https://bugs.kde.org/show_bug.cgi?id=381326
--- Comment #3 from John Reiser <jrei...@bitwagon.com> --- (In reply to Julian Seward from comment #2) > for (z=p; n; n--, z++) if (*z) *z=0; > > What is the point of this? Why not just assign zeroes unconditionally? > Is this some game with the MESI protocol, to avoid unnecessary RFOs? [It's been 30 years, so some of this is hazy ...] i386 did not have on-die cache (4kB of on-CPU cache in the i486 was a HUGE improvement), the write buffer was shallow (one? zero?), and the CPU execution pipeline was very short (little penalty for branching: decoding only.) A memory operation took 4 cycles; you could afford 2 instructions to avoid a write, especially if those two instructions fit into one memory read operation as input to the decoder. -- You are receiving this mail because: You are watching all bug changes.
