shamoe has quit [Quit: Connection closed for inactivity]
jacklsw has quit [Ping timeout: 255 seconds]
vrldvibeveb has joined #riscv
vrldvibeveb has left #riscv [#riscv]
jacklsw has joined #riscv
<courmisch>
I think maybe there is a dispropportionate emphasis on vectors among enthusiasts because it's one of those things that's not just recompiling
<courmisch>
the others being kernel and (re)compilers
solol has joined #riscv
davidlt has joined #riscv
mlw has quit [Ping timeout: 240 seconds]
sukrutb has quit [Ping timeout: 256 seconds]
jfsimon1981_b has quit [Remote host closed the connection]
BootLayer has joined #riscv
esv has quit [Ping timeout: 268 seconds]
esv has joined #riscv
mlw has joined #riscv
Bluefoxicy has quit [Ping timeout: 255 seconds]
Bluefoxicy has joined #riscv
prabhakarlad has joined #riscv
solol has quit [Quit: Client closed]
danilogondolfo has joined #riscv
EchelonX has joined #riscv
EchelonX has quit [Client Quit]
EchelonX has joined #riscv
stylefish has joined #riscv
stylefish has quit [Remote host closed the connection]
jacklsw has quit [Ping timeout: 260 seconds]
m5zs7k has quit [Ping timeout: 264 seconds]
aburgess has quit [Ping timeout: 252 seconds]
m5zs7k has joined #riscv
pecastro has joined #riscv
esv has quit [Ping timeout: 255 seconds]
foton has quit [Quit: %Bye, bye, ...%]
esv has joined #riscv
esv_ has joined #riscv
esv has quit [Ping timeout: 268 seconds]
aburgess has joined #riscv
esv_ is now known as esv
m5zs7k has quit [Ping timeout: 268 seconds]
foton has joined #riscv
m5zs7k has joined #riscv
<gurki>
courmisch: a good vector implementation is a strict requirement if you want to compete performance wise. depending on your background you might care more or less ofc :)
<gurki>
ofc these vector units dont help if you cannot keep them fed due to other problems
Leopold has quit [Remote host closed the connection]
Leopold has joined #riscv
<courmisch>
gurki: that depends on the workloads or benchmarks. there are plenty of tasks that get little or no benefits from vectors, or that can be accelerated some other ways (multithread, DSP, GPGPU...)
<courmisch>
so I don't think it's *generally* a strict requirement
<sorear>
it's a nice to have for users, but a strict requirement for managers because the competition has it
psydroid has joined #riscv
<courmisch>
ofc it's nice to have and in some cases necessary even
<courmisch>
but I feel that it gets overrepresented here and on the subreddit because, unlike the other ways to accelerate computational works, it is tied to the ISA
<courmisch>
and I am partially at fault myself
prabhakarlad has quit [Ping timeout: 250 seconds]
prabhakar has quit [Ping timeout: 252 seconds]
solol has joined #riscv
<sorear>
kind of sad that 29 years after VIS we still have that tight coupling and lack of standardization
solol has quit [Quit: Client closed]
<geertu>
sorear: It's valuable IP. no one wants it standardized :-(
aburgess has quit [Ping timeout: 252 seconds]
foton has quit [Quit: %Bye, bye, ...%]
foton has joined #riscv
foton has quit [Remote host closed the connection]
prabhakarlad has joined #riscv
foton has joined #riscv
foton has quit [Client Quit]
cousteau has joined #riscv
foton has joined #riscv
BootLayer has quit [Quit: Leaving]
cousteau_ has joined #riscv
cousteau has quit [Read error: Connection reset by peer]
aburgess has joined #riscv
foton has quit [Quit: %Bye, bye, ...%]
cousteau_ has quit [Quit: ♫ I can't forget the day I shot that network down ♫]
<clemens3>
and notebookcheck maybe went a bit over the top
BootLayer has joined #riscv
davidlt has quit [Ping timeout: 276 seconds]
prabhakarlad has joined #riscv
solol has joined #riscv
<bjdooks>
no-one has yet learned forking stuff is hard, and software isn't free
<jn>
this article is saying "RISC-based technology" as if it was something even remotely new -_-
<bjdooks>
or even reduced these days
<jn>
and also (unsurprisingly) doesn't even start to formulate a concept of *what* the US would or should ban exactly
<jn>
"hello china, you can't implement this spec now, thanks"? or what
<jn>
the VisionFive2 amazon link is just funny though. completely missed the opportunity to point out that you, dear reader, can be part of this story and buy a RISC-V chip/board made in china
<jn>
it even would have been an opportunity to discuss details: AFAIK StarFive SoCs are based on some CPU cores licensed from SiFive, right?
