* 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
* Re: [libre-riscv-dev] Clarifying the difference between Kazan and Mesa 3D Vulkan
From: Luke Kenneth Casson Leighton @ 2020-03-12 14:10 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <20200312135353.jt2lgatbvjmcd53s@topoi.pooq.com>
On Thu, Mar 12, 2020 at 1:53 PM Hendrik Boom <hendrik@topoi.pooq.com> wrote:
> Makes sense, because our graphics *will* run in (a) CPU(s).
> And is there a Mesa that takes Opengl commands and calls Vulcan?
there are "adapters" out there, yes. OpenGL ES 2 is the most complete
one (that i could find last time i looked). there are also
libre-licensed gallium3d-to-vulkan adapters, etc. etc.
basically it all revolves around vulkan, now.
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
* Re: [libre-riscv-dev] Clarifying the difference between Kazan and Mesa 3D Vulkan
From: Hendrik Boom @ 2020-03-12 13:53 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <20200217000021.GB1920@localhost>
On Mon, Feb 17, 2020 at 05:30:26AM +0530, Veera wrote:
> On Sun, Feb 16, 2020 at 12:12:48PM -0800, Cole Poirier wrote:
> > Hi all, but I think particularly Jacob and Luke. I’m having trouble figuring out the different roles that Kazan and Mesa 3d Vulkan are going to play in the 3Dgpu. I’m sure this will be obvious in hindsight but could you help me understand the purpose of each and the differences between the two? (Or point me to the relevant mailing list thread, I’ve tried looking here: http://lists.libre-riscv.org/pipermail/libre-riscv-dev/2020-January/003302.html, but I’m still having difficulty).
> >
>
> OpenGL as a 3D Graphics API had origin in SGI Graphics. Which was closed source.
> Mesa was the first open source software based OpenGL implementation.
>
> Vulkan is the new Graphics API released by Khronos.
> Currently Vulkan is implemented in Mesa for hardware drivers only.
> And Kazan is first software based Vulkan implementation for Mesa meaning
> a vulkan program runs entirely in CPU using kazan.
>
> If I know correctly, libre-riscv will support Vulkan using kazan.
Makes sense, because our graphics *will* run in (a) CPU(s).
And is there a Mesa that takes Opengl commands and calls Vulcan?
-- hendrik
_______________________________________________
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 13:35 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
Luke Kenneth Casson Leighton <lkcl@lkcl.net> changed:
What |Removed |Added
----------------------------------------------------------------------------
NLnet milestone|--- |NLnet.2019.02
--- Comment #5 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
hm.
misaligned is tricky. am just reading POWER 3.0B, book III, 6.5.8 (page 1074),
it says that quad-word (lq, stq) is allowed to raise an "alignment" exception.
there's several others as well (including when write-through is required or
cacheing is inhibited) however at least if we don't have to deal with 8-byte
misalignment that makes things a *lot* simpler.
--
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] New: create a "ring" system which allows pad locations to be specified conveniently
From: bugzilla-daemon @ 2020-03-12 11:57 UTC (permalink / raw)
To: libre-riscv-dev
http://bugs.libre-riscv.org/show_bug.cgi?id=217
Bug ID: 217
Summary: create a "ring" system which allows pad locations to
be specified conveniently
Product: Libre Shakti M-Class
Version: unspecified
Hardware: PC
OS: Linux
Status: CONFIRMED
Severity: enhancement
Priority: ---
Component: Hardware Layout
Assignee: lkcl@lkcl.net
Reporter: lkcl@lkcl.net
CC: libre-riscv-dev@lists.libre-riscv.org
NLnet milestone: ---
see http://bugs.libre-riscv.org/show_bug.cgi?id=178#c206
we need a way to be able to conveniently and easily specify where, for any
given block, its inputs and output points (known as "Pads") are to be located.
examples:
* auto-specifying that all "inputs" (to be identified by a matching pattern in
the pin name "_i") are to go on NORTH, and that all "outputs" (matching the
pattern "_o") are to go on SOUTH
* signals a[0..15] are to go onto the LEFT side of NORTH whilst signals
b[0..15] are to go on to the RIGHT side of NORTH
* signals a[0..15] and b[0..15] are to be interleaved (zip(a,b)) with a spacing
gap of 5 between a[0],b[0] and a spacing gap of *20* between a[0],b[0] and
a[1],b[1] ...
