Details
-
Task
-
Status: Closed
-
Major - P3
-
Resolution: Fixed
-
None
-
None
-
None
-
Fully Compatible
-
Execution Team 2021-07-12, Execution Team 2021-07-26, Execution Team 2021-08-09
Description
Regular Simple-8b will encode blocks in the buckets the following way
Nr ValueBits Elems Padding
|
1 1 60 0
|
2 2 30 0
|
3 3 20 0
|
4 4 15 0
|
5 5 12 0
|
6 6 10 0
|
7 7 8 4
|
8 8 7 4
|
9 10 6 0
|
10 12 5 0
|
11 15 4 0
|
12 20 3 0
|
13 30 2 0
|
14 60 1 0
|
The four padding bits in selector 7 and 8 can be used to extend the amount of selectors. We want to use them to describe a bit shift where large values with many trailing zeros can be compressed efficiently.
SelectorNr PaddingValue Elems ValueBits ShiftBits
|
7 1 9 2 4
|
7 2 8 3 4
|
7 3 7 4 4
|
7 4 6 5 4
|
7 5 5 7 4
|
7 6 4 10 4
|
7 7 3 14 4
|
7 8 2 24 4
|
7 9 1 52 4
|
Extend Simple-8b encoder/decoder to handle these extended selectors.
Extend calculation of optimal selector to handle these values too.