DetessCast.h File
DetessCast Node — fused CVU kernel that detessellates and casts BF16→FP32. More...
Included Headers
#include "builder/Node.h"
#include "builder/NodeContractConfigurable.h"
#include "builder/NodeContractProvider.h"
#include "builder/OutputSpec.h"
#include <nlohmann/json.hpp>
#include <memory>
#include <optional>
#include <string>
#include <vector>
Namespaces Index
| namespace | simaai |
| namespace | neat |
| namespace | nodes |
Classes Index
| struct | DetessCastOptions |
|
Construction options for a DetessCast Node. More... | |
| class | DetessCast |
|
Fused CVU kernel Node: Detess followed by Cast (BF16→FP32). More... | |
Description
DetessCast Node — fused CVU kernel that detessellates and casts BF16→FP32.
Combines an MLA-layout untessellate with a BF16→FP32 cast in a single CVU pass. Inserted after MLA stages on the BF16 output path when downstream wants natural HWC/CHW float tensors — fusing the two stages saves a DDR round-trip versus running Detess then Cast.
- See Also
"The dtype contract" page in /concepts/dtype_contract
Generated via doxygen2docusaurus 2.0.0 by Doxygen 1.9.8.