<rqou>
whereas most linux installers are hacked iso-mbr-uefi hybrids
<lain>
or buy an isostick! :P
<rqou>
uefi seems to have made booting windows easier and booting linux much harder
<rqou>
also i'm installing on a super ancient bios machine
<lain>
rufus will do it
<rqou>
why is windows not a magic mybrid iso?
<rqou>
*hybrid
<lain>
I've never really looked into it
<rqou>
how come after all of this wintel bullshit booting is still so hard?
<lain>
because you're on an old bios machine :>
<lain>
but mostly: bioses are poorly written
<lain>
this also applies to uefi
<rqou>
IME uefi is worse
<lain>
>pc firmware< is poorly written
<rqou>
er, Apple not-U EFI is worse :P
<lain>
as I understand it, bios/uefi vendors typically see if the latest windows iso boots, and if so, they ship it :P
<lain>
there's not a whole lot of testing done
<rqou>
IME: pure BIOS --> works, pure UEFI --> works, any hint of touching CSM --> blows up
<rqou>
and apple not-U EFI --> blows up
<rqou>
apparently apple EFI is really weird though
<rqou>
er, rufus seems to be a windows tool
<rqou>
i'm on linux
<rqou>
according to stackoverflow ms-sys should work?
<rqou>
i guess i should appreciate how much effort went into making linux installers
<rqou>
either way it seems bootstrapping from "not having windows" to "having windows" isn't well tested in general
<rqou>
i guess MS just doesn't care about that
scrts has joined ##openfpga
pie_ has joined ##openfpga
<lain>
again, if you extract an iso, it works on uefi. I think it works on some bios? but it's been years since I had to care about that :P
<lain>
also rufus should work in wine, fwiw
scrts has quit [Ping timeout: 252 seconds]
<pie_>
?
<pie_>
if you put the contents of an iso on a fat filesystem and mark it bootable yeah
<pie_>
(i think)
<pie_>
something something
<lain>
yeah
<pie_>
probably need an appropriate bootloader on it
<lain>
for uefi I don't think the mbr matters, but for bios I forget what the procedure is
<lain>
I just use an isostick :V
<pie_>
i just use unetbootin
<pie_>
>.>
<lain>
yeah
<lain>
that works too
<lain>
was trying to think of the name of that one :P
<lain>
rqou: use unetbootin
<lain>
or does that work for windows isos?
<lain>
I don't remember
<lain>
whatever :P
<rqou>
got it working with ms-sys
<rqou>
now I see the value of your product :p
<lain>
hehe
<lain>
speaking of which, I need to dump all the (horrible) source code on github
<lain>
it's so bad :<
<rqou>
it's nice to see that MS fixed their magic sector dependencies some time in the last decade or so
<rqou>
for extra fun this machine no longer has an ODD because it's on indefinite loan to a friend of mine
<lain>
haha
<rqou>
within my friend group I'm the "has excess hardware" guy :P
m_t has joined ##openfpga
<rqou>
wtf Windows changed oobe again
azonenberg_work has quit [Ping timeout: 252 seconds]
azonenberg_work has joined ##openfpga
<rqou>
the Windows oobe feels so condescending
qu1j0t3 has quit [Ping timeout: 260 seconds]
scrts has joined ##openfpga
<rqou>
wtf Mozilla why is the Firefox default install still 32-bit
<lain>
didn't they have trouble doing 64bit builds?
<lain>
I remember something about the 64bit build requiring more ram than they had available on the build boxen or something, due to C++ template abuse
<lain>
or maybe I'm thinking of something else
<lain>
it was a long time ago, in any case
<rqou>
bullshit
<rqou>
I've used the nightlies a decade ago
<rqou>
and win64 Firefox is an official release now
<lain>
like I said, it may have been something else
<lain>
I thought it was firefox though
<lain>
and again this was like 10+ years ago :P
<rqou>
I think they did blow up the build boxen at one point
<rqou>
but right now win64 is "officially" supported
<whitequark>
right now that's USB2 so 720p max, but looks like GigE is in development
<whitequark>
ah no 720p60 or 1080p30 max
<whitequark>
and 1080p60 on series 7
* felix_
hopes that he'll be working for the hdmi2usb/axoim projects this year; that depends on getting some funding though...
scrts has quit [Ping timeout: 255 seconds]
pie_ has joined ##openfpga
<rqou>
I'm aware of the hdmi2usb project
<rqou>
the current iteration like you said doesn't have enough bw
<rqou>
1GbE also isn't fast enough for uncompressed 1080p60
scrts has joined ##openfpga
<rqou>
even usb3 (ahem, USB 3.1 gen 1 (5gbps)) isn't enough
<rqou>
er, wait NVM it might just barely be fast enough
<rqou>
either way azonenberg has convinced me that 10gbe is superior
<rqou>
hrm hdmi2usb is even more underpowered than I thought
<rqou>
it can't even do 720p30 uncompressed
<rqou>
480mbps raw is just way too little bw
<rqou>
720p60 already exceeds 1gbps
<rqou>
and I personally consider that the bare minimum
<rqou>
( (1280x720) pixels/frame * 24 bits/pixel * 60 frames/sec is already at least 1.3 gbits/sec)
<rqou>
hrm, unfortunately 4Kp60 just barely exceeds 10gbe
<whitequark>
um
<whitequark>
what?
<whitequark>
why the hell are you transferring it in RGB and not say YUYV?
<rqou>
because devices output RGB?
<whitequark>
oh right hdmi
<whitequark>
well
<whitequark>
just convert it to YUYV :p
<rqou>
er, i think some devices output YUYV
<rqou>
but that can't save you bits in general
<rqou>
if you're source is RGB888 then each pixel must have at least 2^24 possible values
<whitequark>
why not?
