Daanct12 has joined #msm8937-mainline
<
Danct12[m]>
<jojo_autoboy[m]> "it's kinda weird how the 8937 is..." <- it's actually big.little, sort of lol
<
jojo_autoboy[m]>
hmmm
<
Danct12[m]>
one is clocked higher than the other, and it's they're clustered
<
jojo_autoboy[m]>
how exactly
<
jojo_autoboy[m]>
cursed
<
jojo_autoboy[m]>
so i guess it can suffer from the same interconnect issues?
<
Danct12[m]>
s/it's//
<
Danct12[m]>
jojo_autoboy[m]: maybe
<
jojo_autoboy[m]>
ive seen this for a few other qcoms
<
jojo_autoboy[m]>
i know the 8956 and i assume the 8939 too
<
Danct12[m]>
i guess this is why we need the interconnect driver to work
<
jojo_autoboy[m]>
probably
<
Danct12[m]>
i already have a interconnect driver
<
Danct12[m]>
the issue now is that if i load it i get blue screen 🙃
<
jojo_autoboy[m]>
run a benchmark
<
Danct12[m]>
i have yet to enable cpu freq on my kernel
<
Danct12[m]>
2spooky4me
<
jojo_autoboy[m]>
rip
<
Danct12[m]>
does cpufreq works on qcom without it's own driver
<
Danct12[m]>
as in just PSCI is enough
<
jojo_autoboy[m]>
¯\_(ツ)_/¯
<
jojo_autoboy[m]>
does this even have psci
<
Danct12[m]>
it does
<
jojo_autoboy[m]>
oh ok
<
Daanct12>
/buffer 46
<
hacker420[m]>
<jojo_autoboy[m]> "image.png" <- hm
<
hacker420[m]>
So maybe reducing the CPU count to 4 in DTS could speed it up?
<
hacker420[m]>
oh lol it flies at least on the bare kernel
<
hacker420[m]>
yeah it's WAY faster
<
hacker420[m]>
think disabling the 4 cores on 8937 at least until it's properly set up is the way to go
<
hacker420[m]>
for now
<
hacker420[m]>
even phosh is somewhat more usable
<
hacker420[m]>
score is the same if not even lower
<
hacker420[m]>
a bit
<
hacker420[m]>
granted the benchmark is from 2011
<
hacker420[m]>
so ehhh
<
hacker420[m]>
and I doubt geekbench would run on musl
<
travmurav[m]>
hacker420[m]: I'm afraid you have missed by 40 years
<
hacker420[m]>
the problem is I don'
<
hacker420[m]>
* I don't think there is any other sane CPU bench for alpine
<
travmurav[m]>
I think drhy/whet can mostly live in local cache and doesn't need much ram/icc so its's pretty limited
<
hacker420[m]>
phoronix test suite...
<
travmurav[m]>
I guess ~800 then is just what an a53 cluster can do at 1G
<
Daanct12>
time to come back to msm8937 mainline
<
Danct12[m]>
what did i miss
<
hacker420[m]>
<Danct12[m]> "what did i miss" <- limiting 8937 to 4 cores makes it faster
<
Danct12[m]>
anything else
<
hacker420[m]>
don't think so
<
hacker420[m]>
<travmurav[m]> "I guess ~800 then is just what..." <- hm?
<
hacker420[m]>
I have 2.8GB available now
<
Danct12[m]>
nice overclock
<
hacker420[m]>
now I understand
<
hacker420[m]>
lmao
<
hacker420[m]>
I thought he was talking about ram not cpu frequency
<
hacker420[m]>
actually does it show cpu frequency in the current state?
<
Danct12[m]>
force updated msm89x7/6.6, there is no change other than editing commit message
<
Danct12[m]>
hacker420[m]: im sure clock doesnt even work yet lol
<
Danct12[m]>
Danct12[m]: new commits and fun things happens in msm89x7/6.6-experimental
<
Danct12[m]>
s/experimental/develop/
<
hacker420[m]>
<Danct12[m]> "new commits and fun things..." <- where is it
<
hacker420[m]>
or did you not push it yet
<
Danct12[m]>
hacker420[m]: i didnt push it yet
<
hacker420[m]>
what is in these new commits
<
Danct12[m]>
nothing other than trouble
Daanct12 has quit [Quit: WeeChat 4.1.1]
<
hacker420[m]>
<Danct12[m]> "nothing other than trouble" <- fun
<
Danct12[m]>
we grew up in the fun house
<
hacker420[m]>
danct spill the beans, what's in that branch