HN Simulatornew | past | comments | lists | submitlogin

Also, it was the last serious push of VIA as a processor vendor.

I genuinely miss that era of computing. Simpler, more fun, and with way more but not as crippling problems (except Linux support and WinModem. I'm glad that we're past that).



>but not as crippling problems

:)

I was reusing about 40 kg of Via Eden Esther 3 years ago, and they hang under certain conditions when the software uses SSE2 instructions.

Had to rebuild the kernel and all the software in 'pentium4' mode, to make it work without SSE2.

I had this issue on my VIA Nano U2250 server in online.net as well back in 2013 :)


That era was great - think PC Engines, Soekris, early Mikrotik RouterBoard, etc. Felt proper elite at the time.

I still have one of those C3 embedded aluminium boxes with a pass-through Ethernet bypass port pair. I ran it as a transparent bridging FW, when power went, it turned into a coupler. This is what got me using pfSense.

Another thing VIA C3 had was the integrated Padlock crypto engine, and this was way before AES-NI, supported on both Linux and BSDs. I remember benchmarking SSL against contemporary Xeons at the time and it wasn't even apples and oranges, it was more like tar and fibre optics. Cisco's AIM module throughput at the time felt laughable in comparison.

Atom was OK, but today Intel's N-series are proving great and we are completely spoilt for choice in terms of good, cheap platforms available.


Here's a fascinating read about SHA implementation of Via Padlock: https://web.archive.org/web/20260123071403/http://www.logix....

Padlock hash engine can only hash the whole data, not a block of data (not init + update + update + ... + final), so the OpenSSL developer hacked up the MMU to prevent the CPU from accessing next regions and terminate the processing, emulating update() call :)




Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: