* [libre-riscv-dev] [Bug 325] create POWER9 TRAP pipeline
From: bugzilla-daemon @ 2020-06-03 1:28 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-325-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=325
--- Comment #27 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Cole Poirier from comment #26)
> Thanks for your assisstance, it made things much easier, and more robotic :)
> I just pushed a new commit with your suggested changes. If this is now
> complete is the next step writing unit tests?
yes... but it's not complete. every single line needs to be translated,
which means paying attention to every line. it's quite... tedious.
you missed several: i highlighted them for you.
--
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] [OpenPOWER-HDL-Cores] microwatt decoder tables: M-Form and X-Form switched RS and RB
From: Luke Kenneth Casson Leighton @ 2020-06-03 1:28 UTC (permalink / raw)
To: Libre-RISCV General Development; +Cc: openpower-hdl-cores
In-Reply-To: <20200603004052.GA725968@thinks.paulus.ozlabs.org>
On Wed, Jun 3, 2020 at 1:41 AM Paul Mackerras <paulus@ozlabs.org> wrote:
> On Tue, Jun 02, 2020 at 01:10:37PM +0100, Luke Kenneth Casson Leighton wrote:
> > this therefore leaves me wondering if something similar was done for
> > POWER, where we've simply missed these patterns because they're just
> > not documented.
>
> As far as I can see, there are patterns like this to some extent, but
> not uniformly, and not as much as one would like.
>
> For example, with the condition register logical ops (crand etc.),
> there are 4 bits of the instruction word that give the truth table for
> the operation.
yes i saw that trick: utterly cool :) actually using the instruction
field as a 4-bit lookup.
> However, the fixed-point logical ops don't have that.
possibly because all permutations (particularly if you include immediates)
would simply have taken up too much of the major opcode space.
> > however... is this fact encoded *in the instructions*? i.e. is the
> > instruction set encoded such that:
>
> Appendix D (Opcode Maps) of the ISA is useful for investigating this
> sort of question.
ah good point.
> The add and sub instructions are on page 1365, with
> sub instructions in the first column (bits 1-5 of the instruction word
> being 01000) and add instructions in the third column (01010). The
> ops that use CA for the carry input seem to be in rows 5-8 and 21-24
> (instruction bits 6-10 containing x01xx). The ops that have the carry
> input set to one all have bit 6 of the instruction word = 1.
interesting.
> > if this principle has been applied across the board, in the ISA
> > design, the POWER instruction decoder phase could be drastically
> > simplified. right now, the decoder is a massive 4 deep nested series
> > of Switch/Case statements, and consequently the number of gates
> > involved is enormous compared to other RISC architectures.
>
> I don't know of such a principle being applied in recent times, though
> there may have been such a principle in the very early days (early
> 90s).
RISC-V was based on the principle of "learning from past mistakes",
and a thorough analysis shows that one hell of a lot of thought has
gone into the ISA design. i can say that it does indeed make for
a much simpler decoder. the 16-bit C encoding for example, on first
sight, looks like a mess (bits all over the place), however when
implemented in hardware, several bits fall straight through without
needing to be MUXed, regardless of which Function Unit actually
needs those registers / sub-operations / etc.
unfortunately they did not leave enough room for future extension.
LD-with-update is not possible to add, without some inconvenience,
for example.
> I had hoped that the logic simplification algorithms that the
> synthesis tools have would automatically detect the patterns, but
> maybe that's asking a bit much.
if you're paying USD $90m (or renting the tools @ USD 250,000 a week)
it's perfectly reasonable! the rest of us - using yosys? mmmm... :)
i've found that yosys can do a reasonable job of reducing gate delay.
(reducing longest paths). however that *increases* gate count, not
reduces it.
i did wonder if there is a special mode for yosys "synth" command
to ask it to optimise for reduced gate count, instead.
> It would probably help if our decode
> tables indicated the don't-care entries rather than just putting 0 for
> table entries that have no effect.
when converting to CSV format, i went over all of them: i didn't find
many opportunities.
