[Libre-soc-dev] ls180 update

Luke Kenneth Casson Leighton lkcl at lkcl.net
Sun Apr 18 10:16:17 BST 2021


On Sun, Apr 18, 2021 at 2:57 AM Jean-Philippe Turcotte
<jp2k5 at hotmail.com> wrote:
>
> Hello everyone,
>
> I have been lurking on the lists since more than a year, and have finally decided to subscribe here. I am sadly not a programmer. Except for a year of computer science more than a decade ago, my background is in management and education. However, I am a tech enthusiast, and I find your project exciting, inspiring, and sorely needed. I hope you succeed.

thanks Jean-Philippe, and welcome

> By the way, I accept to abide with the Charter for all my interactions with the team. Also, I am a total noob with mailing lists, so I apologize if I screw up.

:)

> This piqued my interest. Could this be run in the cloud?

yes - the specs for the first phase (yosys turns HDL into gate-level
BLIF, coriolis2 turns that back into VHDL, yosys turns that into
Verilog, verilator compiles that to c++) would be that each machine
would require at least 128 GB of RAM, and to complete in a reasonable
time-frame at least 100 machines would be required.

for the second phase (coriolis2 completes place-and-route, coriolis2
extracts P&R as VHDL, yosys turns that into Verilog, verilog compiles
that to c++) would require machines with 5x those resources and
require 10x those numbers of machines.

that's 1,000 machines, with 512 GB of RAM in each machine (1 TB of RAM
may be needed).

> I took a glance at Verilator’s docs, but I cannot tell if we can just throw more threads at it, and expect it to go faster.

paradoxically, there are reports online that if threading is used
verilog simulations can go slower.

> Also, would that affect the simulation?

we have to first get to the point where there is a binary to execute
at all.  the resultant executable for the 2nd stage is expected to be
over a gigabyte in size.

> Am I correct in my assumption that this is a two-stage process generating C++, and that this generated code can then be compiled as usual with any C++ compiler?

yes.

> Again, I am not a coder, but I can run commands. If it is as simple as using many threads to get the C++ in a reasonable amount of time, and then use a distributed compiler, I can give it a shot.

if you happen to have access to a cluster of 1,000 machines with 512
GB of RAM in each machine, great!

i will ask NLnet if they have anything.

l.



More information about the Libre-soc-dev mailing list