| Update 6/30/01 After a couple of conference calls with Broadcom (who makes the HPNA chipset in the HPRO200 and just about every other HPNA 2 product) and Linksys, we discovered the reason why my test results were slower than Broadcom's. Upgrading the HPRO to Linksys 1.37.4 BETA firmware fixed the UDP streaming test lockups and doubled Ethernet WAN to HPNA LAN TCP throughput results, but results were still significantly slower then Broadcom's. But after I forced the Ethernet adapter on my test setup's WAN machine to 10BaseT Half duplex (instead of the usual autonegotiate setting), routing performance improved significantly, more closely matching Broadcom's expected results. I also updated the NETGEAR PA301 HPNA test client to the 2.31 firmware [download here] that was posted in May, but didn't find that it made that much difference in the test results. You should not encounter problems with normal use with Cable or DSL modems, but if you are using the HPRO as a LAN to LAN router, you should force the WAN side connection to 10Mbps/ Half duplex. New test results are given below, but I am also including the older results for reference purposes. OLD RESULTS Given my experience with the HPB200 bridge, I was all set to breeze through the performance testing and put this review (and myself) to bed. I started out my testing using the same Netgear PA301 10x PCI HPNA 2 adapter in the same computer that I used for the B200's tests and got the results below. (Details of how we tested can be found here.) [Tests run with Ver 1.37.1 firmware] | Test Description | Qcheck Transfer Rate (Mbps) | Qcheck Response Time (msec) [10 iterations 100byte data size] | Qcheck UDP stream [10S@500kbps] | | (Actual throughput- kbps) | (Lost data- %) | | LAN Adapter: NETGEAR PA301 10Mbps HPNA | | WAN - LAN | 0.6 | 6 avg 22 max | 480 | 4% | | LAN - WAN | 1.5 | 5 avg 7 max | locked up | -- | As you can see, something looked very wrong. Speeds were terrible and the router locked up on the LAN-WAN UDP streaming test consistently, requiring a push of the Reset switch to bring it back on-line. "Maybe it's the Netgear adapter" I thought, although I'd used the very same adapter in my other HPNA testing. So I tried a Linksys USB200HA Phoneline 10M adapter. That yielded the following results: [Tests run with Ver 1.37.1 firmware] | Test Description | Qcheck Transfer Rate (Mbps) | Qcheck Response Time (msec) [10 iterations 100byte data size] | Qcheck UDP stream [10S@500kbps] | | (Actual throughput- kbps) | (Lost data- %) | | LAN Adapter: Linksys USB200HA 10Mbps HPNA | | WAN - LAN | 1.2 | 7 avg 12 max | 329 | 33% | | LAN - WAN | 2.0 | 7 avg 17 max | locked up | -- | Better, but still nothing to write home about, with the router still locking up on the LAN-WAN UDP tests. To round out the testing, I tried a Netgear FA310TX 10/100 Ethernet card which turned in the results below: [Tests run with Ver 1.37.1 firmware] | Test Description | Qcheck Transfer Rate (Mbps) | Qcheck Response Time (msec) [10 iterations 100byte data size] | Qcheck UDP stream [10S@500kbps] | | (Actual throughput- kbps) | (Lost data- %) | | LAN Adapter: Netgear FA310TX 10/100 Ethernet | | WAN - LAN | 0.9 | 3 avg 12 max | 499 | 0% | | LAN - WAN | 6 | 2 avg 23 max | 490 | 9% | Not very good at all. Curiously, the LAN-WAN UDP test only locked up one out of 3 times and this time the Transfer Rate looked like I'd expect it to look. I finally ran an old-fashioned web transfer test from WAN-LAN, to double check the poor numbers and got a reading of 0.8Mbps, which was within the range of difference that I normally get between timed transfers and the Qcheck Transfer Rate test. Oh yeah, I also connected the Netgear and Linksys HPNA adapters and ran some quick tests between them (both on the LAN side of the router) and got about 4.5Mbps Transfer Rate. (Updated 5/30/01) This is about right due to the slower speed of the USB HPNA adapter. If you have both HPNA1.0 and 2.0 devices on your network, your 2.0 clients will not reach 10Mbps speeds. This is not a problem with the HPRO, but the way that the HPNA spec works. See this for the explanation! So, try as I might, I couldn't get the HPRO to turn in a decent performance! NEW RESULTS Routing - HPNA LAN Client [Tests run with Ver 1.37.4 BETA firmware & 10Mbps/Half duplex WAN client] | Test Description | Qcheck Transfer Rate (Mbps) | Qcheck Response Time (msec) [10 iterations 100byte data size] | Qcheck UDP stream [10S@500kbps] | | (Actual throughput- kbps) | (Lost data- %) | | LAN Adapter: NETGEAR PA301 10Mbps HPNA | | WAN - LAN | 3.0 | 5 avg 7 max | 499 | 0% | | LAN - WAN | 2.9 | 5 avg 7 max | 404 | 18% | Routing - Ethernet LAN Client [Tests run with Ver 1.37.4 BETA firmware & 10Mbps/Half duplex WAN client] | Test Description | Qcheck Transfer Rate (Mbps) | Qcheck Response Time (msec) [10 iterations 100byte data size] | Qcheck UDP stream [10S@500kbps] | | (Actual throughput- kbps) | (Lost data- %) | | LAN Adapter: NE2000 Compatible 10BaseT Ethernet | | WAN - LAN | 4.9 | 3 avg 4 max | 499 | 0% | | LAN - WAN | 5.2 | 3 avg 4 max | 499 | 0% | While I was at it, I ran HPNA / Ethernet Bridging tests to see how well that part of the box worked. Results show it to be somewhat of a weak link and not symmetrical in performance. Remember the following test does not include the router performance! Bridging - HPNA & Ethernet LAN Clients [Tests run with Ver 1.37.4 BETA firmware & 10Mbps/Half duplex WAN client] | Test Description | Qcheck Transfer Rate (Mbps) | Qcheck Response Time (msec) [10 iterations 100byte data size] | Qcheck UDP stream [10S@500kbps] | | (Actual throughput- kbps) | (Lost data- %) | | ETH LAN Adapter: UMAX 10BaseT PCMCIA Ethernet HPNA LAN Adapter: NETGEAR PA301 PCI | | HPNA -> Ethernet | 2.4 | 4 avg 9 max | 465 | 7% | | Ethernet -> HPNA | 5.2 | 4 avg 8 max | 499 | 0% | If you have both HPNA1.0 and 2.0 devices on your network, your 2.0 clients will not reach 10Mbps speeds. This is not a problem with the HPRO, but the way that the HPNA spec works. See this for the explanation! |