00:07
raster has joined #panfrost
00:10
alyssa has quit [Quit: leaving]
01:04
raster has quit [Quit: Gettin' stinky!]
01:48
stikonas has quit [Ping timeout: 260 seconds]
02:03
rando25892 has quit [Ping timeout: 260 seconds]
02:10
rando25892 has joined #panfrost
03:21
archetech has quit [Quit: Leaving]
03:25
archetech has joined #panfrost
04:36
vstehle has quit [Ping timeout: 256 seconds]
05:41
rak-zero has joined #panfrost
06:00
vstehle has joined #panfrost
06:10
vstehle has quit [Quit: WeeChat 2.9]
06:12
vstehle has joined #panfrost
06:38
archetech has quit [Quit: Leaving]
07:26
nlhowell has joined #panfrost
07:26
nlhowell has quit [Remote host closed the connection]
07:26
nlhowell has joined #panfrost
07:27
nlhowell has quit [Remote host closed the connection]
07:28
nlhowell has joined #panfrost
07:28
nlhowell has quit [Client Quit]
08:12
raster has joined #panfrost
08:58
<
macc24 >
icecream95: thanks
09:28
warpme_ has joined #panfrost
10:13
icecream95 has quit [Ping timeout: 260 seconds]
11:27
nlhowell has joined #panfrost
11:36
<
chewitt >
all the screensaver/visualisations (shadertoy based) in Kodi are now working :)
11:36
<
chewitt >
this means I officially have nothing to nag about
11:37
<
chewitt >
huge kudos to the team!
11:44
stikonas has joined #panfrost
12:05
<
robmur01 >
HdkR: a new syscall solely for the purpose of abusing the compat ABI that may not even be available? I'm gonna guess "not popular"...
12:10
<
HdkR >
Guess I might need to double check how `in_compat_syscall` works
12:10
<
HdkR >
But we do already have a whole class of syscalls on x86-64 to determine if the syscall is x86-64 or x32
12:14
<
robmur01 >
x32 is a whole other mess though
12:15
<
HdkR >
I've heard that one nasty solution would be to have a custom memory allocator for my entire application and thieve that full 64bit address space, but leave the 32bit space alone for ioctls to be forced to stick memory in the low end
12:16
<
HdkR >
But that's a lot of work and I don't know what new problems it'll pose
12:31
<
robmur01 >
I guess what you really want is the thread flag/address space/etc. bits of ILP32, but without the syscall mess that everyone hates, since presumably the code for marshalling syscalls between the emulation and the host
*is* actually LP64
12:34
<
HdkR >
A lot of the random state tracking needs to be emulated anyway, since x86->ARM64, just something as opaque as ioctl is hard
12:47
<
HdkR >
macc24: Userspace x86-64 (and soon 32bit x86) emulator for running Linux applications on an ARM64 host
12:47
<
macc24 >
how fast is it?
12:48
<
HdkR >
Needs to be faster. Some games run full speed
12:48
<
macc24 >
bookmarked
12:50
<
HdkR >
We're never going to receive native ports of most any game on Steam. Need an emulator that is worthwhile to run them on ARM
12:50
<
macc24 >
yeah i already heard of box86 and tested it a bit
12:51
<
HdkR >
FEX does a lot of stuff like box86, but we are going to cover 32bit and 64bit x86
12:52
<
HdkR >
Just have additional pains since we are running 32bit apps in a 64bit environment
12:52
<
macc24 >
doesn't linux support running 32bit binaries on 64bit host without issues?
12:53
<
HdkR >
Sure, if you're wanting to run x86 apps on x86-64 or AArch32 apps on ARMv8
12:54
<
HdkR >
But x86 on ARM64 is a different beast :P
12:55
<
HdkR >
It's also not a good idea to support AArch32 as the host for FEX, we lose all the new ISA improvements
12:55
<
macc24 >
b-but my 2015 laptop isn't arm64
12:56
<
HdkR >
Your 2015 laptop isn't the target audience :P
12:57
<
HdkR >
Emulating strong memory model x86 on ARMv8.1 is already a pain, don't want to make it too much worse
12:57
<
HdkR >
and....ugh, unaligned atomics
12:59
<
HdkR >
Fun enough, ARMv8.6 solves a bunch of the strong memory model problems. Sadly we won't have ARMv8.6 until post-Hercules
13:00
<
HdkR >
Doesn't solve unaligned atomics that cross cachelines, but can't win them all
13:05
<
macc24 >
post-what?
13:06
<
HdkR >
Hercules is Cortex-A789
13:06
* macc24
is waiting for arm cortex-a2137
13:07
<
macc24 >
ok.... does FEX run quickly on cortex-a73?
13:07
<
HdkR >
and Apple's SoC cheats by having an MSR that enables global strong memory model for the process
13:07
<
HdkR >
ehhh depends on what you're running
13:08
<
HdkR >
It really wants A75 minimum
13:09
<
HdkR >
Almost all x86 instructions turn in to atomic loadstores
13:09
<
HdkR >
and you need ARMv8.1 for atomic memory ops to clean up a lot of them
13:13
<
macc24 >
so it wouldn't work on a73
13:13
<
HdkR >
It'll run on base ARMv8.0, but don't expect decent perf :P
13:14
<
macc24 >
how fast will it run on just armv8?
