Libre RISC-V Mailing list Archives (accessible via public-inbox)
 help / color / Atom feed
* Re: [libre-riscv-dev] MMU + TLB idea
From: Luke Kenneth Casson Leighton @ 2020-05-08  0:40 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <6a63afa2-f425-9490-be9f-a05e34cadc9c@gmail.com>

On Friday, May 8, 2020, Michael Nolan <mtnolan2640@gmail.com> wrote:

> I remember some time ago working on a powerpc core that (I think) had a
> partly software MMU. It got me thinking: we could have a TLB and the
> machinery to look up entries in it and translate the addresses in hardware,
> and implement the page table walking and such in software (via a trap).


ha, funny.  i went through this idea twice.  once in 2012 and again 18
months ago.


>  This would I think cut down the work in implementing the MMU by making it
> so we don't need to implement the page table walking machinery in hardware,
> and make it a bit easier to test.


yes.

from a unit test perspective i like it.   it reduces the dependent
components for each test.



> The downside of course would be that TLB misses would be slower, but I
> think it might be a decent trade-off for our first attempt at the MMU.


yyyeahh.  a TLB (basically a cache for MMU lookups) is probably the
critical component, first in the chain.

without a TLB, every MMU lookup (which is *all* memory lookups) can take 4
or more cycles on *every* memory access, both I and D.

so we will definitely need a TLB.  there exists code which i moved to the
"unused" directory that implements one, so we do not need to duplicate that
work (simply adapt it to POWER).

this code was translated from ariane:
https://git.libre-soc.org/?p=soc.git;a=blob;f=src/unused/TLB/ariane/tlb.py;h=72b67a2dcc56aa76aca561f5da3345a99c36c2ab;hb=HEAD#l40

note how small the TLB is - only an 8 entry CAM! this is pretty normal, it
saves power and also means it can do single cycle requests.

the wikipedia page on TLB says that intel do i think three nested levels of
TLB Cache (!) going from 8 entries (single cycle) up to i think 2048
entries (multi cycle), to avoid missing clock cycles.


the one downside of a software MMU: now a linux kernel driver (and
potentially u-boot) supporting that design concept is on the list of
dependencies.

that in turn, to ensure a reduced dependency on the hardware team, means
implementing the software MMU in an emulator (pearpc, dolphin, qemu, gem5,
other).

this would allow the new MMU linux kernel driver to be developed and tested
in advance, then the *exact* same linux kernel binary can be run on the
newly developed hardware, and we are not trying to test 3 newly developed
"unknowns" at once (a guaranteed nightmare).

so on first glance it seems like it would save time.

unfortunately, despite the usefulness of not having an MMU at the unit test
level, as an end goal for a final product, due to the additional *software
development*, it actually *adds* time!

however if there is *already* support for a POWER "soft MMU" in the linux
kernel, we can try it out.

honestly though on balance, by conforming to the RADIX MMU already
suppported for POWER in the linux kernel, we join a "well tested" code path.

with Paul Mackerras working on that for microwatt, (right noe), if we
simply wait a while we will have working code that we can by-rote translate
to nmigen. and i kinda enjoy those mind numbing tasks :)

yes, i thought that a softmmu would save time, too, Michael :)

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

* [libre-riscv-dev] nMigen Hierachical Issues
From: Yehowshua @ 2020-05-08  2:19 UTC (permalink / raw)
  To: Libre-RISCV General Development

I noticed this a couple days back.

https://github.com/m-labs/nmigen/issues/328 <https://github.com/m-labs/nmigen/issues/328>

Does that affect cookie cutter VLSI layout?

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] daily kan-ban 07may2020 update
From: Luke Kenneth Casson Leighton @ 2020-05-08  2:26 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <CAC2bXD6EWrcrxbva=XeUokF3Csu4t0cF0LWVw8mCwZUJ55Ku7g@mail.gmail.com>

On Friday, May 8, 2020, Jacob Lifshay <programmerjake@gmail.com> wrote:

> I attended the OpenPower virtual coffee meeting, there was an interesting
> suggestion to see if we could move to a later tapeout than october, to
> avoid burnout due to deadline crunch.


i've written to Staf to see what is available.  if you recall we have an
agreement to let him use the Oct2020 slot.  so he is "2nd in queue".

there were other ideas to do a simpler processor for that slot.  however
these would likely go to the *back* of the queue (as a new request for that
slot: "3rd in queue").



