Hi all,

We, the PLCT Lab and Tarsier Team[1], are about to start two projects
on July 1, 2021:
1. Porting Firefox to RV64GC, including spidermonkey w/ JIT support.
2. Porting Chromium browser and ChromiumOS to RV64GC.

All the roadmaps, development meetings, CI/BuildFarm, source codes and
issue trackers will be open in July. All the codes will be submitted
to upstreams eventually.

The motivation and necessity behind these projects is that we expect
there will be RISC-V laptops coming out next year. The ISCAS is
seeking to build 2000 RV64GC laptops before the end of 2022. I am
focusing on the software side. There are still a few 'big absence' in
the RISC-V software ecosystem which should be supported when the
hardwares is ready. The availability of mainstream browsers is
currently the highest demand we heard.

Collaborations are welcome! Feel free to name the 'big absence' in the
RISC-V software ecosystem so that we can do some porting work with
priority.

[1] Tarsier Team is a new team in ISCAS, which is focusing on porting
and optimizing Linux distributions for RISC-V.


--
Best wishes,
Wei Wu (吴伟)

HI Wei:
  I thought it was just a joke when you talk about the RISC-V laptop,
but now I realize you are actually serious about this ;-)

  I'm interested in ChromiumOS project, it's an aggressive fastforward.
one question, have you tried to talk to people from RIOS lab?
They seems have a similar goal to get a ChomiumOS working,
It would be great to combine the force and collaborate together.

Yixun

Hi Drew,


On Mon, Jun 7, 2021 at 2:04 AM Drew Fustini <pdp7pdp7@gmail.com> wrote:

That is very exciting!

Any information about the SoC the laptops would use?

Not yet. T-Head, Starfive, and Sifive are in my candidate list.

I would love to get invovled in the effort.

Great! Let's wait one week to collect enough feedbacks and start to
prepare the first coordinating/planning meeting.

Thank you,
Drew

On Sun, Jun 6, 2021, 10:08 Wei Wu (吴伟) <lazyparser@gmail.com> wrote:

Hi all,

We, the PLCT Lab and Tarsier Team[1], are about to start two projects
on July 1, 2021:
1. Porting Firefox to RV64GC, including spidermonkey w/ JIT support.
2. Porting Chromium browser and ChromiumOS to RV64GC.

All the roadmaps, development meetings, CI/BuildFarm, source codes and
issue trackers will be open in July. All the codes will be submitted
to upstreams eventually.

The motivation and necessity behind these projects is that we expect
there will be RISC-V laptops coming out next year. The ISCAS is
seeking to build 2000 RV64GC laptops before the end of 2022. I am
focusing on the software side. There are still a few 'big absence' in
the RISC-V software ecosystem which should be supported when the
hardwares is ready. The availability of mainstream browsers is
currently the highest demand we heard.

Collaborations are welcome! Feel free to name the 'big absence' in the
RISC-V software ecosystem so that we can do some porting work with
priority.

[1] Tarsier Team is a new team in ISCAS, which is focusing on porting
and optimizing Linux distributions for RISC-V.


--
Best wishes,
Wei Wu (吴伟)

--
Best wishes,
Wei Wu (吴伟)

Hi Yixun,

On Mon, Jun 7, 2021 at 6:01 AM Yixun Lan <yixun.lan@gmail.com> wrote:

HI Wei:
I thought it was just a joke when you talk about the RISC-V laptop,
but now I realize you are actually serious about this ;-)

Sure. It will always sound like a joke until we build one. I'm serious
and super excited about this idea :)

I'm interested in ChromiumOS project, it's an aggressive fastforward.

Great!

one question, have you tried to talk to people from RIOS lab?
They seems have a similar goal to get a ChomiumOS working,
It would be great to combine the force and collaborate together.

I've heard of the effort from RIOS Lab and had several talks with RIOS
in July, 2019. Unfortunately their work is not open source (yet?).

According to the report in Jan, 2021, RIOS was able to boot the linux
kernel, and start a shell running d8 on PicoRio board. No Graphics or
other important components were demonstrated. The PLCT Lab had done
similar work, porting AOSP kernel and shell to RV64GC Platform[1]. So
I am kinda sure that compared with the whole work volume, we still
have tons of work to do.

Definitely we should combine all the contributors we can find. I'll
try to talk to the engineers from RIOS again. :)

[1]: The PLCT Lab had stopped AOSP porting projects and turned to
support T-HEAD after the T-Head open source their full stack porting.

Yixun

--
Best wishes,
Wei Wu (吴伟)

On Mon, Jun 7, 2021 at 10:07 AM Morton Yu <morton.yu@starfivetech.com> wrote:

This is definitely a very promising direction.

