Commit graph

7 commits

Author SHA1 Message Date
Hannes Karppila
1c885cf0f1
codegen optimization: symbolic fuel-vm interpretation (#7109)
## Description

This PR adds an initial support to symbolic fuel-vm interpretation to
improve codegen. The symbolic execution is designed to be extended in
later PRs. In it's current form, it provides three distinct
improvements:

1. If a temporary (allocated) register contains a value already held in
another register, we can use that other register instead. This reduces
register pressure, and removes unnecessary `mov` ops.
2. If a register already has the value we're setting it to, we can
eliminate the instruction doing so.
3. For conditional jump instructions, if the value of the condition is
known, one of the branches can be elminated. In such cases, the
instruction is either removed or replaced with an unconditional jump.
Dead code elimination pass can then potentially remove the
now-unreachable branch.

In addition to these changes, this PR takes the optimization level into
account for asm generation. For debug builds, we only run the
optimization passes once. But for release builds we run the optimization
passes multiple times, until we either cannot eliminate any more
operations, or until a fixed upper limit of rounds. For instance, this
means that if the new pass added in this PR manages to eliminate a
branch, and then dead code elimination removes that code, it could allow
the interpretation pass to do even more work.

To account for new optimization passes, I also organized them to mulple
files, so that the ones which create private types are scoped well and
thus easier to read.


## Impact

project | size before | size after | size reduction
-|-|-|-
mira-v1-core | 89.384 KB | 83.480 KB | 6.6%


## Checklist

- [x] I have linked to any relevant issues.
- [x] I have commented my code, particularly in hard-to-understand
areas.
- [x] I have updated the documentation where relevant (API docs, the
reference, and the Sway book). No changes required!
- [x] If my change requires substantial documentation changes, I have
[requested support from the DevRel
team](https://github.com/FuelLabs/devrel-requests/issues/new/choose)
- [x] I have added tests that prove my fix is effective or that my
feature works. Just updated pre-existing ones!
- [x] I have added (or requested a maintainer to add) the necessary
`Breaking*` or `New Feature` labels where relevant.
- [x] I have done my best to ensure that my PR adheres to [the Fuel Labs
Code Review
Standards](https://github.com/FuelLabs/rfcs/blob/master/text/code-standards/external-contributors.md).
- [x] I have requested a review from the relevant team or maintainers.

---------

Co-authored-by: Vaivaswatha N <vaivaswatha.nagaraj@fuel.sh>
Co-authored-by: Joshua Batty <joshpbatty@gmail.com>
2025-05-05 10:50:08 +03:00
Hannes Karppila
37235f2e7d
codegen: Use idiomatic fuel-vm ops for memcpy and addr_of (#6795)
## Description

SIgnificantly reduces bytecode sizes by transforming the following
commonly-used operations when the constant value fits into the immediate
part of the `*i` variant of the instruction:

* `MOVI(tmp, copy_len); MCP(dst, src, tmp)` → `MCPI(dst, src, copy_len)`
(in `compile_mem_copy_bytes`)
* `MOVI(dst, offset); ADD(dst, dst, DATA_SECTION)` → `ADDI(dst,
DATA_SECTION, offset)` (in `addr_of`)

## Checklist

- [ ] I have linked to any relevant issues. (none that I know of)
- [x] I have commented my code, particularly in hard-to-understand
areas.
- [ ] I have updated the documentation where relevant (API docs, the
reference, and the Sway book). (no changes required)
- [ ] If my change requires substantial documentation changes, I have
[requested support from the DevRel
team](https://github.com/FuelLabs/devrel-requests/issues/new/choose)
- [ ] I have added tests that prove my fix is effective or that my
feature works. (partially; some values in tests got smaller. we don't
seem to be tracking the size and size regressions in tests)
- [x] I have added (or requested a maintainer to add) the necessary
`Breaking*` or `New Feature` labels where relevant. (not relevant)
- [x] I have done my best to ensure that my PR adheres to [the Fuel Labs
Code Review
Standards](https://github.com/FuelLabs/rfcs/blob/master/text/code-standards/external-contributors.md).
- [x] I have requested a review from the relevant team or maintainers.
2025-01-16 10:49:25 +11:00
Vaivaswatha N
9741809659
Add configurables section offset in the preamble (#6709)
Re-enabling this feature after a bugfix. (reverted PR: #6522)

---------

Co-authored-by: Sophie Dankel <47993817+sdankel@users.noreply.github.com>
Co-authored-by: IGI-111 <igi-111@protonmail.com>
2024-11-18 12:33:47 +05:30
IGI-111
d7dd104dac
Bump to v0.66.4 (#6688) 2024-11-03 17:12:45 +01:00
Vaivaswatha N
3ff63a59c1
EncodeBufferAppend: grow_if_needed should allocate sufficiently (#6686) 2024-11-02 11:47:58 +01:00
Vaivaswatha N
0555973393
Add configurables section offset in the preamble (#6522)
The preamble now contains 8 bytes of offset to the configurables
section. If there is no configurable const in the data-section, then the
value of this integer will be equal to the size of the binary itself.

This also means that we now sort the data-section to have all the
non-configurables first, and then the configurables.

The preamble final asm looks like this (the offset to configurables is
0'd out here in the example):
```
;; ASM: Final program
;; Program kind: Script
.program:
move $$tmp $pc
jmpf $zero i10
DATA_SECTION_OFFSET[0..32]
DATA_SECTION_OFFSET[32..64]
CONFIGURABLES_OFFSET[0..32]
CONFIGURABLES_OFFSET[32..64]
lw   $$ds $$tmp i1
add  $$ds $$ds $$tmp
```
The preamble bytecode looks like this:

```
0x00000000 MOVE R60 $pc                                    ;; [26, 240, 48, 0]
0x00000004 JMPF $zero 0xa                                  ;; [116, 0, 0, 10]
0x00000008                                                 ;; [0, 0, 0, 0, 0, 0, 2, 40]
0x00000010                                                 ;; [0, 0, 0, 0, 0, 0, 0, 0]
0x00000030 LW R63 R60 0x1                                  ;; [93, 255, 192, 1]
0x00000034 ADD R63 R63 R60                                 ;; [16, 255, 255, 0]
...
```

---------

Co-authored-by: Sophie Dankel <47993817+sdankel@users.noreply.github.com>
Co-authored-by: IGI-111 <igi-111@protonmail.com>
2024-10-29 20:16:33 +01:00
Sophie Dankel
7ab1565498
feat: Add --raw-logs option for forc-test (#6528)
## Description

Closes https://github.com/FuelLabs/sway/issues/6015

- Adds the `--raw-logs` option
- Removes the `decode` option as this is now the default for `--logs`
- Shortens `--pretty-print` to `--pretty` 

## Checklist

- [x] I have linked to any relevant issues.
- [ ] I have commented my code, particularly in hard-to-understand
areas.
- [ ] I have updated the documentation where relevant (API docs, the
reference, and the Sway book).
- [ ] If my change requires substantial documentation changes, I have
[requested support from the DevRel
team](https://github.com/FuelLabs/devrel-requests/issues/new/choose)
- [x] I have added tests that prove my fix is effective or that my
feature works.
- [ ] I have added (or requested a maintainer to add) the necessary
`Breaking*` or `New Feature` labels where relevant.
- [ ] I have done my best to ensure that my PR adheres to [the Fuel Labs
Code Review
Standards](https://github.com/FuelLabs/rfcs/blob/master/text/code-standards/external-contributors.md).
- [x] I have requested a review from the relevant team or maintainers.
2024-09-13 15:33:18 -07:00