>
> I also weighed in on nmigen signed/unsigned sign extension/zero
> extension/truncation operators:
> https://github.com/nmigen/nmigen/issues/381


ah as long as they are explicit rules that in no way change current
behavior (like the signed shift one which caused unit tests to inexplicably
fail for no reason right when we do not need the hassle) we're good.



> still slowly getting through the load/store unit stuff, progress may be a
> little slow since I have a bunch of stuff I need to do at home as well.


ok.

remember, address is divided as:

* bits 0 to 3 plus len turned into 16 bit mask
* bit 4 is for odd-even L1 cache bank selection.
* bits 5 thru 11 go through a 2D AddrMatch, as part of the DMs
* bits 12 and above need checking in 1D, *only* in the L0CacheBuffer.

however if you completely ignore *everything* and i mean everything other
than PortInterface (and the prototype L0CacheBuffer) and just do a
comparison on the full bits 5 thru 48 we can add the information from
AddrMatch later.

the task needed really is genuinely much simpler than it seems. estimated
about... 100 lines of code.

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] nMigen Hierachical Issues
From: Luke Kenneth Casson Leighton @ 2020-05-08  2:36 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <88B9E0E6-23AD-49C3-92B8-3824772ABEED@gatech.edu>

On Friday, May 8, 2020, Yehowshua <yimmanuel3@gatech.edu> wrote:

> I noticed this a couple days back.
>
> https://github.com/m-labs/nmigen/issues/328 <https://github.com/m-labs/
> nmigen/issues/328>


the problem i think is that the WIDTH parameter could potentially radically
alter the submodule.

thus, efforts to "analyse" the submodules and get them to share the same
"include" (if i uderstand him correctly this is what he wants) would need
literally a full AST tree walk to determine if they were identical.

Does that affect cookie cutter VLSI layout?


yes.  pass in different parameters or take in global python state which
produces different AST, we cannot safely expect the instances to be the
same.

if *we* know that the AST is the same, we might take the risk.



the "paranoid" solution is to create external imports.

external imports nmigen is forced to stop at the boundary of the import.

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 07may2020 update
From: Staf Verhaegen @ 2020-05-08  8:41 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <CAPweEDzUw8jNQaZ4eJDQQwBrt1-uMPEyzdE8TN-VrRijJAxVoA@mail.gmail.com>

[-- Attachment #1.1: Type: text/plain, Size: 1762 bytes --]

Luke Kenneth Casson Leighton schreef op vr 08-05-2020 om 03:26 [+0100]:
> On Friday, May 8, 2020, Jacob Lifshay <programmerjake@gmail.com> wrote:
> > I attended the OpenPower virtual coffee meeting, there was an interestingsuggestion to see if we could move to a later tapeout than october, toavoid burnout due to deadline crunch.
> 
> 
> i've written to Staf to see what is available.  if you recall we have anagreement to let him use the Oct2020 slot.  so he is "2nd in queue".
> there were other ideas to do a simpler processor for that slot.  howeverthese would likely go to the *back* of the queue (as a new request for thatslot: "3rd in queue").

As a backup plan I had in mind to tape-out some other CPU like a SwerV RISC-V core from Western Digital. Alternatively some of the budget could be used so I can look into multi-port RAMs and register files; maybe I can have a closer look at a PLL also. These could first be tested on 0.35um which costs a lot less (ca. $5000 vs. ca $20000).

There is question about funding. It takes about 3 months after tape-out to get the chips and then the debugging of the chip can only begin which typically also takes a lot of time. So if we delay the tape-out the completion of the testing of the chip will miss the deadline for the current NLNet funding. I am currently supposed to do the testing of the chip but I won't start that if I don't know if I can get the funding for it.

0.18um TSMC MPW runs are done at least once a month. You can find the schedule online: https://europractice-ic.com/wp-content/uploads/2020/05/General-MPW-EUROPRACTICE-200505-v8.pdf
So last two tape-out dates this year are 18/11 and 2/12 (I think 2/12 was not there in a previous version of the file).
greets,
Staf.



