* Re: [libre-riscv-dev] next tasks
From: Hendrik Boom @ 2020-03-12 14:32 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <20200220093921.635c47f72ea2faf85dd352b7@gmx.com>
On Thu, Feb 20, 2020 at 09:39:21AM +0200, Lauri Kasanen wrote:
> On Wed, 19 Feb 2020 14:14:16 +0000
> Luke Kenneth Casson Leighton <lkcl@lkcl.net> wrote:
>
> > we only have a few months in which to hang together all of the pieces
> > that have been put together, to make a 64-bit 180nm chip. because
> > we're not doing this in verilog i believe it is reasonably doable, if
> > we take steps to cut "unnecessary" things out i.e. not do too much.
> > chop out L2 cache for example.
>
> What about the C/C++ simulator/interpreter timeline? It's not a blocker
> for the hw tapeout, but it is a blocker for my work, which probably
> mostly won't make it for the simple Oct tapeout anyway.
What is it you want with this? An implementation of C/C++ for the new SoC?
Or a simulation of the new SoC instruction set in C/C++? Would it be good
enough to have a simulator of the new SoC in any widely available programming
language?
-- hendrik
>
> - Lauri
>
> _______________________________________________
> libre-riscv-dev mailing list
> libre-riscv-dev@lists.libre-riscv.org
> http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 158] NLNet 2019 Formal Correctness Proofs toplevel 2019-10-032
From: bugzilla-daemon @ 2020-03-12 14:39 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-158-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=158
--- Comment #5 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
Joost, hi,
please see in http://bugs.libre-riscv.org/show_bug.cgi?id=158#c4 the
"Schedule A" to be attached to the Memorandum of Understanding for
the Formal Correctness Proofs Proposal, 2019-10-032
many thanks.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] next tasks
From: Luke Kenneth Casson Leighton @ 2020-03-12 14:43 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <20200312143244.g6ns5j4am2jf3yhl@topoi.pooq.com>
On Thu, Mar 12, 2020 at 2:32 PM Hendrik Boom <hendrik@topoi.pooq.com> wrote:
>
> On Thu, Feb 20, 2020 at 09:39:21AM +0200, Lauri Kasanen wrote:
> > On Wed, 19 Feb 2020 14:14:16 +0000
> > Luke Kenneth Casson Leighton <lkcl@lkcl.net> wrote:
> >
> > > we only have a few months in which to hang together all of the pieces
> > > that have been put together, to make a 64-bit 180nm chip. because
> > > we're not doing this in verilog i believe it is reasonably doable, if
> > > we take steps to cut "unnecessary" things out i.e. not do too much.
> > > chop out L2 cache for example.
> >
> > What about the C/C++ simulator/interpreter timeline? It's not a blocker
> > for the hw tapeout, but it is a blocker for my work, which probably
> > mostly won't make it for the simple Oct tapeout anyway.
correct. it's on a separate track.
> Or a simulation of the new SoC instruction set in C/C++?
to start from an existing power isa simulator (gem5 for example), that
simulator *happens* to be written in c or more likely c++, therefore
simulated vector and VPU instructions *have* to be written in c / c++.
otherwise we need to create - from scratch - an entire new POWER ISA
cycle-accurate instruction simulator, which would be a genuinely
stupid thing to spend time on given that there are a number of such
simulators in existence already. including qemu.
l.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] EU digital self-sufficiency talk
From: Lauri Kasanen @ 2020-03-12 15:26 UTC (permalink / raw)
To: libre-riscv-dev
Hi,
I read today that there's talk about EU's digital self sufficiency, and
it's likely there will be funding in the next 2021-2027 budget period.
That sounds encouraging for a project such as this. Would be good for
nlnet/etc to ping their EU contacts, making sure that a CPU design is
kept in mind just like web browsers and Linux distros.
