Here’s a short version of the last few weeks. The bad news: the first attempt at shipping all GnuBee Personal Cloud 1 units from Hong Kong to Crowd Supply’s warehouse ended in complete failure. The good news: as of today, all GB-PC1 units are again on their way from Hong Kong to Crowd Supply’s warehouse, this time with a much higher chance of actually arriving!
We learned the hard way that international shipping is not easy - don’t let anyone tell you otherwise. Sure, it can go off without a hitch, but when it doesn’t, things go bad really quickly.
We learned this firsthand when our GnuBee Personal Cloud 1 shipment arrived in Los Angeles from Hong Kong and was subject to a random US customs inspection. Unfortunately, the inspection revealed something we’d overlooked - the custom USB-to-serial cable we’d made wasn’t properly labeled as being "Made in China".
This was enough to stop the shipment dead in its tracks. Suddenly, our shipping company (SF Express) was telling us we only had two choices: pay to return all the goods to Hong Kong or pay (a more modest sum) to have them destroyed in Los Angeles. We had 30 days to figure it out.
Over the next few days, we realized our problems were compounded by SF Express’s handling of the situation. In particular, SF Express never notified the recipient (Crowd Supply) of the problem, they stopped answering our calls and emails, and they abruptly sent back the entire shipment to Hong Kong without our consent, which was particularly annoying given we were on the verge of working out a solution involving a customs agent applying the appropriate "Made in China" labels to the goods being held in Los Angeles.
We met the returned shipment in Hong Kong, applied the labels, found a better courier (UPS), dealt with a typhoon, and today shipped again all units to Crowd Supply’s warehouse. We’re hoping our shipping woes are over - we’ll keep you posted over the next few days either way!
Well, it seems like I have made an error in the benchmarking. I discovered this when submitting the patch to LEDE/OpenWrt and ended up talking to a few LEDE devs at length. I’m still trying to get to the bottom of what happened, but as far as I can tell, overclocking the processor does not change the benchmarks. It appears the clock is incorrect when overclocking. Since all the benchmarks involve measuring time, this means the benchmarks are likely incorrect. So far, I suspect a problem with the kernel code, but I haven’t yet nailed it down. I’ll send out more updates on the benchmarks as I make progress.
Thanks for your patience and support!