[-- Attachment #2: Type: text/plain, Size: 177 bytes --]

_______________________________________________
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 07may2020 update
From: Yehowshua @ 2020-05-08  8:51 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <ce6b15151c8dc03dec058988a808d04753c5acd9.camel@fibraservi.eu>



> On May 8, 2020, at 4:41 AM, Staf Verhaegen <staf@fibraservi.eu> wrote:
> 
> 0.18um TSMC MPW runs are done at least once a month. You can find the schedule online: https://europractice-ic.com/wp-content/uploads/2020/05/General-MPW-EUROPRACTICE-200505-v8.pdf <https://europractice-ic.com/wp-content/uploads/2020/05/General-MPW-EUROPRACTICE-200505-v8.pdf>
> So last two tape-out dates this year are 18/11 and 2/12 (I think 2/12 was not there in a previous version of the file).
> greets,
> Staf.


Hello Staf,

What about 2021? Are there slots for say March 2021?
And worst case scenario, I can acquire the needed funds for testing the chip from alternative funding sources no problem.

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] daily kan-ban 07may2020 update
From: Luke Kenneth Casson Leighton @ 2020-05-08  9:08 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <ce6b15151c8dc03dec058988a808d04753c5acd9.camel@fibraservi.eu>

On Fri, May 8, 2020 at 9:41 AM Staf Verhaegen <staf@fibraservi.eu> wrote:

> As a backup plan I had in mind to tape-out some other CPU like a SwerV RISC-V core from Western Digital.

which would test the Cell Library (and the Libre development workflow)
very well.  except because it would come from the NLNet budget, it
would remove the opportunity for us to have an NLNet-funded tape-out.


> Alternatively some of the budget could be used so I can look into multi-port RAMs and register files; maybe I can have a closer look at a PLL also.

a Professor at LIP6.fr is happy to design one.


> These could first be tested on 0.35um which costs a lot less (ca. $5000 vs. ca $20000).

if we drop the slot, we should talk that through with NLNet.  they're
critical components so would be good to have.  also that clock-gating
cell.

> There is question about funding. It takes about 3 months after tape-out to get the chips and then the debugging of the chip can only begin which typically also takes a lot of time. So if we delay the tape-out the completion of the testing of the chip will miss the deadline for the current NLNet funding. I am currently supposed to do the testing of the chip but I won't start that if I don't know if I can get the funding for it.

ok so what you're saying is: if we try to do a later date, then we're
outside of the window for not just NLNet funding proposals, but
funding from NLNet *at all*, due to the EU Horizon 2020 conditions
that NLNet-funded projects be 100% completed Q3/4 of 2021.

> 0.18um TSMC MPW runs are done at least once a month. You can find the schedule online: https://europractice-ic.com/wp-content/uploads/2020/05/General-MPW-EUROPRACTICE-200505-v8.pdf

so everyone is aware: these general 180nm MPW runs are not EUR 600 /
sq.mm (the subsidised cost via Staf).  they're 2-3x that level: around
EUR 1100 to 1500


Yehowshua wrote:

> And worst case scenario, I can acquire the needed funds for testing the chip from alternative funding sources no problem.

this is good - it would however not be at the EUR 600 / sq.mm level,
it would be at double that, and also likely need Staf's paid-for
assistance (if you're up for that, Staf?) to act as a firebreak for
the NDAs.

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] minimum viable ASIC
From: Luke Kenneth Casson Leighton @ 2020-05-08  9:50 UTC (permalink / raw)
  To: Libre-RISCV General Development

i'd like to discuss what the minimum viable ASIC would look like,
that, if we're happy to do so, we could hit the October 2020 deadline
with it and not be completely stressed.

my take on this is, which i've added here (do edit, comment etc.
https://libre-soc.org/180nm_Oct2020/)

