OffloadEngines [FreeBSD WIKI]

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== Offload Engines - Working Group – BSDCan 2012 ======
May 9th, 2012 9:00-12:00 EDT

{{devsummit/201205/offloadengines/offload_wg_2012.pdf?linkonly|Working Group Presentation}}

==== Attendees ====

  * [[/JimHarris|JimHarris]] (session leader)
  * [[/DaichiGoto|DaichiGoto]]
  * David Somayajulu (Qlogic)
  * Jason %%McMullan%% (Netronome)
  * [[/NeelNatu|NeelNatu]]
  * Mike Karels (%%McAfee%%)
  * [[/EdMaste|EdMaste]]
  * [[/RyanStone|RyanStone]]
  * Noah Robbin (Netronome)
  * [[/JustinGibbs|JustinGibbs]]
  * [[/JohnBaldwin|JohnBaldwin]]
  * [[/MarcelMoolenaar|MarcelMoolenaar]]
  * [[/PhilipPaeps|PhilipPaeps]] (tentative)
  * [[/KenMerry|KenMerry]]
  * Jeremiah Lott (Avere)
  * [[/WarnerLosh|WarnerLosh]]
  * [[/ErmalLuci|ErmalLuci]]
  * Chris Buechler (pfSense)
  * Scott Ullrich (pfSense)

==== Overview ====

Storage and network applications can benefit from CPU offload and/or acceleration engines for DMA, memfill, XOR/PQ.  Currently, %%FreeBSD%% does not have an infrastructure in place where these engines can be registered with the kernel, nor applications can generically take advantage of such engines if they exist.

The purpose of this working group would be to identify the following:
  * Critical areas in the kernel where offload/acceleration would most benefit
  * High-level architecture for an offload interface
  * Offload/acceleration engines that could be supported in %%FreeBSD%% %%**%% Note: I am currently working on a driver for Intel's IOAT v3 engines which support XOR/PQ, DMA, memfill
  * If/how much should this infrastructure coincide with crypto(9)

==== Agenda ====

| 9:00-9:15 | Introductions |
| 9:15-10:30 | Offload Engine Brainstorm |
| 10:30-11:00 | Break |
| 11:00-11:45 | Next Steps and Opens |
| 11:45-12:00 | Review Follow-up Items |

----

{{tag>Historical}}