<bjdooks>
I think so, Sophgo have also licensed them
<bjdooks>
a ban would hurt sifive more than anyone else at this point
<jn>
yeah, if implementations are banned, chinese SoC companies could just switch to Alibabas implemention
<jn>
if specs are banned.... does that even work?
<jn>
(or T-Head, i should say)
<bjdooks>
i'm guessing that means future IP transfer, banning a spec is difficult, banning say sifive from selling to .cn is much easier
<jn>
something that this garbage article really should have mentioned as an important distinction, because it is what makes RISC-V even notable and different from ARM or MIPS
Noisytoot has quit [Remote host closed the connection]
<bjdooks>
and arm64 is basically mips anyway
<sorear>
there's been some fearmongering that US persons and entities will be forbidden from exporting technology by contributing to the spec, but it's all meaningless without details and primary sources
Noisytoot has joined #riscv
heat has joined #riscv
<sorear>
love how the most recently linked article mentions Sunway as if it was in any way relevant to rv
jacklsw has joined #riscv
<conchuod>
I think it deliberately misleads sorear
<solol>
wytb sorear?
<bjdooks>
colloquially - clickbait
<solol>
Japan/Korea/Arabian countries/Russian countries don't accept US money so they're probably the places...
<solol>
That was news a couple months ago.
solol has quit [Quit: Client closed]
<courmisch>
not a legal expert by any means whatsoever, but I doubt that the US could restrict their citizen contributing to specifications unless it repeals the first amendment. It's not the same as exporting.
<courmisch>
I do suppose that they could restrict licensing of IPs partly or wholly american
shamoe has joined #riscv
<courmisch>
FOSDEM talk submitted. Now hoping that it gets rejected so I don't have to write the slides!
jmdaemon has quit [Ping timeout: 260 seconds]
<courmisch>
(of course implying that there should be better talks for the moderators to pick from)
<gurki>
never underestimate the clusterfuck of us export regulations :3
ntwk has quit [Read error: Connection reset by peer]
solol has joined #riscv
sakman_ has joined #riscv
sakman has quit [Ping timeout: 264 seconds]
sakman_ is now known as sakman
<muurkha>
courmisch: it'll be interesting to see
<courmisch>
muurkha: IIUC, RV only gets half a day, and if they go with their planned 45 minutes length, that will be like 4 talks... that's a high bar
<gurki>
good luck then :)
<courmisch>
more like don't hold your breath
<muurkha>
sorear: yeah, the author confusing "RISC" with "RISC-V" clearly indicates that the author has no idea what they're talking about
<muurkha>
it's probably worth mentioning in this context that PRC has not only their own ARM manufacturing but also their own amd64 manufacturing: Zhaoxin
<muurkha>
they aren't suitable for HPC, but the Sunway chips certainly are
<muurkha>
> China has a domestic policy to "replace all foreign hardware and software from its public infrastructure with homegrown solutions" by 2023 (the so-called 3–5–2 policy).
<clemens3>
the article also said china can't buy latest x86 chips, which imho is bs.. i know the 4090 is on the sanction list, which is kinda funny in itself, as lot's of kids have it, and they now can't sell to china, lol.. but x86_64 chips, are latest 5nm chips also on list? would surprise me actually
<clemens3>
but anyway, all bad news is good news for risc-v anyway
<davidlt>
clemens3, well, Loongson related companies are on US entity list, but there are also export restrictions from China too IIRC
<davidlt>
As soon as SOPHGO SG2042 showed up the clock started ticking