Currently, we (StarFive) have a publicly available platform called StarLight, which is useful to support early software porting. We are busy working on a much higher performance Chip at Cortex-A76 level which is an even better fit for this application.

I have one BeagleV on my desk (thanks Drew!). It's promising. Looking
forward to collaborate with StarFive!

Please stay tuned.



Morton



发件人: <software@lists.riscv.org> 代表 Drew Fustini <pdp7pdp7@gmail.com>
日期: 2021年6月7日 星期一 上午2:04
收件人: "Wei Wu (吴伟)" <lazyparser@gmail.com>
抄送: "software@lists.riscv.org" <software@lists.riscv.org>
主题: Re: [RISC-V] [software] [Meta Project] Porting Firefox, Chromium/ChromiumOS to RV64GC, and the 2K RISC-V Laptop Project



That is very exciting!



Any information about the SoC the laptops would use?



I would love to get invovled in the effort.



Thank you,

Drew



On Sun, Jun 6, 2021, 10:08 Wei Wu (吴伟) <lazyparser@gmail.com> wrote:

Hi all,

We, the PLCT Lab and Tarsier Team[1], are about to start two projects
on July 1, 2021:
1. Porting Firefox to RV64GC, including spidermonkey w/ JIT support.
2. Porting Chromium browser and ChromiumOS to RV64GC.

All the roadmaps, development meetings, CI/BuildFarm, source codes and
issue trackers will be open in July. All the codes will be submitted
to upstreams eventually.

The motivation and necessity behind these projects is that we expect
there will be RISC-V laptops coming out next year. The ISCAS is
seeking to build 2000 RV64GC laptops before the end of 2022. I am
focusing on the software side. There are still a few 'big absence' in
the RISC-V software ecosystem which should be supported when the
hardwares is ready. The availability of mainstream browsers is
currently the highest demand we heard.

Collaborations are welcome! Feel free to name the 'big absence' in the
RISC-V software ecosystem so that we can do some porting work with
priority.

[1] Tarsier Team is a new team in ISCAS, which is focusing on porting
and optimizing Linux distributions for RISC-V.


--
Best wishes,
Wei Wu (吴伟)

--
Best wishes,
Wei Wu (吴伟)

Hi,

Em dom., 6 de jun. de 2021 às 19:08, Wei Wu (吴伟)
<lazyparser@gmail.com> escreveu:

The motivation and necessity behind these projects is that we expect
there will be RISC-V laptops coming out next year.

Huh, exciting news :-)

So my secret desire in 2014-2015 when I started to look at RISC-V,
that my next laptop would not be a x86-based one and hopefully have a
RISC-V one, might not be totally hopeless... but better be quick
because my current laptop is aging quickly :-)

The ISCAS is
seeking to build 2000 RV64GC laptops before the end of 2022. I am
focusing on the software side. There are still a few 'big absence' in
the RISC-V software ecosystem which should be supported when the
hardwares is ready. The availability of mainstream browsers is
currently the highest demand we heard.

Collaborations are welcome! Feel free to name the 'big absence' in the
RISC-V software ecosystem so that we can do some porting work with
priority.

Yes, you're totally right. So I've been working on this for a long
time, one of the main person (and the earlierst) pushing for a Debian
port.

There might be other missing pieces LibreOffice or Java/OpenJDK
(there's support and it's been available for years, but not with JIT
optimisations AFAIK), but the complexity of the browsers and
JavaScript engines and the absolutely need for them in basic desktop
usage is probably the top of the list. If anyone has other priorities
I'm also interested to know, to see what we can do about it.


Speaking of Debian, for a couple of years it's been in a plateau of
having ~95% of the archive built at all times, see
https://buildd.debian.org/stats/graph-ports-quarter-big.png , jumping
up from about 80% after gaining support for Go and Rust a while ago.
It's at the top of secondary architectures, right there with
PowerPC-64-bigendian, which benefits from the huge
PowerPC-64-littleendian. So that's more than 14K source packages
ready to use, probably more than 30K counting arch-independent
packages like libs of Python/Perl/etc.

Not sure about the current progress of others like Fedora, but I
assume that their level of support is about the same, everybody tries
to push support the the projects upstream so all distributions (or
kernels/OSs) benefit.