* a Wishbone interface.  this allows us to drop *directly* into
already-written Litex "SOC" infrastructure (leaving all of us in the
core team free to focus on the essentials)
* the dependency matrices are essential.
* a Branch Function Unit is essential (minimum of 1)
* Load/Store Function Units are essential
* so are multiple register file files (SPRs, Condition Regs, 32x INT Regs)
* the integer pipelines (integer and logic instructions) are essential
(the FP ones not so much)
* a very very basic Branch Prediction system (fixed, but observing
POWER branch "hints")
* a very very basic Common Data Bus infrastructure.
* a TLB and MMU are not strictly essential (not for a proof-of-concept ASIC)
* neither in some ways is a L1 cache
* interfaces we need as a bare minimum include GPIO, EINT, SPI and
QSPI, I2C, UART16550, LPC (from Raptor Engineering) and that actually
might even be it.

secondary priorities (which are approximately in priority order)

* a PLL (this is quite a lot however it turns the ASIC from a 24mhz
design into a 300mhz design)
* a TLB and MMU (in combination with a PLL if it is GNU/Linux OS
capable we have an actual viable *saleable product*, immediately)
* dual L1 Caches with the 2x 128-bit-wide L0CacheBuffer to merge 8x LD/STs
* multiple Common Data Buses to / from the RegFile along with a 4x
"Striped" HI/LO-32-ODD/EVEN access pattern.
* multi-issue
* PartitionedSignal-based integer pipelines
* an FP regfile and associated FP pipelines
* SV compliance
* 128x INT/FP registers
* GPU-style opcodes - Jacob you mentioned Texturisation opcodes as
being more important than e.g. SIN/COS.
* additional interfaces such as RGB/TTL, SDRAM, HyperRAM, RGMII,
SD/MMC, USB-ULPI
* a pinmux

so the *absolute* basic - minimum viable ASIC - is to be honest really
not a lot.

now, i was contacted by Rudi, from http://asics.ws, he has some time
available, he's been doing ASIC design for 20+ years.  he's more than
happy to take care of the interfaces for us.  we define the set, he
will get on with it, design the Wishbone Registers etc. for us.

given what's already been achieved, that *really does* leave us with a
very small amount of work to do.  the integer pipelines are something
that's really quite straightforward to do, and basically "track" the
work that Michael's been doing on the simulator.

one essential aspect of the simulator work is: unit tests have to be
done.  those *exact* same unit tests we can use to test the hardware
version, either against the simulator *or* against qemu, it's all the
same.

if MarketNext find people fast, we will have a whole batch of people
who can help hammer through the tasks, which will give us the space to
focus on "difficult" bits.

thoughts appreciated.

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] minimum viable ASIC
From: Staf Verhaegen @ 2020-05-08 10:29 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <CAPweEDw862-S1=Au1YbsKNy0629pa56eShp+2o_Avd96Auz8mw@mail.gmail.com>

[-- Attachment #1.1: Type: text/plain, Size: 321 bytes --]

Luke Kenneth Casson Leighton schreef op vr 08-05-2020 om 10:50 [+0100]:
> 
> * a PLL (this is quite a lot however it turns the ASIC from a 24mhz
> design into a 300mhz design)

Why only 24MHz without PLL ? You should have problems getting external
clock frequencies up to 100MHz without a problem inside a chip.


[-- Attachment #2: Type: text/plain, Size: 177 bytes --]

_______________________________________________
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] minimum viable ASIC
From: Luke Kenneth Casson Leighton @ 2020-05-08 10:42 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <7fcce2dc2715c268c1029783a83ebcd814c489b9.camel@fibraservi.eu>

On Friday, May 8, 2020, Staf Verhaegen <staf@fibraservi.eu> wrote:

>
> Why only 24MHz without PLL ? You should have problems getting external
> clock frequencies up to 100MHz without a problem inside a chip.
>

really? great!  and that's driven from an external 100mhz clock?  it seems
a little high, i guess i am used to SoCs which all run from a 24mhz XTAL.

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 07may2020 update
From: Luke Kenneth Casson Leighton @ 2020-05-08 10:46 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <CAPweEDxSFSJYU-GzRtp62PZo-=QFCHvnD3isrO8Kee4=o+MZgw@mail.gmail.com>

staf tells me he has to focus on a tapeout deadline for a few days.

