thelounge7571340 has quit [Remote host closed the connection]
phil_ has joined #asahi
philm has quit [Ping timeout: 480 seconds]
jacksonchen666 has joined #asahi
cmalvi has joined #asahi
jacksonchen667 has joined #asahi
jacksonchen666 has quit [Ping timeout: 480 seconds]
jacksonchen667 is now known as jacksonchen666
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
phil_ has quit [Remote host closed the connection]
jacksonchen666 has quit [Ping timeout: 480 seconds]
jacksonchen666 has joined #asahi
cmalvi has quit [Quit: cmalvi]
jacksonchen666 has quit [Ping timeout: 480 seconds]
Gaspare has joined #asahi
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
IcaroDextris has joined #asahi
nela has quit [Ping timeout: 480 seconds]
olebra has quit [Quit: rebooting, smell ya later]
borisk has joined #asahi
Stroller has joined #asahi
borisk has quit [Quit: Page closed]
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
Gaspare has quit [Quit: Gaspare]
nela has joined #asahi
borisk has joined #asahi
caleb has joined #asahi
julio7359 has joined #asahi
caleb has quit []
<borisk>
marcan: Would there be any reason to think that M1 Max under Asahi is not running at full throttle? I am running the 6.0rc6 kernel and the C++ compilation performance is surprisingly underwhelming compared to i9-12900K, especially considering the differences in the memory bandwidth (which C++ compilation is quite sensitive to, IMO).
<jannau>
numbers might be helpful, you're comparing a 8+8 code desktop processor vs. a 8+2 laptop processor. the CPU can not use the full memory bandwidth of the m1 max
r0ni has joined #asahi
<marcan>
geekbench says the intel has about 50% better performance
<marcan>
considering it has more cores and easily eats 190W at full CPU throttle, while the M1 Max will push 60W or so, that makes the M1 Max more than 2x as efficient as the intel
<marcan>
thus: get an M1 ultra and it'll be better than that intel and use less power :p
<marcan>
also, I don't think compilation is memory bandwidth bound at all?
<marcan>
it should be memory *latency* bound
<marcan>
OTOH, there is a memory latency related tweak we are not applying right now, because it didn't make much of a difference for the tests I did. I think it mostly only affects certain single-core type workloads though
<marcan>
(increasing the memory power management timeout to reduce latency on first access after some idle time when the CPU is busy)
<marcan>
I imagine all cores doing stuff at once should be sufficient to keep the RAM awake anyway
r0ni has joined #asahi
<borisk>
marcan, jannau: thanks for the feedback. I should have mentioned M1 Max is in Mac Studio, not a laptop. The numbers I get is that M1 Max is about ~30% slower, depending which compiler I use (GCC or Clang). Might be a good idea to try Apple Clang under Mac OS to make sure there are no drastic discrepancies.
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
nicolas17 has joined #asahi
IcaroDextris has quit [Quit: Connection closed for inactivity]
chengsun_ has joined #asahi
chengsun has quit [Ping timeout: 480 seconds]
<marcan>
30% slower sounds completely expected for these CPUs
<marcan>
so I really don't see what the problem is
<marcan>
you're comparing a 16 core processor that can turbo at 5.2 GHz and burn power vs. a 10 core processor that maxes out at 3 GHz multicore and uses 1/3 of the power
<marcan>
if anything the Max being only 30% slower is pretty awesome
<marcan>
borisk: ^
<j`ey>
also it seems the i9 has 24 threads too
<_jannau_>
also are you cross-compiling on either the intel or the m1 system? if not you're comparing different workloads
<marcan>
j`ey: yeah (though that probably helps intel close the IPC gap with Apple more than anything else, since Apple are much better at that)
<marcan>
(much better OoO on Apple)
chengsun_ has quit [Quit: Quit]
<j`ey>
and what about RAM sizes? on the 2 machines
chengsun has joined #asahi
guillaume_g has quit []
<borisk>
marcan: ok, thanks for confirming this is not unexpected
<borisk>
_jannau_: this is native compilation on both. Yes, I realize I am not comparing the same workloads, especially if optimization is involved (since it's optimizing for different architectures). Though I see the same difference with debug builds.