One ref:
https://soylentnews.org/article.pl?sid=20/02/27/1057230
- Lauri
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 178] first coriolis2 tutorial, workflow and "test project" page
From: bugzilla-daemon @ 2020-03-12 16:15 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-178-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=178
--- Comment #208 from Jean-Paul.Chaput@lip6.fr ---
(In reply to Jean-Paul.Chaput from comment #207)
> Created attachment 35 [details]
> Coriolis2 ioring.py with explicit pad positioning
> > Second, I don't think that introducing an additional file with a special
> > format would do us any good. I suppose that 'coriolis2.ioring' is ok as it
> > is for passing pad configuration between stages. If I am just not aware of
> > the issues with it, please point me in a right direction.
>
> the issue with ioring.py "as-is", is that the procedure for laying out the
> pads is entirely automatic, spaced-out evenly, because it's designed
> *exclusively* for doing Quad Flat Packages (QFPs).
>
> what it can't do is something like this:
>
> https://cdn.sparkfun.com/r/600-600/assets/7/a/6/9/c/51c0d009ce395feb33000000.
> jpg
>
> that's an *uneven* distribution of the pads.
Wrong! ;-) You can specify the exact position of each pad with the ioring.py
file. I did put an example in attachement #35. It is not part of the toolkit
as the related example is the adder with AMS 350nm (under NDA).
I did not completely follow the thead about CSV file, but in my experience,
I don't see the need of introducing a new file format (even if its simple
CSV), directly write in Python instead.
> > What I did thought is necessary, is to eliminate the need for the user to
> > configure the pads at all, or at least to configure them separately from
> > top-level logic. I thought we can just set up formal rules (like putting
> > pins with names ending on '_i' to the left, '_o' − to the right), and
> > implement the reusable 'create_pads()' to inspect the list of pins according
> > to these rules and generate 'coriolis2.ioring'.
>
> hmmm.... that may actually be useful in some cases. i think... i think we
> need a class of some kind, with the ability to derive from it and create
> alternative layouts.
In my experience again, there is too many different cases so that an
automated
way works for for enough of them. And I'm very reluctant to put semantic in
the pad names, it may clash with the HDL in too many ways.
Lastly, if I'm not mistaken, you want to put information about the pads at
nMigen level. I think it should be avoided. nMigen is for the design
(behavioral or high leval description), and information pertaining to
the layout should be kept at Coriolis level.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 217] create a "ring" system which allows pad locations to be specified conveniently
From: bugzilla-daemon @ 2020-03-12 16:39 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-217-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=217
--- Comment #1 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
moving discussion from http://bugs.libre-riscv.org/show_bug.cgi?id=178#c208
(In reply to Jean-Paul.Chaput from comment #208)
> (In reply to Jean-Paul.Chaput from comment #207)
> > Created attachment 35 [details]
> > Coriolis2 ioring.py with explicit pad positioning
>
> > > Second, I don't think that introducing an additional file with a special
> > > format would do us any good. I suppose that 'coriolis2.ioring' is ok as it
> > > is for passing pad configuration between stages. If I am just not aware of
> > > the issues with it, please point me in a right direction.
> >
> > the issue with ioring.py "as-is", is that the procedure for laying out the
> > pads is entirely automatic, spaced-out evenly, because it's designed
> > *exclusively* for doing Quad Flat Packages (QFPs).
> >
> > what it can't do is something like this:
> >
> > https://cdn.sparkfun.com/r/600-600/assets/7/a/6/9/c/51c0d009ce395feb33000000.
> > jpg
> >
> > that's an *uneven* distribution of the pads.
>
> Wrong! ;-)
ah *ha*!
> You can specify the exact position of each pad with the
> ioring.py
> file. I did put an example in attachement #35. It is not part of the
> toolkit
> as the related example is the adder with AMS 350nm (under NDA).
okaay. so the positions are specified in coordinates along that edge.
excellent.
> I did not completely follow the thead about CSV file, but in my experience,
> I don't see the need of introducing a new file format (even if its simple
> CSV), directly write in Python instead.
the format of the ioring.py (which we didn't know about) covers it.
and, that leads me to investigate more thoroughly (grep -r "ioring.py"
src/coriolis-2.x) and i find this:
cumulus/src/plugins/chip/Configuration.py
ta-daaaa, ChipConf does pretty much (nearly) everything we need, jock.
which would leave just some "convenience" routines that help us to find
and construct the pad names, on which side.
