ntwk has joined #riscv
cousteau has quit [Quit: ♫ I can't forget the day I shot that network down ♫]
pecastro has quit [Ping timeout: 260 seconds]
KREYREN has joined #riscv
KREYREN has quit [Remote host closed the connection]
KREYREN has joined #riscv
wingsorc has joined #riscv
wingsorc has quit [Ping timeout: 260 seconds]
Tenkawa has quit [Quit: Was I really ever here?]
zkrx has quit [Ping timeout: 264 seconds]
Armand has quit [Remote host closed the connection]
Armand has joined #riscv
naoki has joined #riscv
nlhcd has joined #riscv
zkrx has joined #riscv
test924 has quit [Quit: Leaving]
test924 has joined #riscv
KREYREN has quit [Remote host closed the connection]
naoki has quit [Quit: naoki]
nlhcd has quit [Quit: Quit]
jmcgnh has quit [Excess Flood]
jmcgnh has joined #riscv
BootLayer has joined #riscv
mlw has joined #riscv
aredridel has quit [Ping timeout: 272 seconds]
ntwk has quit [Ping timeout: 264 seconds]
TMM has joined #riscv
Armand has quit [Quit: Leaving]
coldfeet has joined #riscv
hightower2 has quit [Remote host closed the connection]
<
courmisch>
is there a trick to extract signs in RVV? I guess VSRA.VI (bits - 1) then VOR.VI 1 works as long as we don't care for zero
<
sorear>
an integer signum function?
<
sorear>
are you doing something that would be better expressed in terms of masks?
<
sorear>
there's also (x > 0) - (x < 0) (branch-free spaceship operator in three instructions) but slt works a bit differently in rvv
<
courmisch>
I need to inject the sign from one vector to another, but in integer
<
courmisch>
well, that's not the only way to do it
<
sorear>
what's the spec for copysign(+1, 0) or copysign(INT_MIN, +1)?
<
courmisch>
values are small and the value for 0 is basically irrelevant because it gets multiplied by zero
<
sorear>
compare, compare, mask-xor, masked-negate?
<
courmisch>
not sure if this works out better with 4 comparisons or with 2 comparisons on absolute values, TBH
<
sorear>
8 iterations? somebody doesn't like vectors very much
<
courmisch>
sorear: video codec macroblocks have the size that they have
dlan has quit [Remote host closed the connection]
<
courmisch>
and H.263 is almost 30 years old, so yeah, 64-bit vectors :/
<
courmisch>
well, technically, the calculation of d is in 16-bit arithmetic, so 128-bit vectors
dlan has joined #riscv
<
sorear>
i was thinking more "process several macroblocks at once"
<
sorear>
possibly harder with codecs from the last ten years where macroblocks come in a million different sizes and shapes
<
courmisch>
I don't know if that would be possible for deblock
<
courmisch>
for mocomp, I think it would totally fail because the motion vectors can change from one block to the next
<
courmisch>
also you may have one block depending on the previous one, which ruins parallelisation
Kedleston has quit [Ping timeout: 255 seconds]
Kedleston has joined #riscv
pecastro has joined #riscv
ntwk has joined #riscv
psydroid2 has joined #riscv
guerby_ is now known as guerby
BootLayer has quit [Quit: Leaving]
coldfeet has quit [Remote host closed the connection]
jfsimon1981 has quit [Remote host closed the connection]
jfsimon1981 has joined #riscv
llu_ has quit [Ping timeout: 252 seconds]
coldfeet has joined #riscv
Tenkawa has joined #riscv
nlhcd has joined #riscv
<
courmisch>
sorear: there's no magic trick to divide (toward zero) by a power of two, is there? :(
<
courmisch>
GCC gives srai;andi;add;srai in scalar
BootLayer has joined #riscv
Stat_headcrabed has joined #riscv
Stat_headcrabed has quit [Client Quit]
coldfeet has quit [Remote host closed the connection]
ntwk has quit [Read error: Connection reset by peer]
ntwk has joined #riscv
coldfeet has joined #riscv
aredridel has joined #riscv
KREYREN has joined #riscv
___nick___ has joined #riscv
coldfeet has quit [Quit: leaving]
coldfeet has joined #riscv
JanC has quit [Read error: Connection reset by peer]
JanC has joined #riscv
___nick___ has joined #riscv
___nick___ has quit [Client Quit]
___nick___ has joined #riscv
JanC has quit [Remote host closed the connection]
esv has joined #riscv
JanC has joined #riscv
TMM has joined #riscv
nlhcd has quit [Quit: Quit]
nlhcd has joined #riscv
nlhcd has quit [Remote host closed the connection]
KREYREN has quit [Remote host closed the connection]
Stat_headcrabed has joined #riscv
Stat_headcrabed has quit [Client Quit]
ntwk has quit [Read error: Connection reset by peer]
llu has joined #riscv
llu has quit [Ping timeout: 260 seconds]
wingsorc has joined #riscv
Tenkawa has quit [Ping timeout: 255 seconds]
Tenkawa has joined #riscv
llu has joined #riscv
ntwk has joined #riscv
wingsorc has quit [Quit: Leaving]
wingsorc has joined #riscv
BootLayer has quit [Quit: Leaving]
<
sorear>
probably not. wonder why rtz didn't make the cut for vxrm
___nick___ has quit [Ping timeout: 240 seconds]
nlhcd has joined #riscv
nlhcd1 has joined #riscv
nlhcd has quit [Remote host closed the connection]
nlhcd1 has quit [Client Quit]
mlw has quit [Ping timeout: 240 seconds]
ntwk has quit [Read error: Connection reset by peer]
coldfeet has quit [Remote host closed the connection]
edef_ is now known as edef
Maylay_ has quit [Ping timeout: 268 seconds]
Maylay has joined #riscv
ntwk has joined #riscv
pecastro has quit [Ping timeout: 240 seconds]
DesRoin has quit [Ping timeout: 260 seconds]
DesRoin has joined #riscv
naoki has joined #riscv
Noisytoot has quit [Excess Flood]
Noisytoot has joined #riscv