<jrtc27>
looks to be 玄铁C910用户手册.pdf, the user manual
<jrtc27>
(the other one is apparently an integration manual, ie for vendors)
indy has joined #riscv
<mmind00>
jrtc27: thanks for that pointer
hexteex has quit [Ping timeout: 276 seconds]
indy has quit [Ping timeout: 260 seconds]
indy has joined #riscv
hexteex has joined #riscv
hexteex has quit [Remote host closed the connection]
[itchyjunk] has joined #riscv
freakazoid343 has joined #riscv
freakazoid12345 has quit [Ping timeout: 258 seconds]
freakazoid12345 has joined #riscv
freakazoid343 has quit [Ping timeout: 258 seconds]
hendursa1 has quit [Quit: hendursa1]
hendursaga has joined #riscv
<kehvo>
Woo, finally got my longan nano
<kehvo>
Now to figure out what to do with it
frost has quit [Quit: Connection closed]
ASSISTENZA has joined #riscv
ASSISTENZA has quit [Excess Flood]
<xentrac>
that's wonderful!
<xentrac>
jrtc27: that's also wonderful! a lot of Chinese chipmakers don't bother to make English documentation anymore because almost everybody who designs circuits speaks Chinese
<xentrac>
(especially commercially; hobbyists are more diverse)
ASSISTENZA has joined #riscv
ASSISTENZA has quit [Excess Flood]
<jrtc27>
this was a hard requirement from gnu people to accept patches using the custom extensions
<xentrac>
to have them documented in English?
<jrtc27>
I believe
<jrtc27>
at least the features they wanted to upstream support for
* jrtc27
never read the actual thread, just heard reports of it
<xentrac>
google translate came to the rescue it seems
<xentrac>
also though the T-HEAD folks seem to have been pretty open and collaborative, a welcome contrast from the last 30 years of Broadcom and NVIDIA and whatnot
<xentrac>
apparently a bunch of Chinese cellphone companies recently came into compliance with the GPL
<xentrac>
Naomi Wu walked into UMIDIGI's office in her underwear and filmed herself kind of yelling at them, and eventually after some evasion ("No, Ben doesn't work here anymore") they agreed to publish their source
<xentrac>
and then other companies started to follow suit
<jrtc27>
they've been collaborative *now* but where was that when they were adding their own spec-violating page table attributes custom "extension"
<xentrac>
haha
<jrtc27>
so yeah, glad they're not broadcom, but they're hardly a poster child for doing things right
<xentrac>
well, you know a lot more about the history of the situation than I do, I was just going by what's in that thread
<xentrac>
how does the extension violate the spec?
freakazoid343 has joined #riscv
<la_mettrie>
kehvo: i find it strange that longan nano has a microSD port but there seems to be no any instructions how to use it ...i suppose one could insert certain kind of image written on a microSD and use the board that way but...
freakazoid12345 has quit [Ping timeout: 258 seconds]
<jrtc27>
xentrac: they used bits that were marked as reserved and the spec says must be zeroed by software
<jrtc27>
(and that are required for DMA coherence)
<xentrac>
oh, in the page table?
<la_mettrie>
Esmil: thanks, i'll take a look
<jrtc27>
yes
<xentrac>
is there a way they could have added page table attributes without stomping on reserved bits?
<xentrac>
I mean I normally think of ISA extensions as being the sort of thing that reserved bits are reserved *for*
<jrtc27>
the standard-compliant way is to reserve a new satp.mode value in the custom extension space
<jrtc27>
everything else is reserved for *standard* use, not vendor use
<xentrac>
aha, I see. thanks!
* xentrac
rereads the privileged spec
rlittl01_ has quit [Quit: -a- Connection Timed Out]
hendursaga has quit [Remote host closed the connection]
<xentrac>
hmm, here it says all the satp.mode values are "reserved for future use", not "reserved for vendor use"; maybe I'm looking at an outdated version of the spec?
rlittl01 has joined #riscv
hendursaga has joined #riscv
<xentrac>
no, the 20190608 version says the same thing
freakazoid343 has quit [Read error: Connection reset by peer]
freakazoid343 has joined #riscv
freakazoid12345 has joined #riscv
freakazoid343 has quit [Ping timeout: 265 seconds]
freakazoid333 has joined #riscv
freakazoid12345 has quit [Ping timeout: 265 seconds]
<kehvo>
la_mettrie: Right, that's a good excuse for a trip down the SoC docs then I guess :)
<kehvo>
(assuming they're not incomplete ha)
<kehvo>
Turns out the display runs in serial mode, and the bus with prescale of 8, sounds horrible lol
[itchyjunk] has quit [Quit: Leaving]
<xentrac>
what bandwidth does that work out to
<kehvo>
Approx 13.5 Mbit/s, with 160x80 resolution and 12 bits per pixel that amounts to a tad under 80 fps max, if I calculate correctly