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 325] create POWER9 TRAP pipeline
From: bugzilla-daemon @ 2020-06-03 1:35 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-325-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=325
--- Comment #28 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Cole Poirier from comment #22)
> Thanks. I'm not sure how to tranlate c_in
err... errr... that means *sigh* having to look at the CSV files and see if
reg c can be moved to reg a, for MTMSR.
b _definitely_ needs to be in position "b" - so that means that in OP_RFID
this needs to be b not a:
comb += self.o.msr.data[stt:end].eq( --->a<---- [stt:end])
> and in OP_MTMSR this seems to be
> done by just:
>
> ```
> comb += self.o.msr.data.eq(a)
> comb += self.o.msr.ok.eq(1)
> ```
>
> Is this correct?
it is... but we need to check minor_31.csv - ah:
0b0010110010,ALU,OP_MTMSRD,NONE,NONE,RS,NONE
that's wrong. it says "RS goes into reg c" and if you look at
fu/trap/pipe_data.py
we don't *have* an rs.
so i'll go ahead and change that to:
0b0010110010,ALU,OP_MTMSRD,RS,NONE,NONE,NONE
and that will put RS into *a* (which we have)
gimme 1 sec...
--
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 325] create POWER9 TRAP pipeline
From: bugzilla-daemon @ 2020-06-03 1:41 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-325-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=325
--- Comment #29 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Luke Kenneth Casson Leighton from comment #28)
> 0b0010110010,ALU,OP_MTMSRD,RS,NONE,NONE,NONE
>
> and that will put RS into *a* (which we have)
>
> gimme 1 sec...
commit 7e32307f763f693a405a5d1691fb23894bcd3be3 (HEAD -> master, origin/master)
Author: Luke Kenneth Casson Leighton <lkcl@lkcl.net>
Date: Wed Jun 3 02:39:29 2020 +0100
update submodule for ISA tables
done, do make sure to do a "git submodule update" and make sure
"git submodule" reports this:
1970241a6db97d4ace1e053e72f2a3d9462e98b2 libreriscv
--
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 325] create POWER9 TRAP pipeline
From: bugzilla-daemon @ 2020-06-03 1:50 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-325-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=325
--- Comment #30 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
oo... errr... do another git pull - i added this:
# L = self.fields.FormXL[0:-1]
# if e_in.insn(16) = '1' then <-- this is X-form field "L".
so rather than "with m.If(insn[16]):"
you want: "with m.If(L):"
ok?
--
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 325] create POWER9 TRAP pipeline
From: bugzilla-daemon @ 2020-06-03 2:31 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-325-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=325
--- Comment #31 from Cole Poirier <colepoirier@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #27)
> (In reply to Cole Poirier from comment #26)
> > Thanks for your assisstance, it made things much easier, and more robotic :)
> > I just pushed a new commit with your suggested changes. If this is now
> > complete is the next step writing unit tests?
>
> yes... but it's not complete. every single line needs to be translated,
> which means paying attention to every line. it's quite... tedious.
> you missed several: i highlighted them for you.
Thanks for the help!!
Ok, I *think* I have made the necessary changes. Just pushed a commit. I'll
take a look at your comments and make any new recommended changes in the
morning.
--
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 325] create POWER9 TRAP pipeline
From: bugzilla-daemon @ 2020-06-03 3:27 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-325-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=325
--- Comment #32 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
- # TODO translate this, import and use br_ext from branch stage
- # f_out.redirect_nia <= a_in(63 downto 2) & "00"; -- srr0
+ comb += self.o.nia.data.eq(br_ext(a[63:1] & 0))
+ comb += self.o.nia.ok.eq(1)
br_ext already adds the 2 zeros. & in vhdl means "concatenate". however they
go MSB -> LSB order, python is LSB->MSB order
comb += self.o.nia.data.eq(br_ext(a[2:]))
you removed all of these, please do put them back:
- a_i, b_i, cia_i, msr_i = self.i.a, self.i.b, self.i.cia, self.i.msr
- o, msr_o, nia_o = self.o.o, self.o.msr, self.o.nia
- srr0_o, srr1_o = self.o.srr0, self.o.srr1
+ a_i, b_i = self.i.a, self.i.b
- # -- just update EE and RI
- # ctrl_tmp.msr(MSR_EE) <= c_in(MSR_EE);
- # ctrl_tmp.msr(MSR_RI) <= c_in(MSR_RI);
+ L = self.fields.FormX.L[0:-1]
+ with m.If(L):
+ comb += self.o.msr[MSR_EE].eq(self.i.msr[MSR_EE])
+ comb += self.o.msr[MSR_RI].eq(self.i.msr[MSR_RI])
+
here, this is copying the incoming MSR into the outgoing MSR. that's
just wasting time. look again: it says c_in, and we discussed that
this is to be moved in comment #28 to "a".
