Miao Wang <[email protected]> writes: > Hi, > >> 2026年2月7日 17:17，Simon Josefsson <[email protected]> 写道： >> >> Shengqi Chen <[email protected]> writes: >> >>> I'd like to annouce Loong13, an unofficial Debian Trixie port for >>> LoongArch64, built entirely from source without binary blobs. >> >> Yay! >> >> Is anything from non-free-firmware included in the installer ISOs? > > Yes, the installer ISOs includes firmwares from non-free-firmware except > mini.iso, just like the ISOs for the official Debian release for other > architectures.
So built using binary blobs then. I wish terminology about this was a bit more well agreed on. When I read "built without binary blobs" I take that to mean the same as "built with no non-free firmware" but I think exchanges like this are common, so something about the meaning or use of these words are not widely agreed on. >> What hardware are used to build things? Are those machines able to run >> Loong13 without any binary blobs? > > The packages were built on a 32 core single socket 3C6000 server. The machine > is currently running Debian sid from ports since the system was installed > before Loong13 was built. No extra binary blobs are needed to run the system, > except the UEFI firmware. The driver for the embedded NICs on the motherboard > is currently not in the mainline kernel tree and requires an out-of-tree dkms > module. Seems like a good target to run a libre OS on then! Makes me desire such a machine. >> I've only seen 3A6000 boards around, but they are fairly low >> performance, and the more performant 3B6000 doesn't seem commercial >> available for Europe shipping. If anyone knows of a way to source these >> machines, let me know. > > The single-core performance of 3A6000 is much higher than 3C5000 and is a > little lower than Intel CPUs. The performance should be fit for daily normal > use. The single-core performance of 3C6000 should be similar to 3A6000, but > with more cores. It seems that there are some dealers on AliExpress for the > boards. Hmm. Is there any summaries comparing performance for 3A6000, 3B6000, 3C5000, 3B5000 (?), and 3A5000? My mental images has been the performance order is 'C > B > A ; 3600 > 3500' but maybe it isn't that simple. There are some other CPU variants on https://www.loongson.cn/EN/product/list?id=4 too. /Simon
signature.asc
Description: PGP signature