> In my experience again, there is too many different cases so that an
> automated
> way works for for enough of them. And I'm very reluctant to put semantic in
> the pad names, it may clash with the HDL in too many ways.
yes this is why i suggested a pattern-match "helper" routine
> Lastly, if I'm not mistaken, you want to put information about the pads at
> nMigen level. I think it should be avoided. nMigen is for the design
> (behavioral or high leval description), and information pertaining to
> the layout should be kept at Coriolis level.
i was tempted to suggest putting the *chip* information (which pads, which
side) into the structure-of-classes-which-happens-to-be-in-nmigen
why? because the development is quite closely tied together
however a good case can be made for keeping the pads information completely
separate (in the soclayout repository)
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 217] create a "ring" system which allows pad locations to be specified conveniently
From: bugzilla-daemon @ 2020-03-12 16:57 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-217-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=217
--- Comment #2 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
so, jock, just a bit of background:
we have a *massive* class hierarchy (well over 250 class instances, all of
which need to be laid out), which will produce something like 500,000 gates,
which is far too much to do all in one go.
so we want to split down the "tree" of class instances into progressively
smaller blocks, not going completely mad about it, maybe two levels deep. this
will mean that the number of class instances ("blocks") that need to be placed
*by hand* is reduced down to maybe... 50 to 80.
the "top level" is called a "floor plan", and we need to best organise the
routing between all those "blocks", so that their inputs and outputs are all on
the "most efficient side".
the last thing we want, for example, is that to get the I/O to a particular
block, it has to go *all the way round* the outside of a block.
some of the data buses will be a whopping FOUR TO FIVE HUNDRED wires (or more),
which will be completely dominating the centre of the chip, branching out.
sometimes, for smaller blocks, we definitely want the wires to be evenly spread
out across the whole of the side of the block (just like in
benchs/nmigen/ALU16)
sometimes, we want a bus (address / data / control lines) to come in on a
*specific* location - the corner, or the middle - of a block, on one particular
side.
so this is what we need to be able to specify.
also, we need to ensure that the auto-router does *not* ignore the pads (which
will go on the very outer edge of each Block), and that means putting BLOCKAGEs
in, as "rings", around the outside of the Block.
jock: see /alliance-check-toolkit/benchs/nmigen/ALU16/doAlu16.py for an
example.
looking at class ChipConf: Jean-Paul, it is not clear whether ChipConf will
automatically create the full range of BLOCKAGEs nets. i am grepping the
source code "self.corona" and there is no evidence of self.corona actually
being set!
ChipConf.findPowerAndClockNets() *uses* self.corona, however there's no
*setting* of self.corona!
what gives?
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] EU digital self-sufficiency talk
From: Luke Kenneth Casson Leighton @ 2020-03-12 17:06 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <20200312172610.c8a9756e9352422d37f84f65@gmx.com>
On Thu, Mar 12, 2020 at 3:25 PM Lauri Kasanen <cand@gmx.com> wrote:
> I read today that there's talk about EU's digital self sufficiency, and
> it's likely there will be funding in the next 2021-2027 budget period.
> That sounds encouraging for a project such as this. Would be good for
> nlnet/etc to ping their EU contacts, making sure that a CPU design is
> kept in mind just like web browsers and Linux distros.
i forwarded to michiel and bob.
l.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 217] create a "ring" system which allows pad locations to be specified conveniently
From: bugzilla-daemon @ 2020-03-12 17:07 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-217-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=217
--- Comment #3 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(sotto voice: so for what we are doing, we don't use
plugins.chip.Configuration.loadConfiguration, we use
plugins.chip.Configuration.ChipConf *directly*.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 178] first coriolis2 tutorial, workflow and "test project" page
From: bugzilla-daemon @ 2020-03-12 17:14 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-178-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=178
Luke Kenneth Casson Leighton <lkcl@lkcl.net> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|IN_PROGRESS |RESOLVED
Resolution|--- |FIXED
--- Comment #209 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
i'm going to close this one as resolved as far as the tutorial part is
concerned (and we'll work out who gets what), however if it's useful to do so,
carry on discussing :)
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 217] create a "ring" system which allows pad locations to be specified conveniently
From: bugzilla-daemon @ 2020-03-12 17:24 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-217-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=217
--- Comment #4 from Jean-Paul.Chaput@lip6.fr ---
I did miss some posts.
self.corona is an accessor property (defined in Configuration.py,
line 663). It's the first element of the self.coronas list,
which initialized in the checkPads() function at line 1008.