also, those defines MSR_xx do not exist. note the comment at the top:
# TODO: turn these into python constants (just "MSR_SF = 63-0 # comment" etc.)
"""
Listed in V3.0B Book III Chap 4.2.1
-- MSR bit numbers
constant MSR_SF : integer := (63 - 0); -- Sixty-Four bit mode
....
"""
i'm leaving it to you to create those constants.
--
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] googleusercontent looks like it got hacked yesterday
From: Luke Kenneth Casson Leighton @ 2020-06-03 10:08 UTC (permalink / raw)
To: Libre-RISCV General Development
i'm investigating why i can't get log in to google: fail2ban has this
IP address.
DROP all -- 176.193.193.35.bc.googleusercontent.com anywhere
if you invert that IP address it lists a russian domain name.
it's therefore highly likely that someone was messing with world-wide
BGP routing tables yesterday.
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 361] New: all test_pipe_caller.py needs RA=0 tests
From: bugzilla-daemon @ 2020-06-03 12:00 UTC (permalink / raw)
To: libre-riscv-dev
https://bugs.libre-soc.org/show_bug.cgi?id=361
Bug ID: 361
Summary: all test_pipe_caller.py needs RA=0 tests
Product: Libre-SOC's first SoC
Version: unspecified
Hardware: PC
OS: Mac OS
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: ---
the unit tests do not test if RA=0 (which triggers zero-imm).
this needs to be added where appropriate
--
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] daily kan-ban update 03jun2020
From: Luke Kenneth Casson Leighton @ 2020-06-03 12:50 UTC (permalink / raw)
To: Libre-RISCV General Development
yesterday:
* fixed MultiCompUnit FSM for case when there are no registers to
write (branch conditional)
* worked with Michael on Branch, deciding where to put the incoming
SPR registers (port1, port2)
* sorted out regspecs, in particular the "fast" regfile (CTR, MSR, CIA, TAR, LR)
* created a special strategic function that encodes the relationship
between regspecs, PowerDecode2, and the Function Units
* reorganised the */test_pipe_caller.py and test_*_compunit.py unit
tests to share common functions for setting up input
* worked with Cole on TRAP pipeline.
* moved RS to pipeline input position 1, in MTSPR and MTMSR.
morning:
* more pipeline / compunit test reorganisation
* dynamic enable/disable carry-in and XER flags (and test it's ok)
* remove pipe_data.py rdflags() and use strategic function
regspec_decode() instead
today:
* continue helping Cole with TRAP pipeline
* slowly work towards linking pipelines to regfiles, now that
regspec_decode can express the port read/write characteristics, purely
from pipeline specifications (pipe_data.py)
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] daily kan-ban update 03jun2020
From: Tobias Platen @ 2020-06-03 13:28 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <CAPweEDwOG8gQWh_CbrHkmWj1LVaoUc8mCdVDwvkDAze++JnvUQ@mail.gmail.com>
today:
formal verification of the data merger class.
