github.com/antonblanchard.microwatt
1
0
Fork 0
mirror of https://github.com/antonblanchard/microwatt.git synced 2026-01-11 23:43:15 +00:00
Code Releases Activity

countbits: Move more popcount calculation before the clock edge

Browse Source 
Popcount takes two cycles to execute.  The computation of the final
popcount value in the second cycle has showed up as a critical path on
the Artix-7, so move one stage of the summation back into the first
cycle.

Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This commit is contained in:
Paul Mackerras 2025-04-15 09:59:44 +10:00
parent b14dd43ce6
commit 51dd7f578f
1 changed files with 11 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
countbits.vhdl
View File

@ -42,6 +42,10 @@ architecture behaviour of bit_counter is
type fourbit8 is array(0 to 7) of fourbit; type fourbit8 is array(0 to 7) of fourbit;
signal pc8 : fourbit8; signal pc8 : fourbit8;
signal pc8_r : fourbit8; signal pc8_r : fourbit8;
subtype fivebit is unsigned(4 downto 0);
type fivebit4 is array(0 to 3) of fivebit;
signal pc16 : fivebit4;
signal pc16_r : fivebit4;
subtype sixbit is unsigned(5 downto 0); subtype sixbit is unsigned(5 downto 0);
type sixbit2 is array(0 to 1) of sixbit; type sixbit2 is array(0 to 1) of sixbit;
signal pc32 : sixbit2; signal pc32 : sixbit2;
@ -96,6 +100,9 @@ begin
for i in 0 to 7 loop for i in 0 to 7 loop
pc8_r(i) <= pc8(i); pc8_r(i) <= pc8(i);
end loop; end loop;
for i in 0 to 3 loop
pc16_r(i) <= pc16(i);
end loop;
dlen_r <= datalen; dlen_r <= datalen;
pcnt_r <= do_popcnt; pcnt_r <= do_popcnt;
end if; end if;
@ -113,11 +120,13 @@ begin
for i in 0 to 7 loop for i in 0 to 7 loop
pc8(i) <= ('0' & pc4(i * 2)) + ('0' & pc4(i * 2 + 1)); pc8(i) <= ('0' & pc4(i * 2)) + ('0' & pc4(i * 2 + 1));
end loop; end loop;
for i in 0 to 3 loop
pc16(i) <= ('0' & pc8(i * 2)) + ('0' & pc8(i * 2 + 1));
end loop;
-- after a clock edge -- after a clock edge
for i in 0 to 1 loop for i in 0 to 1 loop
pc32(i) <= ("00" & pc8_r(i * 4)) + ("00" & pc8_r(i * 4 + 1)) + pc32(i) <= ('0' & pc16_r(i * 2)) + ('0' & pc16_r(i * 2 + 1));
("00" & pc8_r(i * 4 + 2)) + ("00" & pc8_r(i * 4 + 3));
end loop; end loop;
popcnt <= (others => '0'); popcnt <= (others => '0');