Skip to content

Add best practice: Use BFLY for paired vector addition and subtraction - #49

Open
Rudi7503 wants to merge 1 commit into
ApolloTeam-dev:mainfrom
Rudi7503:feature/bfly-best-practice
Open

Rudi7503 wants to merge 1 commit into
ApolloTeam-dev:mainfrom
Rudi7503:feature/bfly-best-practice

Conversation

@Rudi7503

Copy link
Copy Markdown

Summary

This PR adds a new best-practice guide for using the 68080 AMMX BFLY (Butterfly) instruction (core/best-practices/use-ammx-bfly.md), specifically demonstrating how to accelerate symmetric butterfly stages in FFT, DCT, DWT, and audio processing by replacing split paddw/psubw pairs.

Key Insights & Hardware Evidence

Measurements conducted on physical Vampire V4 Standalone hardware (Core 11350 @ 92 MHz) using the Harness v6 benchmark suite:

  • Single-cycle execution: bflyw executes in 2.6 cycles (including 10-iteration loop overhead), outperforming split paddw/psubw (3.6 cycles) by 28%.
  • Folded EA load: Memory operand mode bflyw (a0),e8,e0:e1 folds effective address fetching with zero cycle penalty (2.6 cycles vs 4.6 cycles for explicit load + padd + psub, a 43% speedup).
  • Format: Conforms to AKB best-practice template, passes check_links.py, check_frontmatter.py, and snippet compiles cleanly with vasmm68k_mot 2.0f.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant