SGEMM-cube approximates FP32 GEMM on Ascend NPUs via a two-component FP16 splitting strategy, recovering near-FP32 accuracy for moderate-range inputs at up to 65.3 TFLOP/s (77% of the three-GEMM peak).
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
SGEMM-cube: Precision-Recovery FP32 GEMM Approximation on Ascend NPUs with FP16 Matrix Engines
SGEMM-cube approximates FP32 GEMM on Ascend NPUs via a two-component FP16 splitting strategy, recovering near-FP32 accuracy for moderate-range inputs at up to 65.3 TFLOP/s (77% of the three-GEMM peak).