SureWest: Speed Updates

The folks at SureWest have been monitoring my blog and saw both Wednesday's and earlier-today's writeups. They actually were a little surprised that my downstream and upstream bandwidth measurement results were as slow as BroadbandReports had suggested; according to them I should be getting at least 50 Mbps in both directions. There's actually a speed test link on SureWest's site, but I'd resisted using it because:

1. It's provided by SureWest, and its objectivity is therefore unclear, and
2. Its Flash-based report format can't be embedded within a web page, as BroadBandReports' PNG image file-based approach easily allows

However, I was sufficiently piqued by SureWest's suggestions (and appeased by their clarification that the speed testing service is actually hosted by Ookla, an independent entity) that after re-running the BroadbandReports test today three consecutive times on my Windows XP SP2-based Dell Inspiron 700m laptop (which I'd also used on Wednesday), tethered to the LAN via a 100 Mbps CAT5 connection, I ran the Ookla-hosted test three times each on two different computers; my laptop and my dual 1.8 GHz G5 Power Mac, running OS 10.3.9 and with an integrated GbE CAT5 transceiver. The 100 Mbps versus 1 Gbps Ethernet difference between the two platforms shouldn't matter in this particular case, as the common 100 Mbps LAN connection coming out of my router is the fundamental bottleneck. Regardless, I came up with some surprising results.

First off, here's today's BroadbandReports data:

Now let's look at the Ookla data:

In comparing the BroadbandReports (both today's and Wednesday's) and Ookla data, two overriding conclusions jump out at me:

1. Ookla's measurement criteria produces lower downstream results on my laptop, compared to BroadbandReports' on the same system (where, as in the past, I was accessing the San Francisco server option using the Java-based test), and
2. Some characteristic(s) of the laptop (O/S? CPU? Ethernet transceiver connection to the rest of the system? A combination of these factors? etc?) translates into consistently slower network performance versus the Power Mac. This is a factor which I'll certainly keep in mind with future LAN- and WAN-based testing I do.

Ideas, folks?

Followup: I should also note, in focusing not only on bandwidth but also on latency, that I get consistent 6 ms ping-to-ack responses from www.yahoo.com via the SureWest fiber connection. Pings to that same domain URL from my AT&T DSL connection (note, which incorporates a Hawking HBB1 QoS processor intermediary, thereby possibly explaining at least some of the delay in this particular case) have 14 ms ack latencies.