<conchuod>
And we got support for setting CROSS_COMPILE bjoto
<conchuod>
btw bjoto if you could check my message from Friday about the PRs that'd be neat 🙂
lucascastro has joined #riscv
<bjoto>
conchuod: Re. CROSS_COMPILE: Oh, nice!
lucascastro has quit [Quit: Leaving]
<bjoto>
conchuod: Uhm, please remind me of the PR... NIPA?
paulk has quit [Read error: Connection reset by peer]
paulk-bis has joined #riscv
jjido has quit [Quit: My laptop has gone to sleep. ZZZzzz…]
<conchuod>
Bjoto yeah, nipa
jay2718 has joined #riscv
<bjoto>
conchuod: Got it. Sorry for the delay!
<conchuod>
nw, thanks
<conchuod>
I think guos entry series is jamming things up - seems to be triggering complete rebuilds for ever allmodconfig
<conchuod>
Every*
knolle has joined #riscv
<conchuod>
also was speaking to bjoto nathanchance last week - he had some suggestions about cutting the clang build times. Will give them a go at some stage
<bjoto>
Oh! That'd be great. I'm using LLVM=1 as my default way of building now...
rsalveti has joined #riscv
joev has quit [Ping timeout: 264 seconds]
joev has joined #riscv
grgy has quit [Remote host closed the connection]
<nathanchance>
conchuod: Something else occurred to me too: Do you use tuxmake’s toolchains, distro ones, or self-compiled for NIPA?
dilfridge is now known as dionysos
grgy has joined #riscv
<conchuod>
I use arnds
<conchuod>
I don't trust distro ones
<conchuod>
And my llvm is from your build script
<nathanchance>
Do you pass —pgo to build-llvm.py? PGO can help close the performance gap between LLVM and GCC
<nathanchance>
I assume that you do, since I seem to recall you running into a problem when doing PGO at one point but I cannot remember…
jacklsw has joined #riscv
BootLayer has quit [Quit: Leaving]
Andre_H has joined #riscv
<conchuod>
Eh, I don't think what I have at the moment is PGOed
<conchuod>
In fact, I am pretty sure that it is not.
<conchuod>
I'm happy enough Nathan once the 4x increased duration in building modules is gone.
<nathanchance>
Might be worth investing in if you are not rebuilding LLVM with enough frequency to erode the savings you get at kernel build time, should make a pretty large impact: https://github.com/ClangBuiltLinux/tc-build/pull/190
<nathanchance>
Those numbers are moderately recent haha
jack_lsw has joined #riscv
aerkiaga has joined #riscv
jacklsw has quit [Ping timeout: 252 seconds]
<conchuod>
I rebuild every point release usually
<conchuod>
So prob worth
<nathanchance>
I should write a universal benchmark script so people can easily evaluate if it is worth it to them
joev has quit [Ping timeout: 260 seconds]
joev has joined #riscv
aerkiaga has quit [Remote host closed the connection]
<palmer>
kito-cheng (IDK if Nelson is in here): looks like binutils/glibc master are unhappy with each other, I'm trying to bisect binutils to see what's up