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

* [libre-riscv-dev] daily kan-ban update 08mar2020
From: Luke Kenneth Casson Leighton @ 2020-05-08 10:56 UTC (permalink / raw)
  To: Libre-RISCV General Development

yesterday we had a discussion with Tim from Raptor CS about interfaces.  if
we have LPC which is the old ISA Memory Bus protocol over only 6 pins then
because it is an actual memory interface we can even boot directly from
it.  start the PC at an address that triggers memory requests that get
fulfilled by an external LPC chip.

if we have that as one of the interfaces (and can run a linux kernel) we
basically have our first customer (and business partner).

i managed to get ST working on the LDSTCompUnit.  LD i should have today.

also i will talk with MarketNext today about the magazine article.

also wrote the page for discussion of "minimum viable ASIC"
https://libre-soc.org/180nm_Oct2020/

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 08mar2020
From: Luke Kenneth Casson Leighton @ 2020-05-08 10:59 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <CAPweEDw-nM8HiGuMaSZPsvf0HBJCCpAfRQjJJOpUZce7uiBOrQ@mail.gmail.com>

writing on behalf of these two people:

Staf is doing a 180nm tapeout for the next few days.

Rudi from ASICS.ws has offered to do all the interfaces (connected to
wishbone).  we need someone to define them for him.




-- 
---
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 07may2020 update
From: Staf Verhaegen @ 2020-05-08 11:02 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <CAPweEDxSFSJYU-GzRtp62PZo-=QFCHvnD3isrO8Kee4=o+MZgw@mail.gmail.com>

[-- Attachment #1.1: Type: text/plain, Size: 1341 bytes --]

Luke Kenneth Casson Leighton schreef op vr 08-05-2020 om 10:08 [+0100]:
> > 0.18um TSMC MPW runs are done at least once a month. You can find the schedule online: https://europractice-ic.com/wp-content/uploads/2020/05/General-MPW-EUROPRACTICE-200505-v8.pdf
> 
> so everyone is aware: these general 180nm MPW runs are not EUR 600 /sq.mm (the subsidised cost via Staf).  they're 2-3x that level: aroundEUR 1100 to 1500
> 
> Yehowshua wrote:
> > And worst case scenario, I can acquire the needed funds for testing the chip from alternative funding sources no problem.
> 
> this is good - it would however not be at the EUR 600 / sq.mm level,it would be at double that, and also likely need Staf's paid-forassistance (if you're up for that, Staf?) to act as a firebreak forthe NDAs.

I don't subsidize the MPW runs I just charge the imec price to NLNet. I don't include normal industrial markup on the price for NLNet and just charge the (projected) time needed for the tape-out.
The cost is a little over the €600/mm² without packaging of the chips. For TSMC tape-out there is a minimum area charge of 25mm². For packaging you can go for semi-automatic packaging with a high cost per chip; prices are here. For high volume you have low per unit packaging cost (<$0.5/unit) but a start-up cost of over $5000.

greets,
Staf.


[-- Attachment #2: Type: text/plain, Size: 177 bytes --]

_______________________________________________
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] minimum viable ASIC
From: Staf Verhaegen @ 2020-05-08 11:10 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <CAPweEDwfqNh-R4EgJP2siEeBZjbqTOZ1svvdk6BBOOuboJeNnA@mail.gmail.com>

[-- Attachment #1.1: Type: text/plain, Size: 629 bytes --]

Luke Kenneth Casson Leighton schreef op vr 08-05-2020 om 11:42 [+0100]:
> On Friday, May 8, 2020, Staf Verhaegen <staf@fibraservi.eu> wrote:
> > Why only 24MHz without PLL ? You should have problems getting externalclock frequencies up to 100MHz without a problem inside a chip.
> 
> really? great!  and that's driven from an external 100mhz clock?  it seemsa little high, i guess i am used to SoCs which all run from a 24mhz XTAL.

There exist clock generator chips, some of them even programmable over a serial bus interface. In the retro world amiga 75MHz accelerator cards or above are not unheard of.
greets,Staf.