* just as in ioring.py, a list of signals is to be evenly-spaced along a given
side.
that first one might be best done by providing a "helper" routine which returns
a subset of pin signals based on regex pattern-matching (not that i like
regex's but hey)
i don't know whether it's best to suggest this is done by way of one class, or
by way of several classes, or by way of "specifiers" that could end up in a
dictionary (like ioring) or something else.
whatever it is, the way to specify what to do needs to be quite compact, and
easily readable.
--
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 11:48 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 #206 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Jock Tanner from comment #205)
> (In reply to Luke Kenneth Casson Leighton from comment #204)
> > ermmm ermermerm i like the idea of keeping things in an external file
> >
> > *thinks*... actually, CSV format would be better, because it's 3 lines of
> > code:
> >
> > import csv # standard python library
> > with open(file_path, 'r') as csvfile:
> > return list(csv.DictReader(csvfile))
> >
> > remember, we need to minimise the dependencies (see HDL_workflow, and see
> > discussion we just had!)
> >
> > in CSV format, it could be:
> >
> > SIDE,pinname,offset,comment
> > NORTH,clk,,,
> >
> > the defaults would be, if there is no offset, to make one. the order of
> > the pins in the CSV file would define the order in which they appear on
> > that side.
> >
> > i think that would work really well. later we can consider actually
> > embedding
> > the CSV file contents *into* the docstring (or a function) of the actual
> > nmigen source code.
>
> Looks like I implied too much.
(nono, we're discussing ideas, here. it's just that unlike other projects
which can freely add arbitrary dependencies, we have to be careful and consider
how a particular task may be achieved. the suggestion of doing something that
jinja would *be* suited to was great).
> First of all, I thought of keeping all the ioring logic intact as of a
> requirement,
not a hard one.
> or at least as a generally good approach.
yes.
> That's why I didn't
> consider adding, say, CSV support into 'coriolis2.ioring' or
> 'Configuration.loadConfiguration()'.
no that would be stepping on the toes of the coriolis2 project.
> 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.
now, if ioring.py was capable of allowing us to specify that "signals a[0..15]
are to be on the LEFT part of the NORTH side", then yes i would say it was
perfect.
this is why i suggested being able to specify the precise location.
> 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.
> Of course, I'm not yet proficient enough in the current workflow to see all
> the right implications. So I have to ask of you to specify my task.
ok, let's discuss this on another bugreport.
--
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 11:23 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 #205 from Jock Tanner <tanner.of.kha@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #204)
> ermmm ermermerm i like the idea of keeping things in an external file
>
> *thinks*... actually, CSV format would be better, because it's 3 lines of
> code:
>
> import csv # standard python library
> with open(file_path, 'r') as csvfile:
> return list(csv.DictReader(csvfile))
>
> remember, we need to minimise the dependencies (see HDL_workflow, and see
> discussion we just had!)
>
> in CSV format, it could be:
>
> SIDE,pinname,offset,comment
> NORTH,clk,,,
>
> the defaults would be, if there is no offset, to make one. the order of
> the pins in the CSV file would define the order in which they appear on
> that side.
>
> i think that would work really well. later we can consider actually
> embedding
> the CSV file contents *into* the docstring (or a function) of the actual
> nmigen source code.
Looks like I implied too much.
First of all, I thought of keeping all the ioring logic intact as of a
requirement, or at least as a generally good approach. That's why I didn't
consider adding, say, CSV support into 'coriolis2.ioring' or
'Configuration.loadConfiguration()'.
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.
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'.
Of course, I'm not yet proficient enough in the current workflow to see all the
right implications. So I have to ask of you to specify my task.
--
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 11:13 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
Luke Kenneth Casson Leighton <lkcl@lkcl.net> changed:
What |Removed |Added
----------------------------------------------------------------------------
See Also| |http://bugs.libre-riscv.org
| |/show_bug.cgi?id=94
--- Comment #4 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Luke Kenneth Casson Leighton from comment #3)
> i am very wary of making any kind of changes to the Dependency Matrices. it
> took months to get an understanding of them, enough to implement correctly.
