Server/Mezz/Update Archive

From OpenCompute
Revision as of 01:09, 16 December 2019 by Damienchong (talk | contribs)
Jump to: navigation, search

Archive of updates

1/19/2018 update

1. Update spec draft to OCP_NIC_3.0_draft_0v63_20180119e_TN

1/18/2018 update

1. Update spec draft to OCP_NIC_3.0_draft_0v62_20180118f_TN

1/16/2018 update

1. Update spec draft to OCP_NIC_3.0_draft_0v60_20180116e_TN

1/15/2018 update

1. Update spec draft to OCP_NIC_3.0_draft_0v57_20180109c_TN

1/2/2018 update

1. Update spec draft to OCP_NIC_3.0_draft_0v50_20171222i_TN

2. Move updates before 12/1/2017 to archive

12/20/2017 update

1. Updated spec draft, overview deck, and mechanical deck

12/5/2017 update

1. Update spec draft to OCP_NIC_3.0_draft_0v01_20171205c_TN

2. Update pin out to OCP 3.0 Pinout Proposal 20171205a_TN

3. Specification draft complete rate (Estimation)

Chapter Draft complete rate Status
1. Overview 70% drafted finishing up / under review
2. Card form factor 60+% drafted finishing up / under review
3. Card Edge and baseboard interface 80+ % drafted Under review
4. Management 0% drafted Need to work on
5. Data Network requirement TBD Same content as Mezz 2.0; check for need to keep this chapter
6. Card routing guideline and SI consideration TBD NC-SI specified on card; PCIe TBD
7. Thermal and Environmental TBD Need edit based on Mezz 2.0

4. Key mechanical updates:

- Agreed on guiding feature and card sizes

- Reduce to 2x sizes (Small and Large form factors; detail see latest spec draft)

To be closed:

- Front bracket / latching mechanism design for small and large form factors

- Right Angle connector vertical offset(3.05mm vs. 3.50mm); Need to decide and close based on Tolerance analysis

- Straddle Mount connector offset and PCB footprint; Need to decide and close based on 1U use case study

5. Coming working sessions and proposal of agenda:

-#7 12/6 : 2hr working session; ME and EE runs in parallel; close on ME opens and clarify the rest of draft work

-#8 12/13: 3-4hr: All draft work due for 1st pass of edit of all chapters. 1st cross review with changes suggested and agreed

-#9 12/20: 3-4hr: Review changes and post 1st draft

11/20 update

1. Updated specification(11/12) and pin definition(11/15) to the "updated doc session"; The spec still needs a lot of work in the coming a few weeks.

2. Thermal deck v7 is uploaded (same material as presented on 11/15)

10/31 update

1. We did 4x working session so far and made good progress in pinout and mechanical profile

- 10/11 11:30am-3pm(PDT) - Done: TE/Dell-EMC/Intel/FB

- 10/18 9am-2pm(PDT) - Done: TE/Dell-EMC/Lenovo/Broadcom/Intel/FB

- 10/27 8am - 11am(PDT) Done: – TE/Dell-EMC/Lenovo/Broadcom/Intel/FB

- 10/31 8:30am - 10am(PDT) Done: - (Pinout specific session) Mellanox / Broadcom / Intel / Dell-EMC / FB

- 11/3 8am-11am(PDT)

- 11/15 9am-5pm(PST)

2. Latest pin out and ME deck are posted at Table above

3. Plan to post WIP specification draft after 11/3 workshop for broader review

4. We will cancel the 2017/11/01 meeting as FB team has a conflict

10/18 Update

- Intel/Dell-EMC/TE/FB joined the spec drafting working session on 10/11 and worked on these topics: framework, sections, pinout, connector, specification drafting logistics.

- With many inputs from community, FB updated Mechanical deck to v3; Key changes are 1) fine tuning module width by 2mm 2) proposing 1 extra sizes, and total 3x sizes with different width. Detail see Mechanical Discussion_v03_20171018

- Pinout is being actively working on. Intel is taking lead on proposing and maintaining the draft; an updated version can be found [here]

- We will move from Fuze to GoToWebinar for project calls in Nov. 10/18 call is likely the last call on Fuze. More details to follow in mailing list and calendar.

9/20 Update

- The subgroup closed on all other major options after a few months of iterations. #14 is the clear preference in subgroup. Details and specification will be worked out around this option.