[-- Attachment #2: Type: text/plain, Size: 177 bytes --]

_______________________________________________
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 304] New: Define minimum viable interface set for 180nm ASIC
From: bugzilla-daemon @ 2020-05-08 11:15 UTC (permalink / raw)
  To: libre-riscv-dev

https://bugs.libre-soc.org/show_bug.cgi?id=304

            Bug ID: 304
           Summary: Define minimum viable interface set for 180nm ASIC
           Product: Libre-SOC's first SoC
           Version: unspecified
          Hardware: PC
                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: ---

define and document, at https://libre-soc.org/180nm_Oct2020/interfaces/
a set of interfaces for a minimum viable 180nm ASIC

-- 
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 07may2020 update
From: Luke Kenneth Casson Leighton @ 2020-05-08 11:18 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <358908cb49ec852d0b8f85b9ce88f7590cf663b5.camel@fibraservi.eu>

On Fri, May 8, 2020 at 12:02 PM Staf Verhaegen <staf@fibraservi.eu> wrote:

> I don't subsidize the MPW runs I just charge the imec price to NLNet. I don't include normal industrial markup on the price for NLNet and just charge the (projected) time needed for the tape-out.
> The cost is a little over the €600/mm² without packaging of the chips. For TSMC tape-out there is a minimum area charge of 25mm². For packaging you can go for semi-automatic packaging with a high cost per chip; prices are here. For high volume you have low per unit packaging cost (<$0.5/unit) but a start-up cost of over $5000.

appreciate the clarification, i assume the link was this
https://europractice-ic.com/wp-content/uploads/2020/05/General-MPW-EUROPRACTICE-200505-v8.pdf

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] minimum viable ASIC
From: Yehowshua @ 2020-05-08 12:43 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <7fcce2dc2715c268c1029783a83ebcd814c489b9.camel@fibraservi.eu>

I talked to Raptor and if Series A doesn’t pan out in August, Raptor is willing to foot 30k for PLL design in exchange for some equity.

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] minimum viable ASIC
From: Luke Kenneth Casson Leighton @ 2020-05-08 12:46 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <7E85F682-C471-4B1D-882A-8E7B54021E18@gatech.edu>

On Fri, May 8, 2020 at 1:44 PM Yehowshua <yimmanuel3@gatech.edu> wrote:

> I talked to Raptor and if Series A doesn’t pan out in August, Raptor is willing to foot 30k for PLL design in exchange for some equity.

that's fantastic.  it would give us (and any other team)
Foundry-independence.  normally the PLL is available as a Standard
Cell from a Foundry... but only under NDA.

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] minimum viable ASIC
From: Yehowshua @ 2020-05-08 12:49 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <CAPweEDw862-S1=Au1YbsKNy0629pa56eShp+2o_Avd96Auz8mw@mail.gmail.com>



> On May 8, 2020, at 5:50 AM, Luke Kenneth Casson Leighton <lkcl@lkcl.net> wrote:
> 
> i'd like to discuss what the minimum viable ASIC would look like,
> that, if we're happy to do so, we could hit the October 2020 deadline
> with it and not be completely stressed.
> 
> my take on this is, which i've added here (do edit, comment etc.
> https://libre-soc.org/180nm_Oct2020/)
> 
> * a Wishbone interface.  this allows us to drop *directly* into
> already-written Litex "SOC" infrastructure (leaving all of us in the
> core team free to focus on the essentials)

This is a good call. I talked to Florent from Enjoy Digital last week, and he said he’s willing to help with LiTeX integration for LibreSOC. He also said Enjoy Digital is available for hire depending on how much must be done.

> * the dependency matrices are essential.
> * a Branch Function Unit is essential (minimum of 1)
> * Load/Store Function Units are essential
> * so are multiple register file files (SPRs, Condition Regs, 32x INT Regs)
> * the integer pipelines (integer and logic instructions) are essential
> (the FP ones not so much)
> * a very very basic Branch Prediction system (fixed, but observing
> POWER branch "hints")
> * a very very basic Common Data Bus infrastructure.
> * a TLB and MMU are not strictly essential (not for a proof-of-concept ASIC)
> * neither in some ways is a L1 cache
> * interfaces we need as a bare minimum include GPIO, EINT, SPI and
> QSPI, I2C, UART16550, LPC (from Raptor Engineering) and that actually
> might even be it.
> 
> secondary priorities (which are approximately in priority order)
> 
> * a PLL (this is quite a lot however it turns the ASIC from a 24mhz
> design into a 300mhz design)
> * a TLB and MMU (in combination with a PLL if it is GNU/Linux OS
> capable we have an actual viable *saleable product*, immediately)
> * dual L1 Caches with the 2x 128-bit-wide L0CacheBuffer to merge 8x LD/STs
> * multiple Common Data Buses to / from the RegFile along with a 4x
> "Striped" HI/LO-32-ODD/EVEN access pattern.
> * multi-issue
> * PartitionedSignal-based integer pipelines
> * an FP regfile and associated FP pipelines
> * SV compliance
> * 128x INT/FP registers
> * GPU-style opcodes - Jacob you mentioned Texturisation opcodes as
> being more important than e.g. SIN/COS.

