Bug 727 - Convert SimdSignal to use full SIMT
Summary: Convert SimdSignal to use full SIMT
Status: DEFERRED
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: Jacob Lifshay
URL:
Depends on:
Blocks:
 
Reported: 2021-10-13 02:09 BST by Jacob Lifshay
Modified: 2021-10-13 21:40 BST (History)
2 users (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
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Jacob Lifshay 2021-10-13 02:09:31 BST 
https://bugs.libre-soc.org/show_bug.cgi?id=713#c89
Comment 1 Luke Kenneth Casson Leighton 2021-10-13 21:18:37 BST 
SIMT is a type of broadcast-instruction, a single L1 I-cache that
broadcasts an instruction to multiple execution units that are
otherwise fully-functioning cores.

if you meant "SimdSignal should be converted to full SIMD" that's
always been the plan so it will never need "conversion" to something
that it already is, by design.

either way this bugreport is invalid, so closing.
Comment 2 Jacob Lifshay 2021-10-13 21:37:56 BST 
(In reply to Luke Kenneth Casson Leighton from comment #1)
> SIMT is a type of broadcast-instruction, a single L1 I-cache that
> broadcasts an instruction to multiple execution units that are
> otherwise fully-functioning cores.

You missed my point, which is to convert SimdSignal to have an API like SIMT's from a game programmer perspective, we sent dozens of messages over what exactly I meant in #713.

This is basically the bug for the future improvements that you and David convinced me to put off for later in the virtual coffee meeting yesterday. David explained that you thought those improvements were a good idea, so I'm reopening this bug.
Comment 3 Jacob Lifshay 2021-10-13 21:40:32 BST 
(In reply to Jacob Lifshay from comment #2)
> (In reply to Luke Kenneth Casson Leighton from comment #1)
> > SIMT is a type of broadcast-instruction, a single L1 I-cache that
> > broadcasts an instruction to multiple execution units that are
> > otherwise fully-functioning cores.
> 
> You missed my point, which is to convert SimdSignal to have an API like
> SIMT's from a game programmer perspective, we sent dozens of messages over
> what exactly I meant in #713.

To be clear, this has nothing to do with how GPU hardware implements SIMT, it has to do with what our HDL will look like, like how SIMT appears from a software perspective.