_______________________________________________
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] regfile-to-function-unit connection taking shape
From: Luke Kenneth Casson Leighton @ 2020-06-03 19:21 UTC (permalink / raw)
To: Libre-RISCV General Development
i put together a "non-production core" which is starting to take shape:
the first priority being to get the register file ports connected.
this involves first enumerating them, and just thinking about the data
structures involved was hampering me from making progress.
i therefore decided a while ago to drop that (after first creating the
MultiCompUnit Record system, to make it easier to come back to), and
now, after testing each MCU and being happy that they can handle
and look after their respective pipelines, it's time to connect them
to actual regfiles.
this is being done here:
https://git.libre-soc.org/?p=soc.git;a=blob;f=src/soc/simple/core.py;hb=HEAD
the output from that (which is for read ports only at the moment)
shows exactly what needs to be connected up. the "lanes" designation
is purely arbitrary convenience, however the important thing is that
they are all the same "register". lane 0 for INT is *always* RA.
lane 1 for the INT regfile is *always* RB, and so on.
the next stage will be to introduce a series of PriorityPickers,
which allow one and *only* one of the candidates (RA INT Lane 0)
access to that dedicate Regfile port.
this straight one-to-one relationship means that there is no need
for that cyclic buffer previously described.
also the plan is to very simply only allow one instruction to be
"active" at any one time (issue, wait for read, wait for busy,
wait for write, then back to issue). this avoids the need for
any kind of Dependency Tracking, thus allowing us to actually have
"something working".
l.
-----
regfile read ports: INT
regfile read port INT lane: 0
alu0 ra 0:63 (<< (const 1'd1) (sig reg1)) (sig reg1_ok)
<soc.fu.compunits.compunits.ALUFunctionUnit object at 0x7f6abac0ea58>
cr0 ra 0:63 (<< (const 1'd1) (sig reg1)) (sig reg1_ok)
<soc.fu.compunits.compunits.CRFunctionUnit object at 0x7f6abaa9a2b0>
logical0 ra 0:63 (<< (const 1'd1) (sig reg1)) (sig reg1_ok)
<soc.fu.compunits.compunits.LogicalFunctionUnit object at
0x7f6aba9984e0>
shiftrot0 ra 0:63 (<< (const 1'd1) (sig reg1)) (sig reg1_ok)
<soc.fu.compunits.compunits.ShiftRotFunctionUnit object at
0x7f6aba8c7ef0>
regfile read port INT lane: 1
alu0 rb 0:63 (<< (const 1'd1) (sig reg2)) (sig reg2_ok)
<soc.fu.compunits.compunits.ALUFunctionUnit object at 0x7f6abac0ea58>
cr0 rb 0:63 (<< (const 1'd1) (sig reg2)) (sig reg2_ok)
<soc.fu.compunits.compunits.CRFunctionUnit object at 0x7f6abaa9a2b0>
logical0 rb 0:63 (<< (const 1'd1) (sig reg2)) (sig reg2_ok)
<soc.fu.compunits.compunits.LogicalFunctionUnit object at
0x7f6aba9984e0>
shiftrot0 rb 0:63 (<< (const 1'd1) (sig reg2)) (sig reg2_ok)
<soc.fu.compunits.compunits.ShiftRotFunctionUnit object at
0x7f6aba8c7ef0>
regfile read port INT lane: 2
shiftrot0 rc 0:63 (<< (const 1'd1) (sig reg3)) (sig reg3_ok)
<soc.fu.compunits.compunits.ShiftRotFunctionUnit object at
0x7f6aba8c7ef0>
regfile read ports: XER
regfile read port XER lane: 2
alu0 xer_so 32 1 (& (slice (sig oe) 0:1) (sig oe_ok))
<soc.fu.compunits.compunits.ALUFunctionUnit object at 0x7f6abac0ea58>
regfile read port XER lane: 3
alu0 xer_ca 34,45 2 (== (sig input_carry) (const 2'd2))
<soc.fu.compunits.compunits.ALUFunctionUnit object at 0x7f6abac0ea58>
shiftrot0 xer_ca 34,45 2 (== (sig input_carry) (const 2'd2))
<soc.fu.compunits.compunits.ShiftRotFunctionUnit object at
0x7f6aba8c7ef0>
regfile read ports: CR
regfile read port CR lane: 2
cr0 full_cr 0:31 255 (sig read_cr_whole)
<soc.fu.compunits.compunits.CRFunctionUnit object at 0x7f6abaa9a2b0>
branch0 cr_a 0:3 (<< (const 1'd1) (sig cr_in1)) (sig cr_in1_ok)
