MCPcopy Create free account
hub / github.com/ml-explore/mlx / get_quantized_kernel_wrapped

Function get_quantized_kernel_wrapped

mlx/backend/metal/quantized.cpp:20–34  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

18
19template <typename... Args>
20auto get_quantized_kernel_wrapped(
21 metal::Device& d,
22 const std::string& name,
23 const std::string& func,
24 const std::string& mode,
25 const std::string& type,
26 int group_size,
27 int bits,
28 Args... args) {
29 std::string template_def;
30 std::string fname = ((mode == "affine") ? "affine_" : "fp_") + func;
31 template_def = get_template_definition(
32 name, fname, type, group_size, bits, std::forward<Args>(args)...);
33 return get_quantized_kernel(d, name, template_def, mode);
34}
35
36template <typename... Args>
37auto get_qmm_nax_kernel_wrapped(

Callers 11

qmv_quadFunction · 0.85
qmvFunction · 0.85
qvm_split_kFunction · 0.85
qvmFunction · 0.85
qmmFunction · 0.85
qmm_splitkFunction · 0.85
gather_qmmFunction · 0.85
gather_qmvFunction · 0.85
gather_qvmFunction · 0.85
quantize_dequantizeFunction · 0.85
eval_gpuMethod · 0.85

Calls 2

get_template_definitionFunction · 0.85
get_quantized_kernelFunction · 0.70

Tested by

no test coverage detected