Project update 2 of 10
We’re happy to confirm the GnuBee Personal Cloud 1 will ship with 100% libre software and firmware - there won’t be any binary blobs! This has real-world ramifications, the most immediate of which is it actually lowers the cost of production, which is in part why we can lower our funding goal.
As we originally described on the main campaign page:
In the current prototype, there is exactly one binary blob, but we have a clear path to removing it and plan to remove it before shipping the first production units. In particular, the binary blob is for the ASM1061 PCI-to-SATA bridge. As it turns out, a libre kernel driver for this particular chip was mainlined since our original board design, so it should simply be a matter of removing the associated SPI NOR flash chip and using the kernel driver directly to control the PCI-to-SATA bridge.
Well, we tested the kernel driver and it works quite well, with only a minor hit in performance. Here are some initial read and write test results using the same Western Digital 500 GB WDC WD5000BEVT drive and methodology used in our previous benchmarks:
|Interface||With Blob (using SPI NOR chip)||Without Blob (using libre kernel driver)|
|Read||79.2 MB/s||76.2 MB/s|
|Write||56.4 MB/s||56.2 MB/s|
This means we will completely remove the SPI NOR flash chip used to hold the ASM1061 proprietary blob.
As with most hardware projects like the GB-PC1, there is a delicate balance between the minimum number of units our factory is willing to make at a reasonable price (minimum order quantity or MOQ) and the price per unit. In general, when one goes up, the other goes down. Given GnuBee’s stated goal of "FLOSS Devices at Reasonable Prices," we really want to keep the price as low as possible. Initially, this meant an MOQ of 500 units, which is what the original funding goal of $75,000 represented. We’re trying to make a business out of this, so there is of course some margin built-in so we can cover non-manufacturing costs (like flights), invest in the next round of production without needing another crowdfunding campaign, and begin work on the next product.
In any case, between no longer needing the SPI NOR flash chip and renegotiating with our manufacturing partner, we’re able to reduce our MOQ from 500 to 300 without raising our prices at all. After consulting with the Crowd Supply team to double check our numbers and get their sign-off, we’re setting the campaign goal to $32,000. There won’t be as much left over to put into new products or the next batch, but we’re hoping sales after the campaign will help make up for that.
We have several more updates lined up, including benchmarks for encrypted drives and walk-throughs of some popular use cases for the GB-PC1. In the meantime, please get in touch if you have any questions, comments, or suggestions. Thanks for your support!
Larry and the GnuBee team