Rule is simple: any instance which is not a pad is a corona.
And there should be only one.
I did lost a post about HDL langiage. No time to rewrite it right
now but this is a question we asked ourselves too. Until now,
we are not happy with anything that exists.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 178] first coriolis2 tutorial, workflow and "test project" page
From: bugzilla-daemon @ 2020-03-12 17:25 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-178-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=178
--- Comment #210 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
for this one, i'm going to suggest:
* EUR 250 to jock for the (crucial) questions and for helping get to buster
* EUR 200 to tobias for the early checking of the tutorial
* EUR 700 to be donated to LIP6.fr for the huge amount of bugfixing by JP
* EUR 150 to Staf for his insights
* EUR 1200 to me for the persistent head-banging against brick walls
Cole, as long as you are in Europe, you helped find some early bugs in the
tutorial page, and made some useful edits, however if you are outside of
Europe, most of what NLNet would transfer to you would go in banking fees. we
can if you like work something out.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 216] LOAD STORE buffer needed
From: bugzilla-daemon @ 2020-03-12 17:29 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-216-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=216
--- Comment #6 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
ok i think i have a data structure that will help. it's a (very small) CAM,
with extra "things".
also it critically depends upon a pre-analysis of the LD/ST operation, breaking
it down from this form, where unary is defined as (1<<N):
FunctionUnit# Addr Length
into:
unary(FU#){0} Addr[4:] unary(Addr[0:3] ) (start=0, mask=000bbbb)
unary(FU#){1} Addr[4:]+1 unary(Addr[0:3]&mask) (start=M, mask=bbb0000)
when a write is misaligned across a cache line, it results in the *second*
operation (unary(FU#){1}) being activated, where there would be two separate
and distinct LOADs / STOREs coming from *one* FunctionUnit.
the (start,mask) is so as to be able to shift and mask the data as it comes in
(or out) of the register. this because when it's split into two, obviously you
have to know which portion of each split covers which part of the data
register.
let us assume that there are 4 LOAD/STORE Function Units.
there is a table (a CAM) with **EIGHT** rows, with an *unary* row-matching
by:
unary(FU#0){0}
unary(FU#0){1}
unary(FU#1){0}
unary(FU#1){1}
unary(FU#2){0}
unary(FU#2){1}
unary(FU#3){0}
unary(FU#3){1}
and in each column there is:
* MSBs of address - address[4:]
* *bitmask* of bytes covered by address[0:3]
* (up to) 128 bit data to be LD/ST'd
basically it's a very very small L0 cache, and because it's so small we can do
row-addressing in *unary*.
the key to resolving misalignment is that initial break-down of the "original"
LD/ST into a *PAIR* (plus a bitmask)
the bit that is a bit... awful is: there will be *four* Load-Store Units
creating and contending for up to *EIGHT* simultaneous reads/writes to this
CAM, worst-case (every single one of the operations crosses a cache-line
boundary)
that's far too much.
i think it is more reasonable to limit that to only four, however even a
four-way port-contention system is still one hell of a lot.
writing to this store/buffer/CAM: with four ports (four ways in CAM
terminology) any one LOAD/STORE will be able to write to two separate addresses
in the CAM
AND
up to *four* LOAD/STORE Units will be able to read/write to the *same* address
(same row).
the Memory itself will need individual byte-level enable lines.
then, the "flush" from this small CAM will be able to read/write entire lines
out to the L1 Cache. each line will kiinda be like one single-entry FIFO.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 217] create a "ring" system which allows pad locations to be specified conveniently
From: bugzilla-daemon @ 2020-03-12 17:37 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-217-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=217
--- Comment #5 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Jean-Paul.Chaput from comment #4)
> I did miss some posts.
oops.
> self.corona is an accessor property (defined in Configuration.py,
> line 663). It's the first element of the self.coronas list,
> which initialized in the checkPads() function at line 1008.