<soc.fu.compunits.compunits.BranchFunctionUnit object at
0x7f6abaa185c0>
regfile read port CR lane: 3
cr0 cr_a 0:3 (<< (const 1'd1) (sig cr_in1)) (sig cr_in1_ok)
<soc.fu.compunits.compunits.CRFunctionUnit object at 0x7f6abaa9a2b0>
regfile read port CR lane: 4
cr0 cr_b 0:3 (<< (const 1'd1) (sig cr_in2)) (sig cr_in2_ok)
<soc.fu.compunits.compunits.CRFunctionUnit object at 0x7f6abaa9a2b0>
regfile read port CR lane: 5
cr0 cr_c 0:3 (<< (const 1'd1) (sig cr_in2)) (sig cr_in2_ok)
<soc.fu.compunits.compunits.CRFunctionUnit object at 0x7f6abaa9a2b0>
regfile read ports: FAST
regfile read port FAST lane: 0
branch0 spr1 0:63 (<< (const 1'd1) (sig fast1)) (sig fast1_ok)
<soc.fu.compunits.compunits.BranchFunctionUnit object at
0x7f6abaa185c0>
regfile read port FAST lane: 1
branch0 spr2 0:63 (<< (const 1'd1) (sig fast2)) (sig fast2_ok)
<soc.fu.compunits.compunits.BranchFunctionUnit object at
0x7f6abaa185c0>
regfile read port FAST lane: 3
branch0 cia 0:63 1 (const 1'd1)
<soc.fu.compunits.compunits.BranchFunctionUnit object at
0x7f6abaa185c0>
---
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 329] coriolis2 experiment layout for Dependency Matrices
From: bugzilla-daemon @ 2020-06-03 22:00 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-329-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=329
--- Comment #9 from Jean-Paul.Chaput@lip6.fr ---
(In reply to Luke Kenneth Casson Leighton from comment #8)
> after a little thought, see the diagram in attachment "fu-fu matrix"
> https://bugs.libre-soc.org/attachment.cgi?id=58
Hello Luke,
I'm starting to have automated matrix placement working for the FU-FU
matrix. But the first comparison I can make between matrix placement
and automated placement still got the later winning (by a little).
I have two little remarks about the FU-FU example you gave me:
1. It seems to me that the current example has 4 unit and 3 regs
(inverted in comment #5).
2. Nothing is connected to gowr3_i, seems strange to me.
I did try to generate a bigger FU-FU, but got stuck here:
Traceback (most recent call last):
File "fu_fu_matrix.py", line 169, in <module>
test_fu_fu_matrix()
File "fu_fu_matrix.py", line 166, in test_fu_fu_matrix
run_simulation(dut, d_matrix_sim(dut), vcd_name='test_fu_fu_matrix.vcd')
File "/usr/lib/python3.6/site-packages/nmigen/compat/sim/__init__.py", line
22, in run_simulation
fragment.domains += ClockDomain("sync")
AttributeError: 'FUFUDepMatrix' object has no attribute 'domains'
I should have the latest nmigen, soc & nmutil repositry versions.
Would it be possible for you to generate a FU-FU matrix as close as possible
to the one intended in terms of number of FU and registers (the idea is to
have a representative amount of wires and gates).
Best regards,
PS1: I use the approach of "not getting to the bottom of things but start like
a copying monkey", but always still felt a little bit like cheating. Now,
I will
do it proudly...
PS2: Maybe you have seen this... (personally, I'm at 110 characters)
https://www.phoronix.com/scan.php?page=news_item&px=Linux-Kernel-Deprecates-80-Col
--
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 329] coriolis2 experiment layout for Dependency Matrices
From: bugzilla-daemon @ 2020-06-03 22:18 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-329-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=329
--- Comment #10 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Jean-Paul.Chaput from comment #9)
> (In reply to Luke Kenneth Casson Leighton from comment #8)
> > after a little thought, see the diagram in attachment "fu-fu matrix"
> > https://bugs.libre-soc.org/attachment.cgi?id=58
>
> Hello Luke,
>
> I'm starting to have automated matrix placement working for the FU-FU
> matrix.
superb. do commit it, we can take a look.