(and it was 10 months ago. sigh)
the load/store system is a *multi*-step process: there is the "normal"
(Register-FU and FU-FU hazard system, which takes care of the "usual"
register-based dependencies that all ALUs have including the Branch Unit),
*and* there is the inter-address LD/ST hazard Dependency Matrices.
the memory-function-unit matrices are augmented (differ from) the "normal"
register-style DMs in that they recognise 10 bits of the address as a way to
see if there was a conflict. the logic being "overzealous".
you remember we discussed that last year, jacob, because 10 bits happens to hit
the 1 MEGABYTE boundary of Vulkan Texture blocks?
what i considered doing was having, instead of a straight 10-bit address-line
comparator, shifting that up by 4 bits and having a *bit-map* comparator for
the 4 LSBs.
in other words a byte LOAD would set 1 bit in the 16 bits, a half-word 2 bits,
and so on. (actually to cover misaligned LOADs it would need to be a minimum
23 bits and/or split into *two* separate LOADs).
let us assume that the bitmap width is exactly the same as the cache line width
(16 bits representing 16 byte wide cache lines).
this bitmap - if filled by multiple LOADs - can have a popcount carried out on
it, and that tells us which "line" in the buffer has the highest number of
bytes to perform LOADs.
if the entire bitmap is all 1s, that is a full cache line of LOADs.
--
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 181] test and install public-inbox
From: bugzilla-daemon @ 2020-03-12 1:58 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-181-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=181
--- Comment #31 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to vklr@vkten.in from comment #30)
> Well. The way I thought was to subscribe to mailing list and have that
> delivered
> to a local user, but filter that via exim4 redirection to public-inbox-mda.
ah i would be much happier with something that did not need Maildir or the cron
job (or took up extra disk space)
the local user is up and running (as you can see)
> No need for offlineimap, public-inbox-mirror needed only for initialization
> only.
> No need for another copy of mails in Maildir directory and no need of cron
> job.
>
> I am in testing phase with above setup.
fantastic, sounds much better, and i think eric will really like 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
* [libre-riscv-dev] [Bug 216] LOAD STORE buffer needed
From: bugzilla-daemon @ 2020-03-12 1:54 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 #3 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Jacob Lifshay from comment #2)
>
> Ok. The algorithm I proposed can still be used for scheduling loads/stores
> inside each batch and deciding which cache line to access each cycle, since
> I think the memory dependency matrix currently handles deciding which ops to
> execute each cycle by relying on a slow and simple FIFO algorithm.
nono, it's a bitlevel matrix, capable of multi issue execution and resolving
all conflicts and dependencies in a single cycle.
the ordering is preserved by cascading bits being raised, just like in the
Register-style DMs.
so yes it has the same *characteristics* ad a FIFO bit is not a sequential
binary numbered array, it is *unary* numbering, and the bits (SR Latches)
specify "row X is in front of column Y". actually, "row X is a memory LD
hazard on column Y"
> The queue
> in the algorithm I proposed can be the same HW as the memory dependency
> matrix, so it would end up as a hybrid algorithm.
i am very wary of making any kind of changes to the Dependency Matrices. it
took months to get an understanding of them, enough to implement correctly.
using them as-is, as a component: no problem. modifying them? no - not at this
stage, we simply haven't got time.
> The recursive summary mechanism inspired by carry lookahead will still be
> useful.
can you do this piece? if we spec it up. i can point you at the code it needs
to connect to (minerva L1 cache and LDSTCompUnit).
--
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 181] test and install public-inbox
From: bugzilla-daemon @ 2020-03-12 1:31 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-181-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=181
--- Comment #30 from vklr@vkten.in <vklr@vkten.in> ---
Well. The way I thought was to subscribe to mailing list and have that
delivered
to a local user, but filter that via exim4 redirection to public-inbox-mda.
No need for offlineimap, public-inbox-mirror needed only for initialization
only.
No need for another copy of mails in Maildir directory and no need of cron job.
I am in testing phase with above setup.
