andune , August 26, 2010 1:47 PM
    Would the same limitations be true of an add-on card on X58 chipset? I was under the [mis?]understanding that the additional PCIe slots were also 2.0 compliant and not limited to just the x16 physical slots as on the 1156 platforms...
  • 2 Hide
    dgingeri , August 26, 2010 3:00 PM
    none of this applies to me because I'm running an Asus P6T (x58 chipset) while none of your test boards use an x58.

    I'd like to see 2 more things on this test: an x58 motherboard using a tertiary expansion slot (like a x4 slot on a P6T or a x8 slot on another board in place of a third video card) and the Asus x4 card with both USB3 and SATA 6.0Gb controllers. Does the x4 interface with the switch affect performance? I think it is also especially relevant considering the Asus board is barely more expensive than the USB3 cars without a 6Gb SATA controller and only a x1 interface.

    How would those things fare compared to these solutions in this article?
  • 1 Hide
    awaken688 , August 26, 2010 3:23 PM
    Good article. USB3.0 is definitely a big deal for me. I can't wait to start using it download my RAW images. USB3.0 would fit the bill nicely.
  • 3 Hide
    cyclocommuter , August 26, 2010 3:26 PM
    Why not benchmark the Gigabyte P55A-UD3P or P55A-UD4P motherboards instead of the UD6P? Much more folks use these than the higher end P55s.
  • 2 Hide
    akorzan , August 26, 2010 4:16 PM
    For the majority of folks running p55 systems, we don't want to give up pci express 2.0 lanes from the video card. So why not test the Asus U3S6 which uses 4 pci express lanes. I would love to see it's performance running on 1.1 speeds. Not to mention it's one of the cheapest implementations and comes with Sata 3 (Though the USB would be tested without the SATA being used).
  • 2 Hide
    niknikktm , August 26, 2010 5:03 PM
    Why no X58 motherboards in this evaluation? Would have been nice to see one of the Gigabyte GA-X58 series MB's in this comparison. The UD3R is a TOMS recommended MB.
  • 0 Hide
    JohnnyLucky , August 26, 2010 7:38 PM
    Interesting article. Looks like full impementation is going to take a while longer.
  • 1 Hide
    jakesbuddy , August 27, 2010 1:08 AM
    Back in March I needed an external hard drive, and because I knew I was going to be transferring lots of data I bought the 1TB WD USB 3.0 external hard drive because it was USB 3.0 and it came with the add-in card. It is amazing. Inside the enclosure is a WD Black 1TB hard drive, and transferring files between it and my internal hard drives move at internal SATA speeds. I can't remember the exact numbers, but when it was nearly empty, I was getting read speeds over 110 MB/s, so according to this article, the only real difference is longer access times, which I hadn't really noticed due to the joy of blazing fast USB transfer speeds.
Display more comments
React To This Article