by Andrey Kuznetcov
10/30/2003 | 01:21 PM
USB 2.0 interface has become in inalienable part of the users’ life. Numerous peripheral devices supporting this interface have significantly increased the functionality of our computers. The actual effect resulting from the shift to USB 2.0 (from USB 1.1) pushed forward the development of new technical solutions taking advantage of Hi-Speed USB potential. The best example here will be the ever growing number of external optical storage devices supporting data recording as well.
The theoretical bandwidth of the USB 2.0 interface reaches 480Mbit/s (60MB/s), which is a pretty impressive number I should say. Much more impressive than the previous maximum of 12Mbit/s (1.5MB/s). However, far not everything is so simple. In reality, the contemporary USB 2.0 controllers are simply unable to ensure maximum theoretical performance, that they are supposed to show. Though it is also no calamity yet. The second part of the problem is that the actual controller performance also depends a lot on the manufacturer. It is true for the integrated mainboard USB 2.0 controllers as well as for the ones integrated into the peripheral devices to be connected to them. The final performance will depend a lot on the way these devices get along.
In reality there might be a situation when a high-performance external device connected via the USB 2.0 interface will not prove up to your expectations because the integrated controller in your system will not be able to ensure sufficient bandwidth. When you buy a printer, scanner or any other device complying with the Hi-Speed USB specification, you will not be able to make any hardware changes to it. However, you could make it work at its maximum speed if you select the most effective USB 2.0 controller for your system. Therefore, you will probably ask me: what chip should the controller be based on to ensure the maximum performance for my system? This is exactly what we are going to talk about today.
For our comparative investigation we selected three most popular chips. Two of them were used in USB 2.0 PCI adapters, and in the third case we will talk about a USB 2.0 controller of the ICH5 South Bridge.
Tekram controller card is based on NEC 7200100AGM chip. It features three external and two internal USB 2.0 ports. The controller supports PCI 2.2 interface. Besides the controller card, the retail package also includes a user’s guide, a CD-disk with the drivers and electronic documentation, and a USB 2.0 cable.
This adapter is based on VIA VT6202 chip. There are two external USB 2.0 ports laid out on it. It supports PCI interface. I do not know who is the manufacturer of the card, I assume that it is most likely to be one of the Chinese or Taiwanese companies. The controller is shipped with a driver CD-disk.
Intel 865PE chipset features ICH5 South Bridge, which is going to be the topic of our today’s discussion, because it is exactly the one responsible for the implementation of high-speed USB 2.0 bus.
Of course, we cannot test the South Bridge as is, that is why we took a mainboard based on the above mentioned chipset and tested it.
In order to test the controllers performance and to estimate which of them is the fastest, we used the following software:
As a testbed for our session we used the following system:
We used the following drivers for the controllers:
In order to evaluate the efficiency of the USB 2.0 controllers installed into the mainboard PCI slot and those integrated into the chipset South Bridge, we used an external StorageKingdom EasyHardDisk USB 2.0 20GB and Western Digital WD2500JB HDD installed into Gembird Flexi Drive 2.0 portable chassis.
Before we pass over to testing our USB 2.0 controllers with this hard disk drive, I have to say a few words about the limitations imposed by the Gembird Flexi Drive 2.0 chassis. It turned out that the hard disk drive installed into it cannot be formatted for more than 128GB. Moreover, the documentation for Gembird Flexi Drive 2.0 declared the interface bandwidth of only 40MB/s instead of the 60MB/s theoretically expected from the USB 2.0 interface. All these limitations seem to be implemented on the hardware level, and should be determined by the USB 2.0-IDE bridge used in this chassis.
The HDTach program allowed us to measure the Read Burst Speed of the interface. The advantages of the USB 2.0 bus integrated into the chipset are more than evident here: it leaves not a single chance to the competitors. NEC and VIA controllers work considerably slower, though the NEC one is a little bit faster than VIA.