--
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 206] Implement branch prediction
From: bugzilla-daemon @ 2020-03-12 0:19 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-206-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=206
--- Comment #28 from Jacob Lifshay <programmerjake@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #27)
> from comp.arch TODO get link to preserve attribution
I manually archived the usenet thread as it was a few days ago, see comment #21
--
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 206] Implement branch prediction
From: bugzilla-daemon @ 2020-03-12 0:13 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-206-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=206
--- Comment #27 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
from comp.arch TODO get link to preserve attribution
P
Paul A. Clayton
to
2 hours agoDetails
On Wednesday, March 11, 2020 at 4:31:56 PM UTC-4, Stefan Monnier wrote:
> > the difference between a global history table and a local history table:
> > a global table takes some sort of hash (or lru) of the entire PC, whilst the
> > local table will take say the 10 LSBs?
>
> IIRC the different is that a local history table is indexed by (a hash
> of) the PC of the branch instruction, whereas the global history table
> is indexed by (a hash of) the trace of branch decisions taken before
> reaching the branch instruction.
>
> So a 10bit index could be respectively:
> - for an LHT, the 10 LSb of the address of the branch instruction
> - for an GHT, the direction (taken/not taken) of the last 10 branch
> instructions executed.
>
> Then again, maybe I'm just confused.
Traditional gshare global tables are indexed by an XORing of
branch address bits (typically just the least significant
bits of the branch address) and a taken/not-taken global history
string. More recent advances have hashed larger history strings
and used what is called path history (information from addresses
of branches) rather than just direction (taken/not-taken) history
with more complex hashing of the index.
TAGE-based predictors use partial tags to reduce aliasing and
(I suspect) accelerate training and use GEometric history
lengths (which provides fast training and the ability to use
very long history strings).
(One idea I would like to test — I doubt it would be helpful —
is using bimode-like storage for a TAGE with different predictions
using different slots. This would effectively require doubling
look-up width, but with overlaided skewed associativity the
problem of capacity bias would be reduced. I did try XORing a
tag bit with the prediction or the hysteresis bit for a per
address prediction and it turned out that constructive/non-
destructive aliasing is more common than I supposed [but the
test also did not filter not-taken branches using a BTB]. I
also suspect that dynamic sizing of TAGE tables for various
lengths could also be beneficial.)
Most local history predictors in current use (excluding, e.g.,
that in the Alpha 210264) only use a saturating counter per
address rather than tracking local history strings and
predicting based on those either with a separate table or a
hardwired algorithm.
--
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] processor and soc naming
From: Luke Kenneth Casson Leighton @ 2020-03-11 23:42 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <CAC2bXD5Q6cJ0ukCUfunuO4f34jW8mFUQGThqctCa5YoprYXQYQ@mail.gmail.com>
On Wednesday, March 11, 2020, Jacob Lifshay <programmerjake@gmail.com>
wrote:
>
>
> I was thinking we'd just set them all to redirect to (perhaps specific
> pages in) libre-soc.org.
yyeah that's a really good idea.
>
>
--
---
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] processor and soc naming
From: Jacob Lifshay @ 2020-03-11 23:12 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <CAPweEDxEdfZwK2rDpsJ3g6H+-taqGrrQNRVbxCuEJe-G1s_ynw@mail.gmail.com>
On Wed, Mar 11, 2020, 15:59 Luke Kenneth Casson Leighton <lkcl@lkcl.net>
wrote:
> On Wednesday, March 11, 2020, Jacob Lifshay <programmerjake@gmail.com>
> wrote:
>
> > On Wed, Mar 11, 2020, 14:15 Jacob Lifshay <programmerjake@gmail.com>
> > wrote:
> >
> > I also registered libre-gpu.org
> >
> >
> aw gawd more domains! :)
>
I was thinking we'd just set them all to redirect to (perhaps specific
pages in) libre-soc.org.