> Rule is simple: any instance which is not a pad is a corona.
> And there should be only one.
okaaay. right. yes. sorry, i was searching "self.corona". of course,
properties would not show up.
> I did lost a post about HDL langiage. No time to rewrite it right
> now but this is a question we asked ourselves too. Until now,
> we are not happy with anything that exists.
interesting. yes, on the one hand, the design and layout can be closely
tied. on the other, they are separate.
if you separate them too much, you lose track on the layout side.
here, really, we have the same analogy - and exactly the same problem - between
PCB Schematic and PCB Layout.
often these two get out-of-sync and it is a bitch.
and you just... deal with it.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] next tasks
From: Jacob Lifshay @ 2020-03-12 17:40 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <CAPweEDwHTi+VHrQ5n+KX98hCbVhDGey6U-LgPYe4H1w2WULG9w@mail.gmail.com>
On Thu, Mar 12, 2020, 07:44 Luke Kenneth Casson Leighton <lkcl@lkcl.net>
wrote:
> On Thu, Mar 12, 2020 at 2:32 PM Hendrik Boom <hendrik@topoi.pooq.com>
> wrote:
> >
> > On Thu, Feb 20, 2020 at 09:39:21AM +0200, Lauri Kasanen wrote:
> > > On Wed, 19 Feb 2020 14:14:16 +0000
> > > Luke Kenneth Casson Leighton <lkcl@lkcl.net> wrote:
> > >
> > > > we only have a few months in which to hang together all of the pieces
> > > > that have been put together, to make a 64-bit 180nm chip. because
> > > > we're not doing this in verilog i believe it is reasonably doable, if
> > > > we take steps to cut "unnecessary" things out i.e. not do too much.
> > > > chop out L2 cache for example.
> > >
> > > What about the C/C++ simulator/interpreter timeline? It's not a blocker
> > > for the hw tapeout, but it is a blocker for my work, which probably
> > > mostly won't make it for the simple Oct tapeout anyway.
>
> correct. it's on a separate track.
>
> > Or a simulation of the new SoC instruction set in C/C++?
>
> to start from an existing power isa simulator (gem5 for example), that
> simulator *happens* to be written in c or more likely c++, therefore
> simulated vector and VPU instructions *have* to be written in c / c++.
>
> otherwise we need to create - from scratch - an entire new POWER ISA
> cycle-accurate instruction simulator, which would be a genuinely
> stupid thing to spend time on given that there are a number of such
> simulators in existence already. including qemu.
>
Note that qemu is by-design not a cycle-accurate simulator -- it is
specifically designed to go as fast as possible rather than matching timing
with any particular CPU.
If we decide to create our own simulator, it might be a good idea to write
it in rpython or similar since that has an optimizing JIT trace compiler
that can effectively speed up any interpreter/simulator. That's what's used
to implement PyPy -- it's basically just a python interpreter written in
rpython.
If we decide we don't want rpython, then I would suggest Rust -- I've
already started writing a cycle-accurate CPU simulator to test out my idea
for making a spectre-proof cpu (I stopped since it was taking longer than a
few days):
https://salsa.debian.org/Kazan-team/spectre-resistant-speculative-processor
Jacob
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] Hello!
From: Luke Kenneth Casson Leighton @ 2020-03-12 23:23 UTC (permalink / raw)
To: Chris Harding, Libre-RISCV General Development
In-Reply-To: <CAPweEDy5y_crPcvD6FL0UMvR+h8D_htubO-qZ7jdc+Y8rbVWJg@mail.gmail.com>
On Tuesday, March 10, 2020, Luke Kenneth Casson Leighton <lkcl@lkcl.net>
wrote:
>
> you'll like these
> https://git.libre-riscv.org/?p=riscv-tests.git;a=blob;f=
> isa/rv64ui/sv_addi_predicated_subvl.S;h=ea30797a70675e3fabd73ee66b7ee6
> 8d470aca58;hb=refs/heads/sv
so, that, chris, is an entirely new and innovative design of vectorisation,
which is best described as a "hardware for-loop".
the PC *pauses* whilst a loop counter generates *multiple* instructions,
using the one at the current PC as a template, increasing the register
numbers by one each time round the loop.
no more extending and duplicating ISAs with vector and vector-scalar
opcodes.
no more SIMD madness.
just *scalar* instructions... in a loop.
l.