<borisk>
j`ey: both have 64GB. Specifically, DDR5-4400 with ECC in the i9 case and LPDDR5-6400 in the M1 Max case.
<borisk>
Overall, M1 Max has way, way better and faster memroy/cache subsystem and it's made on a better node, so I sort of expected it will at least match i9. Also, this is subjective (can't measure power draw), but i9 doesn't seem to run especially hot: the fan is not audible and the exhaust air is barely warm, about on par with Mac Studio.
int16h has quit [Quit: WeeChat 3.7]
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
danielnechtan has joined #asahi
chengsun has quit [Ping timeout: 480 seconds]
marvin24 has joined #asahi
r0ni has quit [Quit: Leaving]
marvin24_ has quit [Ping timeout: 480 seconds]
danielnechtan has quit [Quit: WeeChat 3.7]
chengsun has joined #asahi
danielnechtan has joined #asahi
chengsun_ has joined #asahi
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
chengsun has quit [Ping timeout: 480 seconds]
chengsun has joined #asahi
chengsun_ has quit [Ping timeout: 480 seconds]
Gaspare has joined #asahi
<ChaosPrincess>
i seem to have some problems installing m1n1 after a ventura update. i built it with RELEASE=1 and CHAINLOADING=1, did cat m1n1.bin /path/to/kernel/*.dtb and blessed the resulting object with kmutil. But instead of chainloading, it says 'devicetree found but no kernel'
r0ni has joined #asahi
Agua has joined #asahi
<jannau>
ChaosPrincess: that doesn't look ventura related. the 1st stage m1n1 doesn't get DTBs, it needs variables to tell it from where to chainload and the partuuid of the ESP for this installation
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
<ChaosPrincess>
is there a separate plane for the cursor in the dcp driver? the cursor looks really weird, and i do not see a separate plane from skimming the dcp driver code
bestouff has quit [Ping timeout: 480 seconds]
<j`ey>
yes I believe so
<j`ey>
but I guess it's a higher level that decides what goes in what planes
bestouff has joined #asahi
<j`ey>
yeah chadme_ & janna_ were talking about cursor planes this morning on -dev
Agua has quit [Remote host closed the connection]
Agua has joined #asahi
bestouff has quit [Read error: Connection reset by peer]
bestouff has joined #asahi
julio7359 has quit [Remote host closed the connection]
julio7359 has joined #asahi
<jannau>
no, the dcp driver in asahi-wip doesn't have a cursor plane and the cursor looks as expected for me in plasma/wayland
<jannau>
dcp has just the primary plane
<ChaosPrincess>
it looks like expected when stationary, but flickers and jumps when moved
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
<j`ey>
jannau: oh my bad. but DCP itself has planes and the current driver just doesnt use them or?
<jannau>
yes, dcp can blend at least 2 planes (the rpc interface has 4). it would be a good idea to use the second plane for the cursor
<j`ey>
ok cool, so just me misunderstanding the current driver, got it!
<jannau>
no need to redraw the primary plane for mouse movements. the blending has an unfortunate limitation for this use case
nicolas17 has quit [Quit: power outage]
<jannau>
if there are less than 32x32 pixel of the overlay on screen dcp scaling the remaing data to 32 picxels or so
<j`ey>
so just dont move the mouse near the edge of the screen :P
<ChaosPrincess>
jannau: what exactly does dcp do if you move cursor offscreen? not sure i understood what you meant by 'scaling to 32'
<j`ey>
yeah that bit was a bit confusingm sounded like it scaled up the remaining pixels, which would make the cursor bigger if it was near the edge?
