Caribou Developers meetings will from now on be held on Tuesdays, bi-weekly. But still need to determine odd or even weeks to always keep the same pattern.
16:10
→
16:35
Hardware25m
Speakers:
Eric Buschmann(Brookhaven National Laboratory (US)), Shaochun Tang(Brookhaven National Laboratory (US)), Tomas Vanat(Deutsches Elektronen-Synchrotron (DE)), Younes Otarid(CERN)
Tomas + Younes:
Tomas contacted Safiral to clarify details of the service:
Assembly + Testing
PCB electrical tests and impedance matching control
Included by default --> Requested
Optical inspection (AOI) of the soldered components
Included by default --> Requested
X-ray screening
Possible with extra costs --> Discarded
Functional tests
Possible but would lead to extra costs --> Discarded
IPC class 2 for PCB assembly
Pricing and order
Price scaling to 31 boards is linear, which means
776€ / board (includes PCB, assembly, components and services)
Components price might change as the current quotation is one month old. New invoice should be sent to the purchasing office
Request is ready with all 3 quotations and comments on chosen supplier.
Request from Safiral for pre-payment from CERN could lead to additional delays in production.
Will get in touch with the purchasing office to clarify situation
Eric:
Internal discussion at BNL about Caribou v2.0.
Conclusion was, while Max did a lot of work, the design is incomplete with very floaty design choices.
Currently reiterating on requirements and design ideas that are still to be defined
Will prepare a presentation for the coming weeks that shall put all Developers on the same page about the v2.0 design and plans
16:35
→
17:00
Firmware25m
Speakers:
Eric Buschmann(Brookhaven National Laboratory (US)), Tomas Vanat(Deutsches Elektronen-Synchrotron (DE)), Younes Otarid(CERN)
Younes:
Improved the architecture to support multiple boards
Initially architecture was separated into two hierarchical structures, one for each board. This was done because of an issue with locked IPs and imcompatibility between the configured target device and the constraints files included in the synthesis flow. And the correct architecture was picked by defining the top module corresponding to the chosen board.
Disadvantage is that it leads to some code duplication
Solution was:
merge the two structures into a single top file
use generate statements to instantiate board PS IPs
exclude the PS IPs of the non-targeted boards from the synthesis workflow.
This way, the Boreal firmware relies on a single hierarchical structure, and all the needed configuration is handled by the BorealManager
Working on moving the existing common blocks (Patter Generator, Pulse Generator, Syncers, ...)
Included linting into BorealManager
17:00
→
17:25
Software - OS25m
Speakers:
Eric Buschmann(Brookhaven National Laboratory (US)), Mathieu Benoit(Oak Ridge National Laboratory (ORNL)), Simon Spannagel(Deutsches Elektronen-Synchrotron (DE)), Younes Otarid(CERN)
Younes:
Installed PetaLinux build tools
Started looking at the build workflow, working on understanding the needed configurations, and how to include Peary as a layer
Objectives are:
Understand the complete build workflow
Build an image for one ZCU102 (MPSoC) board
Implement a streamlined and automatized build workflow
Reiterate for the ZC706 (Zynq) board to close on support of all boards