Bug 106 - pipeline early-in, early-out on concurrent function units
Summary: pipeline early-in, early-out on concurrent function units
Status: CONFIRMED
Alias: None
Product: Libre-SOC's first SoC
Classification: Unclassified
Component: ALU (including IEEE754 16/32/64-bit FPU) (show other bugs)
Version: unspecified
Hardware: PC Linux
: --- enhancement
Assignee: Luke Kenneth Casson Leighton
URL:
Depends on:
Blocks: 102 114
  Show dependency treegraph
 
Reported: 2019-06-29 15:45 BST by Luke Kenneth Casson Leighton
Modified: 2022-06-20 04:34 BST (History)
1 user (show)

See Also:
NLnet milestone: ---
total budget (EUR) for completion of task and all subtasks: 0
budget (EUR) for this task, excluding subtasks' budget: 0
parent task for budget allocation:
child tasks for budget allocation:
The table of payments (in EUR) for this task; TOML format:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Luke Kenneth Casson Leighton 2019-06-29 15:45:43 BST
the FP ALUs need to be re-useable for INT computation, by allowing
early-in (bypassing normalisation), getting at what would normally
be just the mantissa part of the pipeline, and jumping out.

whilst early-out is easy (due to the muxid), early-in may be challenging.
the simplest method, although it has the strong disadvantage of creating
unnecessary latency, is to just feed in the operands at the front of
the pipeline.

however, a timing-critical method of bringing the muxids from the later
stages up to the front (the Reservation Station) may also work.