<div dir="ltr">Hi Team,<div><br></div><div>Sorry for the long delay after our ptg session. Here are the main takeaways from the meeting (both <a href="https://etherpad.openstack.org/p/cyborg-ptg-pike">f2f and virtual</a> ):</div><div><br></div><div><b>1. Pike Development Goal</b></div><div><b><br></b></div><div>We will shoot for an basic functional Cyborg code base for the Pike release cycle. We will build the initial modules based upon <a href="https://etherpad.openstack.org/p/cyborg-nova-interaction">Roman's proposal</a>. These initial modules are:</div><div><br></div><div><b>API</b>: as abstract as possible, and not necessary for multi-tenancy usage in Pike.</div><div><b>DB</b>: Maintain Cyborg specific device information (for example metadata extracted during the discovery phase)</div><div><b>Drivers</b>: support out-of-tree vendor solution in the long run, and try to come up with a generic driver (maybe based upon virtio interfaces) for Pike to get basic functionalities.</div><div><b>Agent</b>: the "compute" part of Cyborg thats directly talks to the accelerators/dedicated devices.<br clear="all"><div><br></div><div>We have designated the prime sponsors for these modules and look forward to <b>the initial BPs at the next team meeting (which will be Mar 15th, Wed this week)</b></div><div><br></div><div><b>2. Sandboxing proposals</b></div><div><br></div><div>Per agreed in previous meetings as well as ptg, we will setup a sandbox folder for the design proposals (we've covered one from Mellanox and one from Huawei)</div><div><br></div><div><b>3. Specific use case targeting</b></div><div><br></div><div>The task illustrated in section 1 is for building the common code structure, however we still need further work on specific use cases. Although not fully discussed during ptg meeting, <a href="https://drive.google.com/open?id=0B_Dc7PdTARsxc2cwTlJnelctWHM">Harm's proposal</a> will continue to be discussed as the main input for FPGA use case, whereas the <a href="https://wiki.openstack.org/wiki/ScientificWGGPUs">requirements</a> coming out of Scientific WG would be the main one for GPU. </div><div><br></div><div>We are looking at to provide basic solutions for these two use cases in Pike.</div><div><br></div><div><b>Reminder: our bi-weekly meeting resumes on Wed this week :)</b></div><div><br></div><div><br></div><div><br></div>-- <br><div class="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr">Zhipeng (Howard) Huang</div><div dir="ltr"><br></div><div dir="ltr">Standard Engineer</div><div>IT Standard & Patent/IT Prooduct Line</div><div dir="ltr">Huawei Technologies Co,. Ltd</div><div dir="ltr">Email: <a href="mailto:huangzhipeng@huawei.com" target="_blank">huangzhipeng@huawei.com</a></div><div dir="ltr">Office: Huawei Industrial Base, Longgang, Shenzhen</div><div dir="ltr"><br></div><div dir="ltr">(Previous)<br><div>Research Assistant</div><div>Mobile Ad-Hoc Network Lab, Calit2</div><div>University of California, Irvine</div><div>Email: <a href="mailto:zhipengh@uci.edu" target="_blank">zhipengh@uci.edu</a></div><div>Office: Calit2 Building Room 2402</div><div><br></div><div>OpenStack, OPNFV, OpenDaylight, OpenCompute Aficionado</div></div></div></div></div></div></div>
</div></div>