And keep in mind, if we aren’t able to finish the GPU opcodes by October/November/December(according to Staff - these are the available 2020 slots), Raptor has agreed to foot 20k for the silicon, again, in exchange for some equity.

> * additional interfaces such as RGB/TTL, SDRAM, HyperRAM, RGMII,
> SD/MMC, USB-ULPI
> * a pinmux
> 
> so the *absolute* basic - minimum viable ASIC - is to be honest really
> not a lot.
> 
> now, i was contacted by Rudi, from http://asics.ws, he has some time
> available, he's been doing ASIC design for 20+ years.  he's more than
> happy to take care of the interfaces for us.  we define the set, he
> will get on with it, design the Wishbone Registers etc. for us.
> 
> given what's already been achieved, that *really does* leave us with a
> very small amount of work to do.  the integer pipelines are something
> that's really quite straightforward to do, and basically "track" the
> work that Michael's been doing on the simulator.
> 
> one essential aspect of the simulator work is: unit tests have to be
> done.  those *exact* same unit tests we can use to test the hardware
> version, either against the simulator *or* against qemu, it's all the
> same.
> 
> if MarketNext find people fast, we will have a whole batch of people
> who can help hammer through the tasks, which will give us the space to
> focus on "difficult" bits.
> 
> thoughts appreciated.
> 
> l.
> 
> _______________________________________________
> 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] minimum viable ASIC
From: Yehowshua @ 2020-05-08 12:50 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <D725A5C8-7A3C-4403-9BF2-CA73B76CA7F5@gatech.edu>

But really guys, I’m betting the Series A will be just fine, from it, we can hire some talented engineers, and from then on, execution, not finances will be our bottleneck.

What I really need is an “hello world” on FPGA by demo day for August.

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] minimum viable ASIC
From: Yehowshua @ 2020-05-08 12:59 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <CAPweEDw862-S1=Au1YbsKNy0629pa56eShp+2o_Avd96Auz8mw@mail.gmail.com>

Luke, I like the list of priorities. Can we convert it into a 15 week timeline?

We can use the timeline to inform where time/effort should be open. We should also pop the timeline on the wiki.

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] minimum viable ASIC
From: Yehowshua @ 2020-05-08 12:59 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <27527BBC-B545-4D85-A912-C90B7763070A@gatech.edu>



> On May 8, 2020, at 8:59 AM, Yehowshua <yimmanuel3@gatech.edu> wrote:
> 
> should be open.

Spent - not open.
_______________________________________________
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 08mar2020
From: Michael Nolan @ 2020-05-08 13:28 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <CAPweEDw-nM8HiGuMaSZPsvf0HBJCCpAfRQjJJOpUZce7uiBOrQ@mail.gmail.com>

Now that I have branches working in the simulator, I was thinking of 
starting work on the integer ALU. Taking a look at alu_hier.py, I think 
I'll keep the interface but replace the implementation with a pipelined 
one.

--Michael


_______________________________________________
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 08mar2020
From: Yehowshua @ 2020-05-08 13:51 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <CAPweEDw+m3B-7eLOqYBVccTBzED9hbtAwftDdcBR2nZRmuES5w@mail.gmail.com>

OK. Lemme have a go at this.
So I know we need LPC. Discussions with Raptor would have us use LPC to grab the boot loader externally I believe.

Here is some sample code for having the LCP sit as a wishbone slave.

Is Rudi on the mailing list yet?

https://gitlab.raptorengineering.com/raptor-engineering-public/lpc-spi-bridge-fpga/blob/master/lpc_interface.v#L1162 <https://gitlab.raptorengineering.com/raptor-engineering-public/lpc-spi-bridge-fpga/blob/master/lpc_interface.v#L1162>

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

page:   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