Firewire, USB, and Networking Performance - Intel P965: Foxconn P9657AA
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 would be a sensible way to look at USB and Firewire throughput. We utilize a RAM disk as our "server", since memory removes almost all overhead from the serving end. We turn off disk caching on the USB and Firewire side by setting up the drives for "quick disconnect" so our results are consistent.
We use 2GB of system memory with timings of 3-3-3-9 and set our RAM disk to 450MB with system memory at 1550MB. Our standard file is the SPECviewPerf install file, which measures 432,533,504 bytes (412.4961MB). After copying this file to our RAM disk, we measure the time for writing from the RAM disk to our external USB 2.0, Firewire 400, or Firewire 800 drive utilizing our internal Windows based timing program. The copy times in seconds are then converted into Megabits per second (Mb) to provide a convenient means of comparing throughput. Higher rates therefore mean better performance in this particular test.
The most interesting aspect in Firewire and USB throughput tests is the outstanding performance of an external hard drive connected to Firewire 800. Our benchmarks show Firewire 800 is up to 42% faster than a drive connected to the more common Firewire 400, and about 11% faster than the fastest USB 2.0 solution.
We see our Intel ICH8 chipset finally overtake perennial champion NVIDIA in USB 2.0 performance. The Firewire solution from VIA is still slightly faster than the TI solutions normally used on most motherboards.
Ethernet Performance
The current motherboard test suite includes LAN performance measurements. All of these boards utilize PCI or PCI Express based Gigabit controllers with the only difference being the supplier of the core logic.
The Windows 2000 Driver Development Kit (DDK) includes a useful LAN testing utility called NTttcp. We used the NTttcp tool to test Ethernet throughput and the CPU utilization of the various Ethernet Controllers used on the Intel motherboards.
We set up one machine as the server; in this test, an Intel system with an Intel CSA Gigabit LAN connection. Intel CSA has a reputation for providing fast throughput and is a logical choice for our Gigabit LAN server.
On the server side, we used the following Command Line as suggested by the VIA whitepaper on LAN testing:
Ntttcpr -m 4,0,‹server IP› -a 4 -l 256000 -n 30000
On the client side (the motherboard under test), we used the following Command Line:Ntttcps -m 4,0,‹client IP› -a 4 -l 256000 -n 30000
At the conclusion of the test, we captured the throughput and CPU utilization figures from the client screen.
The CPU utilization performance favors the Marvell Gigabit controllers with the Realtek solutions having the highest utilization numbers. However, the Intel solution on the ECS board has the lowest CPU utilization numbers we have seen in testing. The throughput numbers also favor the Marvell and Intel Gigabit controllers with the Realtek options close behind. The Foxconn board utilizes the Marvell 88E8056 solution used on the Gigabyte S3 board.
ASUS is now using the Attansic L1 PCI Express based controller chip and it posted comparable numbers to the more widely used Realtek and Marvell solutions. ASUS utilizes a PCI based Gigabit controller on the 570SLI even though the NVIDIA chipset has native support for Gigabit operations. This decision results in the worse throughput and average CPU utilization numbers. However, even with throughput at 646Mb/s it still exceeds what most home networks are capable of and certainly any DSL or Cable based Internet connection.
ncG1vNJzZmivp6x7orrAp5utnZOde6S7zGiqoaenZH9ygJNobw%3D%3D