Rick’s Rants and Raves

In my first phase of speed testing, I dealt with an unmodified NSLU2. I knew I wanted to apply the Unslung firmware, and probably make it a Turbo slug, but before I did that, I wanted to have some assurance that I would see some performance difference.

First, I installed Unslung, after I tested to be sure I could get into the RedBoot boot loader incase I bricked it.

Next, I enabled telnet, then I verified that my NSLU2 was in-fact “under-clocked.” It was. Mine read 133.7 something. Definitely not 266. Bummer. At least there was hope for some improvement.

Before I cracked the case, I decided to do a quick test of the NSLU2 with the Unslung firmware installed just to see if there was any performance change. There wasn’t. I did three runs of copying files from the laptop to the NSLU2 and the numbers were exactly the same as before.

Finally, I got the nerve to crack the case. I was rather daunted by the description of the procedure but decided to press on. Heck, it was my $80 toy. If I destroyed it, the wife glare would only last as long as my desire to buy a new one.

With a single snip of my wire cutters (albeit a nervous snip) I made my own TurboSlug. The picture on that page is an extemely good macro shot. When I destroyed the resistor, the results were little more than 2-3 grains of sand. I should have taken a picture with something next to it for scale. Anyway, I did it. I put it all back together (gently) and it powered on just fine.

I confirmed that the work had successfully created my TurboSlug.

So, by now you see where this is going. I did another set of speed testing. And the results are in–it is definitely faster than it was, but still not as fast as a PC with a shared drive. Below, tNSLU2 is the improved TurboSlug. Like my previous test, these are average bandwidth in Kilobytes per second (kilo=1024 bytes)

In every transfer test except big files going from the NSLU2 to the laptop, I saw about a 40% performance improvement in the TurboSlug. The big file from NSLU2 to laptop was only about 30% better. However, still a ways off from PC-PC communication.

Is 40% improved performance to void your warranty? You’ll have to decide that for yourself. Is the device worth $80 to serve as a very low power-consuming NAS on your home or small office network? I think so.

Additional thoughts:

One weakness with my methodology was that I didn’t use the same USB hard drive attached to both the NSLU2 and the PC. That would have been the most thorough test to completely isolate the NSLU2′s performance. I figured that since the bandwidth numbers were all much lower than the “theoretical” max of USB 2.0 (40 MB/s) that I wasn’t bumping up against a USB vs EIDE limitation.

Though I don’t mention it explicitly in the rest of the analysis, I’m sure of a couple of other possible variables:

• All networked devices were talking at 100Mb/s, not 10Mb/s. If they had been at 10Mb/s, the max I could have seen on the bandwidth tests would have been 1280 KB/s.
• The drives are definitely working at USB 2.0, not some lower bandwidth because the speed of the NSLU2 is better than USB 1.0 or USB 1.1. I plugged in a USB card reader that is only 1.1 and it was much slower than everything else on the NSLU2 and the PC.

That’s all for now.