--
---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] NLNet Funding Proposals for the Libre RISC-V SoC: call for participation
From: Luke Kenneth Casson Leighton @ 2020-03-13 0:45 UTC (permalink / raw)
To: Discussão do LibrePlanet; +Cc: Libre-RISCV General Development
In-Reply-To: <CAPweEDxP5TqB5Tn8_D0t8TCUTBF2+zSBrDRgVpe=0hRrcYtTmw@mail.gmail.com>
just a quick followup: we got the funding.
therefore: if there is anyone who can program in c/c++ or python who
would like to be the recipient of donations for working on libre
source code that will actually end up creating real silicon, with
skills in assembler, HDL, compilers, formal mathematical proofs,
kernel driver development or simply an attitude to learn whatever it
takes to get the job done: do get in touch.
time for us to say "we don't need permission to own our own hardware".
l.
On Mon, Sep 23, 2019 at 7:56 AM Luke Kenneth Casson Leighton
<lkcl@lkcl.net> wrote:
>
> [please do remove all but libre-riscv-dev when replying, thx]
>
> https://libre-riscv.org/nlnet_proposals/
>
> A series of funding proposals, each for EUR 50,000, have been submitted to NLNet. These are for charitable donations and may be given to either individuals or to Universities (not to Corporations) for completion of an allotted milestone.
>
> The basic idea is to think through and plan for commercial completion all of the tasks that will make the Libre RISC-V SoC a success. It's a lot! We have the following:
>
> a second Vulkan 3D driver, which will be a port of AMDVLK. similar to swiftshader, for the Libre RISCV SoC, except taking into account the Vectorisation, predication and custom accelerated opcodes.
> a video acceleration initiative: with NEON assembler being up to the job of decoding 720p video on recent ARM64 processors, the idea is to design instructions that will do the job and then follow through getting the code upstream.
> two related proposals which, in combination, will result in an actual 180nm ASIC being taped out at TSMC.
> a formal mathematical proof of the hardware design, proving inviolate guarantees of its correctness. this because although auditing the code is possible, it is both laborious, error prone, and could be compromised. mathematical proofs may be run by anyone and are inviolate.
> an augmentation of gcc to support the processor’s parallel and vectorisation capabilities.
>
>
> Yes, really: 180nm ASICs only cost around EUR 600 per square millimetre, and with around 20 or so mm^2 it is completely within the realm of possibility for an NLNet Grant to fund a test ASIC. With each square millimetre being around 40,000 gates in 180nm, that's around 800,000 gates, which is enormous.
>
> We would then have a proven ASIC, and moving up to 40nm or below, which would require around the USD 2m mark, is a far less risky proposition.
>
> The irony is that whilst these funding proposals are quite easy to write, we also need to find people willing to do the work! In particular, we need at least one EU Citizen per project. They do not have to be permanently resident in the EU, they do however need an EU residence. Yes this includes the UK at the time of writing.
>
> This is an extremely important strategic project that puts you - software libre developers - in the driving seat of modern technology instead of picking up the reverse engineering crumbs that fall from the Corporate table with at least a 2 year delay.
>
> If you would like to help and actually receive donations (directly transferred) from NLNet for doing so, please do contact me directly or on libre-riscv-dev.
>
> L.
>
>
>
> --
> ---
> crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
>
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] NLNet Funding Proposals for the Libre RISC-V SoC: call for participation
From: Immanuel, Yehowshua U @ 2020-03-13 1:06 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <CAPweEDwWtgCXscytqhzWRo_Jdt9qvYF8DgKBy6NQXYjw8b9_JQ@mail.gmail.com>
> "we don't need permission to own our own hardware”.
Oooh
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] NLNet Funding Proposals for the Libre RISC-V SoC: call for participation
From: Samuel Falvo II @ 2020-03-13 1:38 UTC (permalink / raw)
To: Libre-RISCV General Development; +Cc: Discussão do LibrePlanet
In-Reply-To: <CAPweEDwWtgCXscytqhzWRo_Jdt9qvYF8DgKBy6NQXYjw8b9_JQ@mail.gmail.com>
Congrats!!