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 216] LOAD STORE buffer needed
From: bugzilla-daemon @ 2020-03-11 23:08 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
Jacob Lifshay <programmerjake@gmail.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |programmerjake@gmail.com
--- Comment #2 from Jacob Lifshay <programmerjake@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #1)
> (In reply to Jacob Lifshay from
> http://bugs.libre-riscv.org/show_bug.cgi?id=215#c5)
> > For a scheduling algorithm for loads that are ready to run (6600-style
> > scheduler sent to load/store unit for execution, no conflicting stores
> > in-front, no memory fences in-front), we can have a queue of memory ops and
> > each cycle we pick the load at the head of the queue and then search from
> > the head to tail for additional loads that target the same cache line
> > stopping at the first memory fence, conflicting store, etc. Once those loads
> > are selected, they are removed from the queue (probably by marking them as
> > removed) and sent thru the execution pipeline.
> >
> > We can use a similar algorithm for stores.
>
> right. ok. thanks to Mitch Alsup, i have a Memory Dependency matrix that
> takes care of discerning and preserving the loads and stores into batches.
> we could if we wanted to do not only TSO, it can handle cross-SMP in a way
> that makes atomic memory either entirely moot or dead trivial. this i need
> a little more research on.
>
> anyway the point is: LOADs as a batch are already identified and hold up any
> STOREs, and vice-versa.
Ok. The algorithm I proposed can still be used for scheduling loads/stores
inside each batch and deciding which cache line to access each cycle, since I
think the memory dependency matrix currently handles deciding which ops to
execute each cycle by relying on a slow and simple FIFO algorithm. The queue in
the algorithm I proposed can be the same HW as the memory dependency matrix, so
it would end up as a hybrid algorithm.
The recursive summary mechanism inspired by carry lookahead will still be
useful.
--
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] processor and soc naming
From: Luke Kenneth Casson Leighton @ 2020-03-11 22:58 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <CAC2bXD6anPZpC=XmYWCuBeYFhJ_c3APJ8T7tLUTemmG9RKmCNA@mail.gmail.com>
On Wednesday, March 11, 2020, Jacob Lifshay <programmerjake@gmail.com>
wrote:
> On Wed, Mar 11, 2020, 14:15 Jacob Lifshay <programmerjake@gmail.com>
> wrote:
>
> I also registered libre-gpu.org
>
>
aw gawd more domains! :)
we need to decide the mailing lists for libresoc btw.
--
---
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
* [libre-riscv-dev] [Bug 216] LOAD STORE buffer needed
From: bugzilla-daemon @ 2020-03-11 22:54 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 #1 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Jacob Lifshay from comment #5)
> (In reply to Luke Kenneth Casson L
> > whew.
> >
> > so that's 128-bit-wide for _textures_... that's on the *load* side. are
> > there any simultaneous (overlapping) "store" requirements? are the
> > code-loops tight enough to require simultaneous 128-bit LD *and* 128-bit ST?
>
> yes and no -- there is code that will benefit from simultaneous loads and
> stores (memcpy and probably most other code that has both loads and stores
> in a loop), however it isn't strictly necessary.
ok.
> It will be highly beneficial to support multiple simultaneous 8, 16, 32, or
> 64-bit loads to a single cache line all being able to complete
> simultaneously independently of alignment in that cache line.
yes i have been thinking about this for a day since getting the LDSTCompUnit
operational. it currently uses a "fake" memory block (for testing) and the
next step is to look at this.
> Also
> misaligned loads that cross cache lines (and possibly page boundaries),
> though those don't need to complete in a single cache access.
ok
> All the above also applies to stores, though they can be a little slower
> since they are less common.
>
> I realize that that will require a really big realignment network, however
> the performance advantages I think are worth it.
yes. we cannot have high performance computation yet no good getting data in
and out.
> For a scheduling algorithm for loads that are ready to run (6600-style
> scheduler sent to load/store unit for execution, no conflicting stores
> in-front, no memory fences in-front), we can have a queue of memory ops and
> each cycle we pick the load at the head of the queue and then search from
> the head to tail for additional loads that target the same cache line
> stopping at the first memory fence, conflicting store, etc. Once those loads
> are selected, they are removed from the queue (probably by marking them as
> removed) and sent thru the execution pipeline.
>
> We can use a similar algorithm for stores.
right. ok. thanks to Mitch Alsup, i have a Memory Dependency matrix that takes
care of discerning and preserving the loads and stores into batches. we could
if we wanted to do not only TSO, it can handle cross-SMP in a way that makes
atomic memory either entirely moot or dead trivial. this i need a little more
research on.
anyway the point is: LOADs as a batch are already identified and hold up any
STOREs, and vice-versa.