Finally, about Debian again, hopefully in the next months we will be
able to get some machines more suitable for datacentres and remote
administration, and we can start to look for "riscv64" to become one
of the main architectures, along with x86_64, arm64 and ppc64el among
others :-)


Cheers.
--
Manuel A. Fernandez Montecelo <manuel.montezelo@gmail.com>

Hi Manuel,

On Mon, Jun 7, 2021 at 3:52 PM Manuel A. Fernandez Montecelo
<manuel.montezelo@gmail.com> wrote:

So my secret desire in 2014-2015 when I started to look at RISC-V,
that my next laptop would not be a x86-based one and hopefully have a
RISC-V one, might not be totally hopeless... but better be quick
because my current laptop is aging quickly :-)

Sure! :-)

The ISCAS is
seeking to build 2000 RV64GC laptops before the end of 2022. I am
focusing on the software side. There are still a few 'big absence' in
the RISC-V software ecosystem which should be supported when the
hardwares is ready. The availability of mainstream browsers is
currently the highest demand we heard.

Collaborations are welcome! Feel free to name the 'big absence' in the
RISC-V software ecosystem so that we can do some porting work with
priority.

Yes, you're totally right. So I've been working on this for a long
time, one of the main person (and the earliest) pushing for a Debian
port.

Debian is one of the most 'ready for RISC-V' Linux distributions I've used.
Although I am not sure how to contribute to downstream/distribution like
debian yet, I will try when the meta project gets started.

Looking forward to collaborate with you (and all other debian contributors) :)

There might be other missing pieces LibreOffice or Java/OpenJDK
(there's support and it's been available for years, but not with JIT
optimisations AFAIK), but the complexity of the browsers and
JavaScript engines and the absolutely need for them in basic desktop
usage is probably the top of the list. If anyone has other priorities
I'm also interested to know, to see what we can do about it.

The JIT for RISC-V in Java/OpenJDK is kinda ready. Check
https://github.com/openjdk-riscv/
for more details. And there is going to have a github/riscv/openjdk repo.
Huawei open sourced the initial support for RV64G. and Alibaba has
ported their codes for RV64GV JIT and GC on Huawei's codebase.
ISCAS is working on adding JIT support for RV32GC.

My colleagues had evaluated the RV64G port on Hifive Unleashed,
using SPECjvm98 benchmark. It works well with one or two failures.
Feel free to try it and open issues when you meet bugs/crashes.

For the LibreOffice, I will add it to my todo list. I have not dug
into it before.

V8 is ready for RV64GC. So does NodeJS (soon, the next release).
Spidermonkey is planned in the roadmap 2021 of PLCT Lab.

Speaking of Debian, for a couple of years it's been in a plateau of
having ~95% of the archive built at all times, see
https://buildd.debian.org/stats/graph-ports-quarter-big.png , jumping
up from about 80% after gaining support for Go and Rust a while ago.
It's at the top of secondary architectures, right there with
PowerPC-64-bigendian, which benefits from the huge
PowerPC-64-littleendian. So that's more than 14K source packages
ready to use, probably more than 30K counting arch-independent
packages like libs of Python/Perl/etc.

Not sure about the current progress of others like Fedora, but I
assume that their level of support is about the same, everybody tries
to push support the the projects upstream so all distributions (or
kernels/OSs) benefit.

Yes. Fedora/Gentoo are also blocked by these key components
mentioned above.

Finally, about Debian again, hopefully in the next months we will be
able to get some machines more suitable for datacentres and remote
administration, and we can start to look for "riscv64" to become one
of the main architectures, along with x86_64, arm64 and ppc64el among
others :-)

Great! By the way, the RVI is planning to build a few RISC-V Labs for
hosting CI/Testing jobs for all open source communities. the first & biggest
Lab would have more than 2k RISC-V boards available for testing purposes.
Hopefully the RISC-V Labs will be ready/online before the end of July.

Cheers.
--
Manuel A. Fernandez Montecelo <manuel.montezelo@gmail.com>

--
Best wishes,
Wei Wu (吴伟)

Hi all,

After talking with Han Mao (the chair of Android SIG) we realized that
there are lots of common open source components in Android and
ChromiumOS/Chromium porting efforts. Han Mao would like to help us
put the porting & optimizing of common parts under Android SIG.

--
Best wishes,
Wei Wu (吴伟)

So I believe that the meta project has just reached its first
milestone: Firefox[1] and Chromium[2] can run on the riscv64 boards
now. (to be clear, these success were not driven by me or the PLCT
Lab. Lots of amateurs and developers from different communities made
these happen. Really amazing!)

And this is just the beginning. From now on, I believe the speed of
development will be significantly accelerated.

[1] https://groups.google.com/a/groups.riscv.org/g/sw-dev/c/81caeTrQWLs
[2] https://groups.google.com/a/groups.riscv.org/g/sw-dev/c/PCGVIH2dDJg

--
Best wishes,
Wei Wu (吴伟)