<whitequark>
you lose half of chroma resolution
<rqou>
right, but i don't want to do that
<rqou>
so e.g. if i have alternating one-pixel-wide columns of red/blue/red/blue YUYV turns it all purple
<rqou>
this is also why i budget for uncompressed data because even if you do have compression there will exist almost by definition *some* input stream that is incompressible
<rqou>
i don't want a "works most of the time" capture device
<rqou>
but yeah, overall i don't want any of the stupid bullshit hacks coming from the "professional video" world
<whitequark>
well, yese, I've precisely suggested YUYV because it does not have any pathologic input
<whitequark>
like, not almost
<whitequark>
there will by definition exist some input that will EXPAND after compression
<rqou>
what about red/blue stripes?
<whitequark>
choose one: accurate representation of red/blue stripes, 4k@60fps
<azonenberg>
To be clear
<whitequark>
personally I am thoroughly uninterested in the former *and* the latter at the same time
<whitequark>
like there is literally no circumstances where that's useful
<azonenberg>
YUV 4:4:4 does not lose resolution, it's an alternate color space
<whitequark>
if you HAVE to debug something just lower the fps
<azonenberg>
YUV 4:2:2 loses resolution
<whitequark>
azonenberg: I meant 4:2:2 (I think)
<whitequark>
hence YUYV
<azonenberg>
ah true
<whitequark>
rqou: it's not like you give up ever streaming RGB if you choose to stream YUYV for certain selections of resolutions and framerates
<rqou>
sure
<rqou>
i don't know if it even matters that much because i don't believe the artix-7 is fast enough to capture 4k input at either 30fps or 60fps
<rqou>
i believe it can do 1080p@60 though
digshadow has quit [Ping timeout: 240 seconds]
<rqou>
azonenberg: how nonstandard is "bonded two 10gbe links together?"
<rqou>
er, without using LACP or any "real" protocol
amclain has joined ##openfpga
<azonenberg>
rqou: highly nonstandard
<azonenberg>
now, FOUR bonded 10gbe links
<azonenberg>
is basically 40gbe
<rqou>
right :P
<azonenberg>
so you'd be best off using the actual 40gbe bonding protocol
<azonenberg>
also when you say two bonded
<azonenberg>
Do you mean packet-level striping a la 802.3ad
<azonenberg>
or byte-level striping a la PCIe
<azonenberg>
b/c i think the latter is what 40g does
<rqou>
ah i see
<rqou>
(still waiting for 25gbe/50gbe/400gbe to make a mess of things even further :P :P )
digshadow has joined ##openfpga
<azonenberg>
lol
<azonenberg>
i think those are 1, 2, 8 lanes of 25 Gbps
<azonenberg>
probably 130/128b coding if i had to geuss
<azonenberg>
but i dont know for sure
<rqou>
that might be right for 25/50 but not for 400
<azonenberg>
1g is 8/10, 10g is 64/66
<rqou>
400 is still in bikeshedding :P
<azonenberg>
ah i see
<azonenberg>
xilinx has 28 Gbps transceivers on their high ned 7 series stuff
<azonenberg>
so i assumed that was designed to work with 25gbe
<rqou>
yeah probably
<rqou>
like how the kintex-7 has a max transceiver speed of exactly 10.3125 gbps :P
<azonenberg>
Lol
<azonenberg>
Yeah that is no accident
massi_ has joined ##openfpga
massi_ has quit [Remote host closed the connection]
digshadow has quit [Quit: Leaving.]
digshadow has joined ##openfpga
scrts has quit [Ping timeout: 240 seconds]
scrts has joined ##openfpga
<whitequark>
rqou: what does "是日光节约时间" say
<whitequark>
it has the odd distinction of being longer than the non-translated "isDST" identifier
<rqou>
on the other hand i've never actually heard anybody use this term
<rqou>
是 - is; 日光 - daylight; 节约 - to conserve; 时间 - time
<whitequark>
oh.
<rqou>
also wtf CN/HK don't do DST
<whitequark>
it's a localized identifier in a programming language
<rqou>
ah ok
<rqou>
that seems quite silly, but it works better than what i usually hear people say to refer to DST
<rqou>
usually i just hear people say "改鐘" "to change clocks"
<rqou>
which isn't specific enough
<whitequark>
oh
<rqou>
er, i usually hear "overseas people who live in the SF Bay Area" say that
<rqou>
i have no idea how people in HK would refer to this american silliness :P
<rqou>
although china has their own time zone problems too
<rqou>
e.g. urumqi is (officially) on the same time zone as beijing despite being quite far away
massi has quit [Remote host closed the connection]
<rqou>
there's a youtube video that mentions that if you hypothetically (you actually can't do this) cross the border from china to afghanistan you get to jump 3.5 hours in time zones
<whitequark>
why can't you do that?
<rqou>
idk geopolitical issues?
<rqou>
wikipedia says "largely due to unrest in its far western province of Xinjiang, which borders the corridor."
<rqou>
this isn't new; the pass has been closed for decades
<whitequark>
ah
<rqou>
also, the china-afghanistan border is literally just this mountain pass and i'm pretty sure most people don't even realize the countries even share a border at all
m_t has joined ##openfpga
<rqou>
chinese land borders mostly seem to be quite in the middle of nowhere
<azonenberg_work>
Lol
<azonenberg_work>
China was tired of being invaded back in the ancient times
<azonenberg_work>
So they expanded until most of their borders were impassable in one way or another
<azonenberg_work>
Himalayas, oceans, etv
<azonenberg_work>
etc*
<azonenberg_work>
and when that didn't work they built the wall