> But the first comparison I can make between matrix placement
> and automated placement still got the later winning (by a little).
that's veery unlikely to be the case at 32x12, 32x25 or, when we go
to GPU DMs, 48x30 or even 128x30. (this is FU-REGs, we are currently
investigating FU-FU, which will still potentially be be up to 30x30)
> I have two little remarks about the FU-FU example you gave me:
>
> 1. It seems to me that the current example has 4 unit and 3 regs
> (inverted in comment #5).
probably :) actually for FU-FU it should be a square matrix.
> 2. Nothing is connected to gowr3_i, seems strange to me.
because if it's not a square matrix, that's what happens.
> I did try to generate a bigger FU-FU, but got stuck here:
>
> Traceback (most recent call last):
> File "fu_fu_matrix.py", line 169, in <module>
> test_fu_fu_matrix()
> File "fu_fu_matrix.py", line 166, in test_fu_fu_matrix
> run_simulation(dut, d_matrix_sim(dut), vcd_name='test_fu_fu_matrix.vcd')
> File "/usr/lib/python3.6/site-packages/nmigen/compat/sim/__init__.py",
> line 22, in run_simulation
> fragment.domains += ClockDomain("sync")
> AttributeError: 'FUFUDepMatrix' object has no attribute 'domains'
>
> I should have the latest nmigen, soc & nmutil repositry versions.
you need the version nmigen from whitequark, not from m-labs.
check the repo url in .git/config, it should be
https://github.com/nmigen/nmigen/
and you can ignore that, anyway - the ilang file-generation should
have happened before that.
> Would it be possible for you to generate a FU-FU matrix as close as possible
> to the one intended in terms of number of FU and registers (the idea is to
> have a representative amount of wires and gates).
let's make it 30x30 - that's pretty damn big. for 180nm we'll likely go
with 1/2 that size.
> Best regards,
>
> PS1: I use the approach of "not getting to the bottom of things but start
> like
> a copying monkey", but always still felt a little bit like cheating.
> Now, I will
> do it proudly...
haha very funny :)
> PS2: Maybe you have seen this... (personally, I'm at 110 characters)
>
> https://www.phoronix.com/scan.php?page=news_item&px=Linux-Kernel-Deprecates-
> 80-Col
sigh there's some severe problems with internet connectivity (much more
sophisticated than a DDOS attack), i still cannot get access to the
majority of news websites. slashdot, theregister, phoronix, zoom, they're
all timing out (!)
if you can see this page https://slashdot.org/~lkcl then you'll see my
comments on this one. summary: it's a serious mistake. this screenshot
shows why: https://libre-soc.org/HDL_workflow/640x-2020-01-24_11-56.png
--
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] Move FHDLTestCase to our utils folder
From: Yehowshua @ 2020-06-03 23:33 UTC (permalink / raw)
To: Libre-RISCV General Development
See here. Whitequark is not maintaining this API.
I can make a bugreport for this.
https://github.com/nmigen/nmigen/pull/399 <https://github.com/nmigen/nmigen/pull/399>
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
* Re: [libre-riscv-dev] Move FHDLTestCase to our utils folder
From: Yehowshua @ 2020-06-03 23:38 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <9B1A70D9-238A-4020-8FF8-74869B061502@gatech.edu>
We would have to move it to the ``utils`` repo.
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
* Re: [libre-riscv-dev] Move FHDLTestCase to our utils folder
From: Luke Kenneth Casson Leighton @ 2020-06-03 23:46 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <75C181BD-5C01-479A-8D98-75E70F80A8FA@gatech.edu>
On Thu, Jun 4, 2020 at 12:38 AM Yehowshua <yimmanuel3@gatech.edu> wrote:
> We would have to move it to the ``utils`` repo.
nmutil, yes. it's accumulating other utilities that are used across
the board and we critically rely on and need.