<jannau>
the cursor gets distorted if it's more than half off-screen with an naive cursor plane. but that is not in the merged dcp driver
Agua has quit [Ping timeout: 480 seconds]
<ChaosPrincess>
and macos solves it by adjusting the contents of cursor plane if it gets too close to the edge?
<jannau>
I don't remember exactly how the cursor get's distorted
<j`ey>
I seem to remember markan showing this on stream at one point, or something similar
<jannau>
ChaosPrincess: wrong. it switches to gpu compositing
<ChaosPrincess>
i guess thats also a solution
<ChaosPrincess>
still kind of a hack tho
jluthra_ has quit [Remote host closed the connection]
jluthra_ has joined #asahi
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
irth has quit [Remote host closed the connection]
irth has joined #asahi
bluetail has joined #asahi
r0ni has joined #asahi
___nick___ has quit [Ping timeout: 480 seconds]
karolherbst has quit [Quit: Konversation terminated!]
karolherbst has joined #asahi
jacksonchen666 has quit [Quit: jacksonchen666]
theKH has joined #asahi
jjido has joined #asahi
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
jjido has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
axboe has quit [Remote host closed the connection]
theKH has quit [Ping timeout: 480 seconds]
ptudor_ has joined #asahi
Agua has joined #asahi
Agua has quit [Remote host closed the connection]
millenialhacker has joined #asahi
ptudor has quit [Ping timeout: 480 seconds]
theKH has joined #asahi
theKH has quit [Ping timeout: 480 seconds]
millenialhacker has quit [Quit: Konversation terminated!]
theKH has joined #asahi
theKH has quit [Remote host closed the connection]
theKH has joined #asahi
chengsun has quit [Ping timeout: 480 seconds]
chengsun has joined #asahi
bcrumb has joined #asahi
<bcrumb>
is there any difference to battery charging when running arch / mac os?
Etrien__ has joined #asahi
bcrumb has quit []
Etrien has quit [Ping timeout: 480 seconds]
SSJ_GZ has quit [Ping timeout: 480 seconds]
Etrien__ has quit [Read error: Connection reset by peer]
Etrien has joined #asahi
chengsun has quit [Quit: Quit]
chengsun has joined #asahi
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]
nicolas17 has joined #asahi
chengsun_ has joined #asahi
<nicolas17>
I was building a raspberry pi disk image with pi-gen (the same scripts they use for the official raspberry pi os images)
<nicolas17>
if you run it on Intel it has to do half the work under qemu-aarch64-user (it creates a base environment, then chroots into it and installs more stuff, and at that point it's running the ARM64 binaries inside the chroot)
<nicolas17>
I have a Lenovo laptop with an Intel i5-8250U CPU (= need to use qemu) and a *mechanical* hard disk, and a MacBook Pro M2 with Apple's fast SSDs and Asahi... any guesses about the perf difference running these scripts? :P
<nicolas17>
(I think the SSD is actually making the biggest difference)
chengsun has quit [Ping timeout: 480 seconds]
chengsun has joined #asahi
<nicolas17>
lenovo: 42 minutes; MBP: 2m30s
<nicolas17>
maybe I should try a ramdisk on the lenovo
chengsun_ has quit [Ping timeout: 480 seconds]
CME_ has joined #asahi
lkron has joined #asahi
<tpw_rules>
honestly you might be surprised how much of that is qemu
CME has quit [Ping timeout: 480 seconds]
bcrumb has joined #asahi
bcrumb has quit []
lkron has quit [Quit: Reconnecting]
lkron has joined #asahi
lkron has quit [Quit: leaving]
bluetail has quit [Ping timeout: 480 seconds]
Gaspare has quit [Quit: Gaspare]
Etrien__ has joined #asahi
amarioguy2 has joined #asahi
Etrien has quit [Ping timeout: 480 seconds]
thelounge7571340 has joined #asahi
thelounge7571340 has quit [Remote host closed the connection]