Re: The market of ASICs (One GigaKey / Second?)
"Dan Oetting" <[email protected]> Mon, 9 Aug 2004 23:51:13 -0600
| Newsgroups | gmane.comp.distributed-net.hardware |
|---|---|
| Message-ID | <[email protected]> |
On Aug 9, 2004, at 10:14 PM, Elektron wrote: > > On 10 Aug, 2004, at 09:16, Dan Oetting wrote: > >> Why so slow! I get 16 bytes in 17 cycles on a G4. I was all excited >> until I found out that IBM apparently already had a patent on the >> method and it couldn't actually run that fast because it is limited >> by the memory bandwidth. > > Using tables or not? I found that using a 12-bit table was fastest, > since a 16-bit table can't fit in my L1 cache. Anything larger than a 8 bit table gets real tricky in AltiVec so I don't use it except for the cleanup. The main loop is just a little bit of polynomial arithmetic which goes real fast when you have 128 bit registers. > I wasn't aware that you could patent CRC either. In the US you can patent anything. _______________________________________________ Hardware mailing list Hardware-Ra3b/QYEcJ3d140v2zMXi0fjHoOT/h/[email protected] http://lists.distributed.net/mailman/listinfo/hardware