MaxGanzII__ has quit [Remote host closed the connection]
<unlord>
running this under gdb, it is in strlen()
naoki has quit [Remote host closed the connection]
MaxGanzII has joined #riscv
foxbat has quit [Ping timeout: 268 seconds]
foxbat has joined #riscv
naoki has joined #riscv
test63 has quit [Ping timeout: 250 seconds]
naoki has quit [Quit: naoki]
davidlt has joined #riscv
heat has joined #riscv
Ellenor has quit [Ping timeout: 255 seconds]
foton has quit [Ping timeout: 268 seconds]
handsome_feng has joined #riscv
naoki has joined #riscv
naoki has quit [Client Quit]
<unlord>
downgrading perf-6.6 to perf-5.10.16 (and patching the ebuild to compile) seems to have fixed it
psydroid has joined #riscv
mps has quit [Ping timeout: 260 seconds]
Ellenor has joined #riscv
davidlt has quit [Ping timeout: 252 seconds]
Leopold has quit [Ping timeout: 264 seconds]
Leopold has joined #riscv
mlw has quit [Ping timeout: 264 seconds]
mps has joined #riscv
mlw has joined #riscv
Stat_headcrabed has joined #riscv
foton has joined #riscv
davidlt has joined #riscv
junaid_ has joined #riscv
Tenkawa has joined #riscv
handsome_feng has quit [Quit: Connection closed for inactivity]
Leopold has quit [Remote host closed the connection]
Stat_headcrabed has quit [Quit: Stat_headcrabed]
Leopold has joined #riscv
test924 has joined #riscv
test924 has quit [Quit: Client closed]
Leopold has quit [Ping timeout: 260 seconds]
test924 has joined #riscv
Leopold has joined #riscv
craigo has joined #riscv
junaid_ has quit [Remote host closed the connection]
Finde has quit [Ping timeout: 245 seconds]
zBeeble has quit [Read error: Connection reset by peer]
zBeeble42 has joined #riscv
Stat_headcrabed has joined #riscv
KREYREN has quit [Remote host closed the connection]
KREYREN has joined #riscv
Stat_headcrabed has quit [Quit: Stat_headcrabed]
billchenchina has joined #riscv
Finde has joined #riscv
junaid_ has joined #riscv
test924 has quit [Quit: Client closed]
test924 has joined #riscv
junaid_ has quit [Remote host closed the connection]
Stat_headcrabed has joined #riscv
<courmisch>
To be clear, I am *not* working on getting upstream kernel (or SBI or uboot) or anything else on the K230.
<courmisch>
at least not at the moment
junaid_ has joined #riscv
crossdev has quit [Remote host closed the connection]
crossdev has joined #riscv
Stat_headcrabed has quit [Quit: Stat_headcrabed]
junaid_ has quit [Quit: leaving]
junaid_ has joined #riscv
junaid_ has quit [Remote host closed the connection]
craigo has quit [Ping timeout: 264 seconds]
BootLayer has joined #riscv
crossdev has quit [Remote host closed the connection]
crossdev has joined #riscv
<unlord>
courmisch: yeah, that is fine, I am happy with this solution
mlw has quit [Ping timeout: 255 seconds]
mlw has joined #riscv
shamoe has joined #riscv
ntwk has quit [Read error: Connection reset by peer]
Stat_headcrabed has joined #riscv
junaid_ has joined #riscv
craigo has joined #riscv
Stat_headcrabed has quit [Quit: Stat_headcrabed]
MaxGanzII has quit [Quit: Leaving]
MaxGanzII_ has joined #riscv
MaxGanzII_ has quit [Quit: Leaving]
Stat_headcrabed has joined #riscv
Stat_headcrabed has quit [Client Quit]
vagrantc has joined #riscv
billchenchina has quit [Quit: Leaving]
test924 has quit [Ping timeout: 250 seconds]
<courmisch>
there are no ways to get integer absolute value in a single vector instruction, are there?