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] daily kan-ban update 03jun2020
From: Cole Poirier @ 2020-06-03 23:58 UTC (permalink / raw)
To: Libre-RISCV General Development; +Cc: Libre-RISCV General Development
In-Reply-To: <CAPweEDwOG8gQWh_CbrHkmWj1LVaoUc8mCdVDwvkDAze++JnvUQ@mail.gmail.com>
On Jun 3 2020, at 5:50 am, Luke Kenneth Casson Leighton <lkcl@lkcl.net>
wrote:
> today:
>
> * continue helping Cole with TRAP pipeline
Thanks Luke, I had a great time working with you on all the TODOs
yesterday, and I feel just a little bit more confident in my handle on
the concepts at play today than I did yesterday.
Today:
continue working on TRAP pipeline fixing the TODOs luke has provided for me
Cole
_______________________________________________
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] daily kan-ban update 03jun2020
From: Luke Kenneth Casson Leighton @ 2020-06-04 0:03 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <B2C7D56F-40CD-4625-8C38-2E8BF8BC2D96@getmailspring.com>
On Thu, Jun 4, 2020 at 12:58 AM Cole Poirier <colepoirier@gmail.com> wrote:
>
> On Jun 3 2020, at 5:50 am, Luke Kenneth Casson Leighton <lkcl@lkcl.net>
> wrote:
> > today:
> >
> > * continue helping Cole with TRAP pipeline
>
> Thanks Luke, I had a great time working with you on all the TODOs
> yesterday, and I feel just a little bit more confident in my handle on
> the concepts at play today than I did yesterday.
:)
> Today:
>
> continue working on TRAP pipeline fixing the TODOs luke has provided for me
cool. remember also to include the comments - don't delete them. the
VHDL comments ( "--" VHDL translated to "#" python) are just as
important to do line-by-line. you'll need to go over the microwatt
execute1.vhdl and the bugreport because i noticed you deleted some of
the python docstrings containing the VHDL, without translating the
VHDL comment strings.
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] daily kan-ban update 03jun2020
From: Cole Poirier @ 2020-06-04 0:09 UTC (permalink / raw)
To: Libre-RISCV General Development; +Cc: Libre-RISCV General Development
In-Reply-To: <CAPweEDxcPG=+BKLJup+dPH4akmnVAK=NyqoA8H_YfFbMD-rC6Q@mail.gmail.com>
On Jun 3 2020, at 5:03 pm, Luke Kenneth Casson Leighton <lkcl@lkcl.net> wrote:
> cool. remember also to include the comments - don't delete them. the
> VHDL comments ( "--" VHDL translated to "#" python) are just as
> important to do line-by-line. you'll need to go over the microwatt
> execute1.vhdl and the bugreport because i noticed you deleted some of
> the python docstrings containing the VHDL, without translating the
> VHDL comment strings.
>
> l.
Yes sorry about that, part was intentional, which I see now was clearly
a silly thing to do, the other part was that I managed to really mess up
the 'resolution' of merge conflicts. I was following the same guide I've
used previously (successfully) for git mergetool vimdiff, and for some
reason my attempt to accept the remote changes in favour of my own did
the opposite... and then some, like deleting a bunch of lines near the
top of the file that I did't touch. This all would not have been a
problem had I done a second review of the changes before pushing them
like I was supposed to. Lesson learned.
Cole
_______________________________________________
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 325] create POWER9 TRAP pipeline
From: bugzilla-daemon @ 2020-06-04 0:29 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-325-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=325
--- Comment #33 from Cole Poirier <colepoirier@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #32)
> - # TODO translate this, import and use br_ext from branch
> stage
> - # f_out.redirect_nia <= a_in(63 downto 2) & "00"; -- srr0
> + comb += self.o.nia.data.eq(br_ext(a[63:1] & 0))
> + comb += self.o.nia.ok.eq(1)
>
> br_ext already adds the 2 zeros. & in vhdl means "concatenate". however
> they
> go MSB -> LSB order, python is LSB->MSB order
>
> comb += self.o.nia.data.eq(br_ext(a[2:]))
I see. Thank you I will make this fix next...