I cannot say that the integrated controller performed better due to wider access bus to the memory subsystem (as we remember, the USB 2.0 controller of the i865PE chipset is connected directly to the memory controller avoiding the PCI bus). The maximum data transfer rate it achieved made only 30MB/s which is considerably lower than the PCI bus bandwidth of 133MB/s.
The same situation can be observed in IOMark test. All obtained results correlate perfectly well with the performance rates from the previous test. Again the indisputable leadership belongs to the integrated USB 2.0 controller of the Intel’s ICH5 South Bridge: the HDD connected to it appeared the fastest. The second prize is again won by NEC 7200100AGM controller, which is pretty far behind the leader. And the last one of the three appeared VIA VT6202.
Unlike the previous drive, this HDD allowed running all the tests we had planned, because there were no problems with formatting it as desired (FAT32).
The data transfer rates measured in HDTach program once again prove the indisputable triumph of the ICH5 chip from Intel. The results appeared even higher than in case of WD250JB HDD, it is probably because of the USB 2.0-IDE controller used in StorageKingdom EasyHardDisk.
Again we can see a noticeable gap between the leader and two other controllers. The NEC 7200100AGM again looks more attractive to us than the rival based on VIA chip.
The results of IOMark test once again confirm our verdict: the integrated solution in Intel ICH5 looks much more attractive than the competitors. All the numbers prove this statement. As for the remaining two competitors, they seem to be in absolutely the same situation as in the previous tests. Again NEC 7200100AGM is notably faster than the stable third prize winner from VIA.
We used FC-Test utility, which is described in detail in our article called X-bit labs Presents: FC-Test for Hard Disk Drives, to create, read, and copy five standard file sets from one HDD folder to another. The read speed results clearly indicate the performance difference. As always, the best performance belongs to the integrated solution.
Intel ICH5 also looks a little better than the competitors during file creation. In fact, the overall situation remained the same, too. The other two controllers retained their positions in this test: NEC came the second, and VIA – the third.
The results of SequentialRead and SequentialWrite patterns once again prove the efficiency of Intel’s integrated solution. Both benchmarks demonstrate its evident advantage over the competitors. The chip from NEC again works faster than the solution based on VIA VT6202. As you can see, some errors popped up as a result of heavy (100%) CPU utilization. Is this a peculiarity of controller drivers for VIA VT6202 and NEC 7200100AGM based solutions?
The classical Ziff Davis test once again demonstrates the superiority of the integrated USB 2.0 bus, although the advantage is not so evident this time. Business Disk WinMark results of ICH5 are not the best ones, as NEC7200100AGM manages to outperform the almost eternal leader here. However, in High-End Disk WinMark everything returns to its usual course, and ICH5 resumes its leadership. VIA VT6202 controller again lags behind the rivals.
Well, you may go farther and fare worse. This is exactly what the whole USB 2.0 controller test session points at. The USB 2.0 bus integrated into the chipset South Bridge appears the fastest, according to our tests. This way, if you are getting an Intel based mainboard, you will automatically get the opportunity to work with external USB 2.0 devices at the highest possible speed.
The situation is a little bit more complicated for mainboards and adapters using NEC 7200100AGM and VIA VT6202 chips. They feature considerably lower practical maximum speed and in some cases it may even tell on the performance of the external devices you are using. I am also talking about those cases when you are going to connect a really high-performance device to your USB 2.0 port, and then suddenly its performance gets limited by the bus bandwidth. If the device simply supports Hi-Speed USB, doesn’t boast any extreme performance and doesn’t require intensive data transfer, NEC 7200100AGM and VIA VT6202 will not cause any performance drops.
The slowest one in our comparison appeared the controller based on VIA VT6202 chip. Although I am pretty sure that the developer is not happy with this either. Actually, the new VT6212 chip should completely cure or at least partially eliminate the huge lag in the interface bandwidth. This could be really nice, however, we will be able to state that only after we actually test the product. So, check out our site for more info next week!