13:15
<
HdkR >
Well it won't run supertuxkart full speed atm
13:17
<
macc24 >
did it use llvmpipe?
13:17
<
HdkR >
There's still work to go for getting it full speed
13:17
<
HdkR >
GL driver is unrelated
13:18
<
HdkR >
It'll use the native x86-64 libGL, or with thunks it'll use the host libGL
13:20
<
HdkR >
I've tested freedreno, turnip, radeonsi, radv, and i965 through the emulated layer. Seemed fine
13:21
<
robmur01 >
how about v8.0, but at 3.3GHz? :D
13:21
<
HdkR >
Someone else tested the Nvidia blob on the x86 host and it seemingly worked
13:21
<
macc24 >
robmur01:
*beats 3.3GHz with 3.4GHz*
13:22
<
HdkR >
If it is 3.3Ghz and a Cortex-A53 then it'll be pretty bad
13:23
<
macc24 >
HdkR: how about 2GHz a73?
13:24
<
robmur01 >
well, a 3.3GHz A53 can probably perform as well as you like, because it's already in fantasy-land :P
13:24
<
robmur01 >
the eMAG, however, is real
13:24
<
macc24 >
robmur01: whats eMAG
13:25
<
HdkR >
Anything in-order would have a rough time, I don't plan on having instruction scheduling at all
13:25
<
macc24 >
well, Cortex A73 is out of order
13:27
<
HdkR >
Give it a couple of months and maybe I'll have a real "user experience" response
13:28
<
HdkR >
But that is also the expectation of atleast ARMv8.1
13:30
<
HdkR >
My personal min-spec for CPU is still Cortex-A78 though. Anything less than that is just sad ;)
13:31
<
macc24 >
do A78 SBCs even exist?
13:32
<
HdkR >
Samsung will have one in a couple months
13:33
<
HdkR >
Nvidia will have one late next year
13:33
<
HdkR >
QCom probably in a few months
13:34
Net147_ has joined #panfrost
13:34
Net147 has quit [*.net *.split]
13:34
urjaman has quit [*.net *.split]
13:34
klaxa has quit [*.net *.split]
13:35
<
macc24 >
can't wait until there's SBC with it
13:35
urjaman has joined #panfrost
13:35
<
HdkR >
I'll definitely be buying a few Nvidia Orin boards
13:36
<
macc24 >
isn't nvidia anti-open source?
13:37
<
HdkR >
Sure, I'll be annoyed every moment of using it. Nobody else making a SoC with 12 Cortex-A78 cores though
13:38
<
macc24 >
i hope that there will be Orin SBC with PCI-E available
13:38
<
HdkR >
Should be, Xavier already did
13:38
<
macc24 >
oh then maybe it can replace my x86 PC
13:39
<
HdkR >
Maybe. Carmel is a bit derpy CPU core
13:41
<
macc24 >
well if it's faster than core 2 quad i will probaby buy it
13:42
<
HdkR >
At least it has core count. It's performance is all over the place depending on workload
13:42
<
macc24 >
all over the place as in?
13:43
<
HdkR >
Worst case its performance can be around Cortex-A55 levels
13:43
<
HdkR >
But on average its more like...A76?
13:44
exit70[m] has quit [Ping timeout: 246 seconds]
13:44
clementp[m] has quit [Ping timeout: 240 seconds]
13:44
l-as has quit [Ping timeout: 268 seconds]
13:44
Ke has quit [Ping timeout: 268 seconds]
13:44
wiizzard has quit [Ping timeout: 268 seconds]
14:01
robmur01 has quit [Remote host closed the connection]
14:02
robmur01 has joined #panfrost
14:05
kinkinkijkin has joined #panfrost
14:16
exit70[m] has joined #panfrost
14:17
wiizzard has joined #panfrost
14:22
clementp[m] has joined #panfrost
14:25
archetech has joined #panfrost
14:28
davidlt has joined #panfrost
14:29
Ke has joined #panfrost
14:29
l-as has joined #panfrost
14:39
<
macc24 >
kinkinkijkin: got a duet?
14:41
<
kinkinkijkin >
not yet, best buy had an issue
14:47
<
macc24 >
there might be a debian derivative for duet Soon™
16:35
Elpaulo has joined #panfrost
17:40
italove has quit [Ping timeout: 240 seconds]
18:29
kinkinkijkin has quit [Remote host closed the connection]
19:50
raster has quit [Quit: Gettin' stinky!]
19:54
Net147_ is now known as Net147
20:33
rak-zero has quit [Ping timeout: 264 seconds]
20:44
uis has joined #panfrost
20:54
davidlt has quit [Ping timeout: 246 seconds]
20:56
raster has joined #panfrost
22:07
raster has quit [Quit: Gettin' stinky!]
23:10
Green has quit [Ping timeout: 256 seconds]