CODA re-expresses most non-attention Transformer computations as GEMM-plus-epilogue programs using a constrained set of composable primitives to keep intermediate results on-chip and cut global memory traffic.
p r e p a r e _ t m a ( 127tma_op = " s2g " , 128e pi _t il e = epi_tile , 129e p i _ s t a g e = epi_stage , 130e p i _ t e n s o r = mPostAct , 131) 132 133if cutlass
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs
CODA re-expresses most non-attention Transformer computations as GEMM-plus-epilogue programs using a constrained set of composable primitives to keep intermediate results on-chip and cut global memory traffic.