> To find the required loads, we can use a network based on recursively
> summarizing chunks of the queue entries' per-cycle ready state, then
> reversing direction from the summary back to the queue entries to tell the
> entries which, if any, execution port they will be running on this cycle.
> There is then a mux for each execution port in the load pipeline to move the
> required info from the queue to the pipeline. The network design is based on
> the carry lookahead network for a carry lookahead adder, which allows taking
> O(N*log(N)) space and O(log(N)) gate latency.
with the Memory DMs already in place, taking care of separating LOAD from STORE
batches, would it be necessary to do that? (i don't know the answer).
also the ordering is not important (Weak Memory Model) and so a buffer will
*only* have LOADs in or STOREs but not both, and it becomes possible to analyse
the buffer and see if a batch is present.
actually if we had a (really) small CAM that mirrored the way the L1 cache
worked, that might work. 2 maybe 4 lines or something.
then you can detect when the lines are fully occupied...
hmmm needs thought
> Loads/Stores that cross a cache boundary can be split into 2 load/store ops
> when sent to the queue and loads reunited when they both complete. They
> should be relatively rare, so we can probably support reuniting only 1 op
> per cycle.
yes.
> RMW Atomic ops and fences can be put in both load and store queues where
> they are executed once they reach the head of both queues.
these luckily Mitch told me you simply set *both* MemRead *and* MemWrite hazard
and the result is fascinatingly they have to be atomic, locking out all other
LDs, STs, and, because of the way the DMs work, preserve the order as well.
so there will *be* no overlap between atomics, LDs, or STs.
now, whether these all go into the same queue, i don't know.
the main thing is, if we have a queue it has to basically not just be a queue,
it has to be a cache as well.
--
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] processor and soc naming
From: Immanuel, Yehowshua U @ 2020-03-11 22:17 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <CAC2bXD6anPZpC=XmYWCuBeYFhJ_c3APJ8T7tLUTemmG9RKmCNA@mail.gmail.com>
Good call
_______________________________________________
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] processor and soc naming
From: Jacob Lifshay @ 2020-03-11 21:21 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <CAC2bXD7xX+_WEj+DT18aZoQcZDr9wd+y5GhnhODA3Otn9oABBg@mail.gmail.com>
On Wed, Mar 11, 2020, 14:15 Jacob Lifshay <programmerjake@gmail.com> wrote:
> I suggest the Libre-SOC 1 for the SoC and Libre-CPU 1 for the processor
> core. later iterations will be numbered 2, 3, etc. I registered
> libre-cpu.org in case we want to use that.
>
I also registered libre-gpu.org
>
_______________________________________________
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] processor and soc naming
From: Jacob Lifshay @ 2020-03-11 21:15 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <327623BB-9D89-4666-A984-5B48FC2EC262@gatech.edu>
On Wed, Mar 11, 2020, 13:37 Immanuel, Yehowshua U <yimmanuel3@gatech.edu>
wrote:
> I’ve seen the term shakti passed around a lot now.
> I know it refers to a class of open source hardware, but what exactly is
> shakti in the context of LibreSOC?
>
From what I understand, Shakti is the common name used for the RISC-V cpu's
that are being designed by the RISE group as part of one of India's
universities: https://shakti.org.in/
Luke originally was working with them and so Libre-SOC is derived/inspired
by the Shakti M Class SOC. Libre-SOC has since evolved into its own
independent thing.
We should probably pick names for our processor core and SoC to
differentiate them from the Shakti M Class and from Libre-SOC as our
organization's name:
I suggest the Libre-SOC 1 for the SoC and Libre-CPU 1 for the processor
core. later iterations will be numbered 2, 3, etc. I registered
libre-cpu.org in case we want to use that.
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] What is Shakti
From: Immanuel, Yehowshua U @ 2020-03-11 20:35 UTC (permalink / raw)
To: Libre-RISCV General Development
I’ve seen the term shakti passed around a lot now.
I know it refers to a class of open source hardware, but what exactly is shakti in the context of LibreSOC?
