[Libre-soc-bugs] [Bug 1155] O(n^2) multiplication REMAP mode(s)

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=1155

--- Comment #37 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Jacob Lifshay from comment #36)
> so, should I start implementing the current design? 

i appreciate it's holidays (i've spent 40% of the past week asleep!)

gentle but very firm reminder that you are to read and re-read the
bug comments *online*, and when you do so you will observe things
that you missed when tired, such as comment #31 *already implementing*
the modulo REMAP Schedule.

> if it's not ready yet I
> think I'll work on the already established parts of the bigint presentation
> while I'm waiting.

good idea.

the next phase here is to design svshape3 to take the largest portion
of possible *most useful* options that will fit into the limited
space.

until that's done (the assessment not the coding)...

basically this can't be rushed.

the number of modes means that svshape3 will need to be more like svindex

 810 # svindex instruction  <a name="svindex"> </a>
 811 
 812 SVI-Form
 813 
 814 | 0:5|6:10 |11:15  |16:20 | 21:25       | 26:31 |  Form    |
 815 | -- | --  | ---   | ---- | ----------- | ------| -------- |
 816 | PO | SVG | rmm   | SVd  | ew/yx/mm/sk | XO    | SVI-Form |
 817 
 818 * svindex SVG,rmm,SVd,ew,SVyx,mm,sk

something like:

 814 | 0:5|6:10 |11:15  |16:20 | 21:25       | 26:31 |  Form    |
 815 | -- | --  | ---   | ---- | ----------- | ------| -------- |
 816 | PO | x   | y     | z    | XX/yx/XX/sk | XO    |TODO-Form |


where the two XXs give some options. although i would be tempted
to cut down z to  range 1-4 and have more bits for modes


 814 | 0:5|6:10 |11:15  |16:20 | 21:25       | 26:31 |  Form    |
 815 | -- | --  | ---   | ---- | ----------- | ------| -------- |
 816 | PO | x   | y     | z/XXX| XX/yx/XX/sk | XO    |TODO-Form |

the modes included from submode and submode2 are EIGHT bits and
that's excluding the offset (!) so it may be necessary to also
design svshape4 *as well* to cover an offset-priority version
just as svshape2 is a 2D offset priority version of the 3D svshape
instruction.

daaaamn.

-- 
You are receiving this mail because:
You are on the CC list for the bug.