On Thu, Mar 12, 2020 at 5:46 PM Luke Kenneth Casson Leighton <lkcl@lkcl.net>
wrote:
> takes to get the job done: do get in touch.
>
I won't be able to focus on it, unfortunately.
--
Samuel A. Falvo II
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] Demystifying unsafe code in Rust
From: Jacob Lifshay @ 2020-03-13 4:56 UTC (permalink / raw)
To: Libre-RISCV General Development
Found this video that explains a lot of the benefits, misconceptions, and
pitfalls for Rust's unsafe:
https://youtu.be/QAz-maaH0KM
One of the more significant clarifications is that unsafe code in Rust is
NOT just C.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] next tasks
From: Lauri Kasanen @ 2020-03-13 6:44 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <CAC2bXD4nb6AA0Anc2w1Rf5f6c6CFXYbqOrD4WyszqZHXOcMZfw@mail.gmail.com>
On Thu, 12 Mar 2020 10:40:16 -0700
Jacob Lifshay <programmerjake@gmail.com> wrote:
> If we decide to create our own simulator, it might be a good idea to write
> it in rpython or similar since that has an optimizing JIT trace compiler
> that can effectively speed up any interpreter/simulator. That's what's used
> to implement PyPy -- it's basically just a python interpreter written in
> rpython.
>
> If we decide we don't want rpython, then I would suggest Rust -- I've
> already started writing a cycle-accurate CPU simulator to test out my idea
> for making a spectre-proof cpu (I stopped since it was taking longer than a
> few days):
> https://salsa.debian.org/Kazan-team/spectre-resistant-speculative-processor
Using a standard language would beat slightly improved speed, for the
usual reasons. The hipster languages have trouble running in many
settings, and we need people to be able to modify the simulator (e.g.
me).
If the simulator was written in hipster language of the week, I may
not even be able to get it running, and it would place a significant
obstacle on my modifying it. More than a decade in C/C++ vs learning an
entirely new language with all its quirks.
It would likely present similar barriers to others who need to modify
it.
(why yes, I have bad experiences with python, ruby, rust, go...)
- Lauri
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] next tasks
From: Jacob Lifshay @ 2020-03-13 7:38 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <20200313084417.fd17b9032b84d390e182aac5@gmx.com>
On Thu, Mar 12, 2020, 23:43 Lauri Kasanen <cand@gmx.com> wrote:
> On Thu, 12 Mar 2020 10:40:16 -0700
> Jacob Lifshay <programmerjake@gmail.com> wrote:
>
> > If we decide to create our own simulator, it might be a good idea to
> write
> > it in rpython or similar since that has an optimizing JIT trace compiler
> > that can effectively speed up any interpreter/simulator. That's what's
> used
> > to implement PyPy -- it's basically just a python interpreter written in
> > rpython.
> >
> > If we decide we don't want rpython, then I would suggest Rust -- I've
> > already started writing a cycle-accurate CPU simulator to test out my
> idea
> > for making a spectre-proof cpu (I stopped since it was taking longer
> than a
> > few days):
> >
> https://salsa.debian.org/Kazan-team/spectre-resistant-speculative-processor
>
> Using a standard language would beat slightly improved speed, for the
> usual reasons. The hipster languages have trouble running in many
> settings,
I can agree that rpython sounds like it is probably a pain to set up,
however I've found Rust to be utterly trivial to set up on all Linux
systems I've tried (other than termux -- the distro designed to run on top
of Android -- and even there it was relatively easy):
assuming you've already installed build-essential (needed for the linker),
just install using the one-line command on https://rustup.rs/ then build &
run the project by running:
cargo run -- <args to my program>
or to build and run the program independently from cargo:
cargo build
target/debug/<program-name> <args to my program>
and we need people to be able to modify the simulator (e.g.
> me).
>
Ok. fair point.
I'd recommend learning Rust anyway since Kazan (the graphics driver) is
written in Rust, as well as some of the other code.
