Performance Tests
I used iperf to measure LAN to WAN and WAN to LAN routing throughput. To establish a baseline, I first connected both Nomad and Mercury to the GP2's LAN ports, ran iperf tests in each direction (Nomad to Mercury, and Mercury to Nomad) and obtained average TCP throughput of 94.1Mbps with both systems running Mandrake Linux 10.1.
I then disconnected the Linksys GP2 from my cable modem and connected Nomad to the product's Internet (WAN) port, leaving Mercury connected to a LAN port. After assigning the router and the laptop each a static IP address, I configured Mercury as a DMZ host using the GP2's web utility. I then ran 3 tests using iperf in each direction and averaged the results to obtain measured LAN to WAN throughput of 24.2Mbps and 26.1Mbps for WAN to LAN.
Although this test does point out that the Linksys WRT54GP2 can't route traffic across its firewall at full 100baseTX speeds, I think it's important to note that it probably wasn't designed to. However, the throughput performance shown here is more than capable of keeping up with anybody's cable or DSL connection - at least those here in the U.S.!
Notes:
- All iperf tests used a TCP window size of 16.0 kBytes and a 10 second test length.
- Nomad: Dell Inspiron 1000 laptop with Celeron 2.2Ghz processor and 224MB of memory
- Mercury: Compaq Presario S5400NX with P4 2.6Ghz processor and 504MB of memory
Wireless Tests
For my wireless tests, I slipped a Dell Truemobile 1300 wireless card into Nomad and booted Windows XP so that I could use Ixia's Qcheck.
As many of you are probably painfully aware, a wireless connection's quality and speed can vary widely and will depend on many factors, including the location you've chosen for your wireless router. So don't place it inside your microwave or near other sources of interference.
I placed the GP2 in my computer "dungeon", which is a spare bedroom upstairs in my two-story wood-frame construction home with all sorts of things like monitors, TV sets and fluorescent lights that should be bad news for wireless connectivity. But I found I could go anywhere in the house with my laptop and stay connected at surprisingly high speeds. How high you ask? Here's a summary of my results, presented in approximate order of expected signal attenuation:
Location | Throughput | Ping | Location Description |
Dungeon | 18.519Mbps | 1mS | Sitting right beside the Linksys WRT54GP2 |
Upstairs Hallway | 19.753Mbps | 1mS | About 10 feet away with a clear line of sight to GP2 |
Upstairs Bedroom | 19.608Mbps | 1mS | About 10 feet away, through 1 interior wall |
Dining room | 18.307Mbps | 1mS | About 15 feet away, through 1 interior wall and a wood frame floor |
Living room | 14.925Mbps | 1mS | About 15 feet away, through 2 interior walls and a wood frame floor |
Bottom of stairs | 15.810Mbps | 1mS | About 20 feet away through 1 interior wall and down a flight of steps |
Den | 19.512Mbps | 1mS | About 20 feet away, through 4 interior walls and a wood frame floor |
Garage | 1 to 14Mbps
high variation |
1mS | About 35 feet away, through 2 interior walls, 1 exterior wall, and a floor |
Test Notes:
- These are all average of 3 tests from each location (except the garage, where reception was too sketchy to be meaningful).
- All Qcheck tests used 1000 kByte data size.
- All distances are estimates as if you could draw a line directly from point to point through walls and floors.
As I said, I was pleasantly surprised by the results, since the GP2 essentially gave me whole-house coverage at speeds better than a 10Mbps Ethernet connection.