- Update main deck: OCP_NIC_3.0_discussion_r10_20170920

- Update mechanical 3D package to r02

- Update mechanical deck to r02

9/6 Update

- Add updated doc session for better tracking of latest docs and community presentations

- Update general deck to v9, thermal deck to v6

- Intend to exit phase 2 by 9/20 with option #14

- Change project name from "OCP Mezz NIC 3.0" to "OCP NIC 3.0" as option #14 is not Mezz style

8/23 Update

HPE's presentation on 8/16 call

8/22 Update

Thermal results and files

- Option #13 with QSFP heat sink

- Option #14 with QSFP heat sink

- Slides - Update of thermal results

- Summary of updated simulation results

8/16 Update

Master Deck r08 dated 8/16/2017

Mechanical update r01 dated 8/16/2017

Mechanical - 3D of option 14 v1

8/2 Update

1. Update to deck

Master Deck r07 dated 8/1/2017

2. 8/2/2017 Meeting Agenda

General update

- Form factor stability goal (2+5+2 goal)

- Update the investigation of #14

- Revisit 1x or 2x form factor (tradeoff between simplicity and flexibility)

- Heads up of stating pin list discussion

Possible project name change

- One of the major option(#14) being looked at is not technical a "Mezz"

- Possible name change of project if the group is clearly trending toward #14

Mechanical intro - by Joshua Held @Facebook

- Josh will talk about coming activities in mechanical design of Mezz NIC

9/25 Workshop @ Dallas, TX

- Call for presentation; topical can be technically, or sharing how your experience with OCP Mezz NIC

- Suggest to drop me an email if you intend to attend (optional) - Free to register; instruction sent in mailing list


1. Update to deck

Master Deck V6 dated 7/25


1. Update to deck

Master Deck V5 dated 7/14

2. Update to Schedule

Schedule update v3 dated 7/14

3. Mezz subgroup will meet @ OCP Engineering workshop Sep 25th, 2017

More info here


See below material for tomorrow's call:

Corrected thermal model: Cases A1-A5

Corrected thermal model: Cases A6-A10

Thermal model with QSFP heat sink: Cases A1-A5

Thermal model with QSFP heat sink: Cases A6-A10

Summary of updated simulation results

Presentation slides

--- Agenda

1. Updates and corrections to thermal models

2. Impact of QSFP heat sink

3. Impact of contact resistance on QSFP heat sink performance

4. Go over a few topics for discussion


See below for material for tomorrow's call:

Summary of simulation results under stand-by power

Presentation slides

--- Agenda

1. Review latest round of simulation results

2. Go over a few topics for discussion


See below for material for tomorrow's call:

Updated thermal model: Cases A1-A5

Updated thermal model: Cases A6-A10

Updated thermal model: Cases D1-D8

Summary of simulation results using updated thermal models

Presentation slides

--- Agenda

1. Review changes to thermal models based on feedback from the community

2. Review latest round of simulation results

3. Go over a few topics for discussion


Here are the update in the past 2x weeks, and material for tomorrow's call:

Updated thermal simulation case deck:Thermal Simulation Feedback_20170517

Update thermal simulation case spreadsheet: Updated Thermal Simulation Cases_20170511

Additional questions to NIC suppliers:

- Standby(S5) power of NIC, or ratio of P_Standby / P_Max

- P_standby of Optical modules


As we kicked off the discussion of Mezz thermal simulation on 4/19, here are the updates:

OCP Mezz 3.0 Thermal Simulation Description

OCP Mezz 3.0 Thermal Simulation Package

Please give your feedback in mailing list and we will talk about it on next call on 5/3.


We have an updated schedule for comment and feedback:

Schedule Update V2

Mezz 3.0 Survey is closed with 5 participants; Summary with company names masked can be found here:

Mezz 3.0 Survey filled 201704016

A few updates were made to the master deck: Mezz 3.0 General Update

Our next call is on 4/19/2017 Wednesday 9am PDT(2nd call of the month). As always, call-in info can be found here.

--- Agenda

1. Sync with community about schedule refresher and get feedback for schedule and milestone

2. Summary of survey

3. John Fernandes will share methodology for thermal simulation

4. Call for more feedback on a few options (9,10,11,12)


1. Ramp NIC survey template version to v03 to have more exact wording. Template v03

2. Due to the limitation of mailing-list with attachment, please send the filled template to Jia will aggregate and update to the wiki based on your selection of the confidential level (cell B6,B7 of the template v03)

3. Schedule proposal v01 is available for review and comment. A few notes:

- Overall it includes 3x stages: Information collection => Narrow down direction => work out details and 1st draft

- The v01 schedule is on the aggressive side, i.e. the best case scenario when everything goes as expected.

- Please review it, give inputs in mailing-list, and discuss in the call next week.


An improved version of survey template v02 is created. All NIC suppliers are strongly recommend to participate to influence Mezz 3.0. Inputs from System supplier and Hyper-scale/CSP are welcome as well.

Agenda for next meeting (4/5)

1. Review overall Mezz 3.0 work schedule

2. Review the boundary with the context of major options

3. Pick major option branches to perform more detailed study


A survey of boundary condition of typical use cases is going on.

Detail see here

Please reply to mailing-list to share your boundary condition of Mezz3.0 or write to Jia Ning


1. Video of 3/17/2017 OCP Server workshop talk - Mezz 2.0 and beyond

2. Slide deck of 3/9/2017 workshop

3. Visco file for placement study

Start Test Code

End Test Code