Firewire and USB Performance

After looking at many options for Firewire and USB testing, we finally determined that an external USB 2.0, Firewire 400, and Firewire 800 hard disk might be a sensible way to look at USB and Firewire throughput.

Our first efforts at testing with an IDE or SATA drive as the "server" yielded very inconsistent results, since Windows XP sets up cache schemes to improve performance. Finally, we decided to try a RAM disk as our "server", since memory removed almost all overhead from the serving end. We also managed to turn off disk caching on the USB and Firewire side by setting up the drives for "quick disconnect" and our results were then consistent over many test runs.

We used 2GB of fast 3-2-2-4 system memory set up as a 450MB RAM disk and 1550MB of system memory. Our standard file is the SPECviewPerf install file, which is 432,533,504 bytes (412.4961MB). After copying this file to our RAM disk, we measured the time for writing from the RAM disk to our external USB 2.0, Firewire 400, or Firewire 800 drive using a Windows timing program written for AnandTech by our own Jason Clark. The copy times in seconds were then converted into Megabits per second (Mb) to provide a convenient means of comparing throughput. Higher Rates, therefore, mean better performance.

Firewire and USB Performance

Possibly the most striking finding in our Firewire and USB throughput tests is the performance of an external hard drive connected to Firewire 800. If you wonder why Firewire 800 matters, just look at the data. Our benchmarks show Firewire 800 is up to 46% faster than a drive connected to the more common Firewire 400, and about 29% faster than USB 2.0.

Our test is just one of many throughput tests, but in this benchmark, it is clear that the VIA Firewire 400 chip is faster than TI's 1394a chip. The NVIDIA nForce4 USB 2.0 controller is slightly faster than Intel's solution. Unlike the other Royal member boards, the Gigabyte 8N SLI Quad Royal does not offer a Firewire 800 option. This is a mistake in our opinion and hopefully, Gigabyte will address this before mass production begins.

Disk Controller Performance Ethernet Performance
Comments Locked

44 Comments

View All Comments

  • DrMrLordX - Thursday, October 13, 2005 - link

    Fine, I'll retract my statement, at least partially. I wasn't reading the statement carefully enough.

    Having looked into the newer 3D1-68GT, it seems to be a more solid product than the original 3D1 card based on 6600s. The original seemed to serve no purpose whatsoever.

  • Calin - Thursday, October 13, 2005 - link

    They made it an Intel board assuming that the more "corporate-oriented" users prefer multiple monitors. I don't know about current performance, but in the recent past, Intel processors smoked the Athlon64 at things like Photoshop. And introduction of dual core processors at prices much lower than AMD's dual core could coax someone into buying such a board.
    I agree that most every normal person would be happy with four processors (powered by two cards), however I remember cases (in Linux) when OpenGL performace fell at half when enabling 2 monitor support on a single video card. This is driving a single monitor, not two. Driving two monitors, it fell even lower.
    So, for every person that WANTS (not that it really really would need) four monitor output from four video cards, this looks like the best choice
  • trooper11 - Thursday, October 13, 2005 - link

    quote:

    And introduction of dual core processors at prices much lower than AMD's dual core could coax someone into buying such a board


    I kind of doubt that since the cost in video equipment does not make this a low cost solution. if a company is willing to shell out for that, they would be willing to shell out for the best in workstation performance, which just happens to be the X2s
  • ElJefe - Wednesday, October 12, 2005 - link

    ever wonder what crack they were smoking though making it an Intel board?

    if you read about modders and gamers , almost 80%+ market share for DIY builders use AMD.

    this board is a waste of technology.

    still cool though.
  • Gary Key - Thursday, October 13, 2005 - link

    Hi,

    The ability to produce this board was due to Nvidia's decision to use a HyperTransport link for the Intel SLI chipset due to the need to have an on-chip memory controller. While it would be feasible to complete a AMD version of the board, the engineering time and product cost would not be acceptable. While I will agree with everyone that the current AMD processor line up offers significantly more performance than Intels, the actual day to day real life experience with both systems is not readily apparent to most people. In fact, I have had people play on my FX55 machine and 840EE machine and nobody could decide clearly which system had the AMD64 in it without benchmarks. This was at both 1280x1024 and 1600x1200 resolutions. While I personally favor AMD for most performance oriented setups, there are some people that still want Intel. After not having an Intel based machine for the last two plus years I have to admit is not as bad as most people make it out to be.
  • Johnmcl7 - Wednesday, October 12, 2005 - link

    Whether you like it or not, the 3D1 was an innovative product, it's not childspplay to stuff both cores together and develop the motherboard support for it.

    John
  • Viper20220k - Wednesday, October 12, 2005 - link

    Yeah, what is up with that.. I would sure like to know also.
  • Wesley Fink - Wednesday, October 12, 2005 - link

    The pictures of the 10-monitor display were supplied, but Gary did hook up every monitor we could, which was 8 if I recall, to test the outputs. To test 10, we needed two more Rev. 2 3D1 cards - our extra pair were Rev. 1 cards - which couldn't be here in time for a review.

    We did verify the ability of the individual 3D1 cards to do what Gigabyte claimed, so there is no reason at all to doubt the 10 claim. One of the key Engineers at Gigabyte works exclusively with AT and THG. All sites use some pictures and diagrams from press kits to save time, but we perform and report our own test results and analysis.

    Yes, we dis ALL of the testing ourselves. Our review took longer because we did much more extensive testing of the board, including quite a bit of overclocking tests to make sure the nVidia dual-core issue we reported in our last Intel SLI review is now fixed in this chipset.

    Gary spent countless hours sniffing out the good and the not so good on this board. We also found the OC capabilities of the shipping BIOS not too exciting, and we wanted to bring you the much improved OC results from the revised BIOS.
  • johnsonx - Wednesday, October 12, 2005 - link

    Wesley,

    I don't think most of your readers actually thought what the subject line of this thread implies. There are always a few who like to throw stones of course.

    In my read of the THG article a few days ago, I found myself thinking that the 10-display shot was from Gigabyte, as they had no detail shots of the display control panel for 10 monitors; nine was the most they were able to get working.

    Like you, I have little doubt that 10 displays will in fact work with this board, but the 9th and 10th would have to come from either a PCI card or a PCIe card running in a x1 slot. Even x1 PCIe is faster than crusty old PCI, but it's still hardly ideal. It'd be nice if 3D1 cards could be coaxed into working in x8 slots (so that'd be 4 PCIe lanes per core - still plenty), as then you could theoretically have 4 3D1 cards for 16(!) displays.

    Thanks for the information on how you did the review testing.

    Regards,

    Dave
  • AmberClad - Wednesday, October 12, 2005 - link

    April Fool's Day already?!

Log in

Don't have an account? Sign up now