Yehowshua
_______________________________________________
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] New: LOAD STORE buffer needed
From: bugzilla-daemon @ 2020-03-11 20:28 UTC (permalink / raw)
To: libre-riscv-dev
http://bugs.libre-riscv.org/show_bug.cgi?id=216
Bug ID: 216
Summary: LOAD STORE buffer needed
Product: Libre Shakti M-Class
Version: unspecified
Hardware: Other
OS: Linux
Status: CONFIRMED
Severity: enhancement
Priority: ---
Component: Source Code
Assignee: lkcl@lkcl.net
Reporter: lkcl@lkcl.net
CC: libre-riscv-dev@lists.libre-riscv.org
NLnet milestone: ---
http://bugs.libre-riscv.org/show_bug.cgi?id=215#c5
--
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 215] evaluate minerva for base in libre-soc
From: bugzilla-daemon @ 2020-03-11 19:25 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-215-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=215
--- Comment #5 from Jacob Lifshay <programmerjake@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #4)
> (In reply to Jacob Lifshay from comment #3)
> > do note that compressed texture decoding needs to be able to load 128-bit
> > wide values (a single compressed texture block),
>
> okaaay.
>
> > so our scheduling circuitry
> > should be designed to support that. They should always be aligned, so we
> > won't need to worry about that in the realignment network.
>
> whew.
>
> so that's 128-bit-wide for _textures_... that's on the *load* side. are
> there any simultaneous (overlapping) "store" requirements? are the
> code-loops tight enough to require simultaneous 128-bit LD *and* 128-bit ST?
yes and no -- there is code that will benefit from simultaneous loads and
stores (memcpy and probably most other code that has both loads and stores in a
loop), however it isn't strictly necessary.
It will be highly beneficial to support multiple simultaneous 8, 16, 32, or
64-bit loads to a single cache line all being able to complete simultaneously
independently of alignment in that cache line. Also misaligned loads that cross
cache lines (and possibly page boundaries), though those don't need to complete
in a single cache access.
All the above also applies to stores, though they can be a little slower since
they are less common.
I realize that that will require a really big realignment network, however the
performance advantages I think are worth it.
For a scheduling algorithm for loads that are ready to run (6600-style
scheduler sent to load/store unit for execution, no conflicting stores
in-front, no memory fences in-front), we can have a queue of memory ops and
each cycle we pick the load at the head of the queue and then search from the
head to tail for additional loads that target the same cache line stopping at
the first memory fence, conflicting store, etc. Once those loads are selected,
they are removed from the queue (probably by marking them as removed) and sent
thru the execution pipeline.
We can use a similar algorithm for stores.
To find the required loads, we can use a network based on recursively
summarizing chunks of the queue entries' per-cycle ready state, then reversing
direction from the summary back to the queue entries to tell the entries which,
if any, execution port they will be running on this cycle. There is then a mux
for each execution port in the load pipeline to move the required info from the
queue to the pipeline. The network design is based on the carry lookahead
network for a carry lookahead adder, which allows taking O(N*log(N)) space and
O(log(N)) gate latency.
Loads/Stores that cross a cache boundary can be split into 2 load/store ops
when sent to the queue and loads reunited when they both complete. They should
be relatively rare, so we can probably support reuniting only 1 op per cycle.
RMW Atomic ops and fences can be put in both load and store queues where they
are executed once they reach the head of both queues.
--
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 215] evaluate minerva for base in libre-soc
From: bugzilla-daemon @ 2020-03-11 15:37 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-215-13@http.bugs.libre-riscv.org/>
http://bugs.libre-riscv.org/show_bug.cgi?id=215
--- Comment #4 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Jacob Lifshay from comment #3)
> do note that compressed texture decoding needs to be able to load 128-bit
> wide values (a single compressed texture block),
okaaay.
> so our scheduling circuitry
> should be designed to support that. They should always be aligned, so we
> won't need to worry about that in the realignment network.
whew.
so that's 128-bit-wide for _textures_... that's on the *load* side. are there
any simultaneous (overlapping) "store" requirements? are the code-loops tight
enough to require simultaneous 128-bit LD *and* 128-bit ST?
--
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