Skip to content

feat: switch to 32-bit cell model with narrow-type load/store#53

Closed
tetsuo-cpp wants to merge 1 commit intocanonfrom
feat/32bit-cell-model
Closed

feat: switch to 32-bit cell model with narrow-type load/store#53
tetsuo-cpp wants to merge 1 commit intocanonfrom
feat/32bit-cell-model

Conversation

@tetsuo-cpp
Copy link
Owner

Summary

  • Switch all arithmetic from i64/f64 to i32/f32 for native GPU throughput (stack stays memref<256xi64> for 64-bit pointer round-tripping)
  • Add narrow load/store words: HF@/HF!, BF@/BF!, I8@/I8!, I16@/I16! (global) and SHF@/SHF!, SBF@/SBF!, SI8@/SI8!, SI16@/SI16! (shared)
  • Update all tests and warpforth-runner for the new cell model

Closes #52

@tetsuo-cpp tetsuo-cpp closed this Feb 22, 2026
@tetsuo-cpp tetsuo-cpp deleted the feat/32bit-cell-model branch February 22, 2026 13:17
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.

Add load/store words for reduced-width memory types

1 participant