<sorear>
nope, none
<sorear>
the bitmanip spec recommends neg + max; compare + masked neg might be faster or slower
<unlord>
no sign extension trick?
test924 has joined #riscv
<sorear>
srai, xor, sub? not sure why you'd do that when you have max
<unlord>
good point
<courmisch>
well there are many ways to do it with two instructions, sure
<courmisch>
but it seems only float supports it in a single instruction
<sorear>
fabs is a pseudoinstruction for fsgnjx
<courmisch>
yes and?
<sorear>
and only works because floats use sign-magnitude representation
<sorear>
integer abs would have to be an actual instruction. propose it if you can demonstrate a large enough improvement
KREYREN has quit [Remote host closed the connection]
<courmisch>
I don't know much about silicon, but adding the sign bit to the value bits does not sound like it would need that much extra, would it? well, it's too late anyway
BootLayer has quit [Quit: Leaving]
<sorear>
conceptually it wouldn't be weird as part of the "vector fixed point" instruction package, since a lot of them do additions and subtractions with a layer of muxes at the end
<mps>
hm, iiuc u74-mc cpu in JH7110 SoC don't have vector extension. cpufetch says 'Extensions: rv64imafdc_zicntr_zicsr_zifencei_zihpm_zba_zbb'
<courmisch>
more accurately, xor each value bit with the sign bit, then add the sign bit, I guess
<sorear>
you can do either way
<sorear>
the fixed-point pipeline will likely be happier with front-loading the negate rather than waiting for sign bit fanout, then an add later in the cycle
<courmisch>
mps: s/iiuc //
<mps>
courmisch: so my guess is ok
<sorear>
why are we talking about jh7110 in a conversation about vectors
<courmisch>
clearly we should be using sign-magnitude instead of that weird two-complements nonsense!! :O
<mps>
sorear: sorry, I didn't wanted to disturb and my question is not related to current discussion
<jrtc27>
(yes, u74 doesn't have v, even the more recent revisions; don't know why that's a noteworthy thing to mention but your understanding is correct)
MaxGanzII_ has joined #riscv
mlw has quit [Ping timeout: 255 seconds]
test924 has quit [Ping timeout: 250 seconds]
mlw has joined #riscv
foton has quit [Read error: Connection reset by peer]
junaid_ has quit [Remote host closed the connection]
hays has quit [Client Quit]
hays has joined #riscv
<sorear>
« We have already defacto forked from the psABI pretty much everywhere » not sure how scary this should be as a psABI maintainer
<conchuod>
I feel like I need to read the whole thread where that quote originates..
<Tenkawa>
conchuod: hey got a question about the Milk-V Mars Patches... I've got a CM board sitting here running successfully with 6.8.0-rc5. The only thing I'm having difficulty tacking down is getting the usb to function (waveshare IO board for informational purposes)... Anything obvious about this setup I should know or current known issues? I noticed you submitted a fair number of patches I'm using on it in mainline.
<Tenkawa>
s/tacking/tracking
<Tenkawa>
Everything else seems to be functioning ok (even NVMe root)
foxbat has quit [Quit: bye]
foxbat has joined #riscv
foxbat has quit [Client Quit]
<conchuod>
Tenkawa: I'm sorry, what did I submit patches for that's being used on a jh7110 system?
<Tenkawa>
It was delegated to you.. (pending this "is" you..)
<conchuod>
Whether it is a vl805 or not I cannot tell obviously, but it has something along those lines.
<Tenkawa>
Even the Star64 realisticly is with a few tweaks
<Tenkawa>
(which I also want to get mine converted one of these days)
<Tenkawa>
ok.. lets see if these new dtbs do any better
prabhakalad has quit [Quit: Konversation terminated!]
prabhakalad has joined #riscv
<Tenkawa>
Well gotta run for now but thanks for all the help. I will be back hopefully with good news after some more testing. Tomorrow will be out due to no time but I plan on taking a look again Sunday I hope. Cheers all
Tenkawa has quit [Quit: Was I really ever here?]
chripo has quit [Quit: ""]
chripo has joined #riscv
hightower3 has joined #riscv
Trifton has quit [Quit: Client exited]
hightower4 has quit [Ping timeout: 268 seconds]
crossdev has quit [Remote host closed the connection]