Additionally, I find Rust to be just as (or more) expressive as modern C++
and it has additional benefits such as the compiler proving that your code
doesn't have undefined behavior (very handy for when you would have 100kloc
of C++ code that occasionally inexplicably segfaults -- been there), a very
powerful traits system (think C++ template metaprogramming but more
extensible and harder to misuse), and a waay better macro system than C
including stuff like procedural macros (running custom code at compile time
to generate source code).
Also, Rust has a very nice standard build system/package manager (cargo),
it's much better than C and C++'s build system mess.
I found Rust to be relatively easy to understand coming from my background
of >7 years of C++ and C.
If the simulator was written in hipster language of the week, I may
> not even be able to get it running, and it would place a significant
> obstacle on my modifying it. More than a decade in C/C++ vs learning an
> entirely new language with all its quirks.
>
> It would likely present similar barriers to others who need to modify
> it.
>
I had picked rpython and Rust as possible languages since python and rust
are already extensively used in the Libre-SOC codebase.
> (why yes, I have bad experiences with python, ruby, rust, go...)
hope they weren't like my experience with Scala -- never got anything to
build after several hours of trying and the JVM is huge.
Jacob
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] next tasks
From: Lauri Kasanen @ 2020-03-13 8:04 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <CAC2bXD62BPZdM8+Y7HRtUwsMy8feD0eZXGNQazpyomgw0fewdQ@mail.gmail.com>
On Fri, 13 Mar 2020 00:38:31 -0700
Jacob Lifshay <programmerjake@gmail.com> wrote:
I don't want to debate languages, but I have to comment on this:
> assuming you've already installed build-essential (needed for the linker),
> just install using the one-line command on https://rustup.rs/
"Pipe this unknown code from the internet straight to your shell"
That is a huge security hole, and a massively stupid idea. On par with
"install this .exe, it will give you naked pics, I promise".
Please do not advocate for such dangerous practices. Some newbies may
actually do it.
- Lauri
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] next tasks
From: Jacob Lifshay @ 2020-03-13 8:25 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <20200313100440.35771c28cefdb56a7e8b9c94@gmx.com>
On Fri, Mar 13, 2020, 01:04 Lauri Kasanen <cand@gmx.com> wrote:
> On Fri, 13 Mar 2020 00:38:31 -0700
> Jacob Lifshay <programmerjake@gmail.com> wrote:
>
> I don't want to debate languages, but I have to comment on this:
>
> > assuming you've already installed build-essential (needed for the
> linker),
> > just install using the one-line command on https://rustup.rs/
> "Pipe this unknown code from the internet straight to your shell"
> That is a huge security hole, and a massively stupid idea. On par with
> "install this .exe, it will give you naked pics, I promise".
>
I totally understand where your coming from, however rustup.rs is part of
the official rust project (closely related to Mozilla), see here for
alternative installation options:
https://github.com/rust-lang/rustup#other-installation-methods
it's similar to installing firefox on windows, where you download and run
the executable.
They have a reputation waay better than the average porn website.
>
> Please do not advocate for such dangerous practices. Some newbies may
> actually do it.
>
They're working on better security:
https://github.com/rust-lang/rustup#security
Jacob
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 218] New: MP3 optimizations
From: bugzilla-daemon @ 2020-03-13 9:49 UTC (permalink / raw)
To: libre-riscv-dev
http://bugs.libre-riscv.org/show_bug.cgi?id=218
Bug ID: 218
Summary: MP3 optimizations
Product: Libre Shakti M-Class
Version: unspecified
Hardware: PC
OS: Linux
Status: CONFIRMED
Severity: enhancement
Priority: ---
Component: Source Code
Assignee: lkcl@lkcl.net
Reporter: cand@gmx.com
CC: libre-riscv-dev@lists.libre-riscv.org
NLnet milestone: ---
Optimizing MP3 code in ffmpeg with new instructions.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
page: next prev latest
Libre RISC-V Mailing list Archives (accessible via public-inbox)
Archives are clonable: git clone --mirror https://inbox.libre-soc.org/libre-riscv-dev
Example config snippet for mirrors
AGPL code for this site: git clone https://public-inbox.org/public-inbox.git