> you removed all of these, please do put them back:
>
> - a_i, b_i, cia_i, msr_i = self.i.a, self.i.b, self.i.cia, self.i.msr
> - o, msr_o, nia_o = self.o.o, self.o.msr, self.o.nia
> - srr0_o, srr1_o = self.o.srr0, self.o.srr1
> + a_i, b_i = self.i.a, self.i.b
>
>
> - # -- just update EE and RI
> - # ctrl_tmp.msr(MSR_EE) <= c_in(MSR_EE);
> - # ctrl_tmp.msr(MSR_RI) <= c_in(MSR_RI);
> + L = self.fields.FormX.L[0:-1]
> + with m.If(L):
> + comb += self.o.msr[MSR_EE].eq(self.i.msr[MSR_EE])
> + comb += self.o.msr[MSR_RI].eq(self.i.msr[MSR_RI])
> +
>
> here, this is copying the incoming MSR into the outgoing MSR. that's
> just wasting time. look again: it says c_in, and we discussed that
> this is to be moved in comment #28 to "a".
Yes, sorry, I found that part confusing, and misinterpreted your change of the
csv to mean that this change was unnecessary, will fix this next next.
> also, those defines MSR_xx do not exist. note the comment at the top:
> # TODO: turn these into python constants (just "MSR_SF = 63-0 # comment"
> etc.)
> """
> Listed in V3.0B Book III Chap 4.2.1
> -- MSR bit numbers
> constant MSR_SF : integer := (63 - 0); -- Sixty-Four bit mode
> ....
> """
>
> i'm leaving it to you to create those constants.
I did this, and restored the convenience variables, and restored the deleted
VHDL comments. It should be in the right state now, so I can move on to making
the two above, next, and next next changes.
--
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] daily kan-ban update 03jun2020
From: Luke Kenneth Casson Leighton @ 2020-06-04 0:32 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <A4AF384A-E5CA-413E-87FC-338016E6F5E7@getmailspring.com>
On Thu, Jun 4, 2020 at 1:09 AM Cole Poirier <colepoirier@gmail.com> wrote:
> Yes sorry about that, part was intentional, which I see now was clearly
> a silly thing to do, the other part was that I managed to really mess up
> the 'resolution' of merge conflicts.
your responsibility to sort out :)
welcome to working on the same file.
suggestion: do regular "git pulls". this doesn't work with rebase, so
you have to use "git stash" followed by "git pull" followed by "git
stash apply" and then work it out from there.
or you can use "git stash show -p {0}" (after doing "git stash list"
to find the right one) prior to "git stash apply" and that will give
you the diff before applying it.
which.... if you want, you can pipe that diff into a file, then use
"patch", which has the advantage of putting rejected patches into
".rej" files, which you can then manually sort out.
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 325] create POWER9 TRAP pipeline
From: bugzilla-daemon @ 2020-06-04 0:34 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-325-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=325
--- Comment #34 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
> go MSB -> LSB order, python is LSB->MSB order
sorry, that should be "python is LSB->MSB+1" order.
--
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] using a stage chain in a stage
From: Jacob Lifshay @ 2020-06-04 0:52 UTC (permalink / raw)
To: Libre-RISCV General Development
I'm trying to build the div pipe, however, the ispec and ospec need to
be different than the currently existing code since that is FP
specific, there needs to be additional io signals that are not
included in all the FP*Data classes. Additionally, I don't want to
just overwrite the existing FPDiv pipe code since I'm not planning on
adding FP div support right away, and the code needs to be refactored
somewhat since the FPdiv pipe code is currently generic over being
16/32/64-bit at compile-time and the new code will need to be able to
dynamically switch between them at runtime, once FP is added again.
Luke, or whoever's available, can you give me an example of a
combinatorial stage chain that is used inside a single stage without
being wired as the inputs and outputs, along with getting all the
pipekls stuff connected properly?
Somewhat unrelated, I moved all the unused code out of the fpdiv
folder and into the unused folder
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] daily kan-ban update 03jun2020
From: Jacob Lifshay @ 2020-06-04 0:58 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <CAPweEDwOG8gQWh_CbrHkmWj1LVaoUc8mCdVDwvkDAze++JnvUQ@mail.gmail.com>
continuing working on the div and mul pipelines
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
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