infini::ops::MoeWna16MarlinGemm Member List

This is the complete list of members for infini::ops::MoeWna16MarlinGemm, including all inherited members.

active_implementation_indices(Device::Type dev_type)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
active_implementation_indices(Device::Type dev_type)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Call(const Handle &handle, const Config &config, const Args &... args)	infini::ops::Operator< MoeWna16MarlinGemm >	static
Call(const Tensor tensor, const Args &... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Call(const TensorLike &tensor, const Args &... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Call(const Handle &handle, const Config &config, const Args &... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Call(const Tensor tensor, const Args &... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Call(const TensorLike &tensor, const Args &... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
clear_cache()	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
clear_cache()	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
config_ptr_	infini::ops::OperatorBase	protected
device_index_	infini::ops::MoeWna16MarlinGemm	protected
device_type_	infini::ops::Operator< MoeWna16MarlinGemm >	protectedstatic
handle_ptr_	infini::ops::OperatorBase	protected
implementation_index_	infini::ops::Operator< MoeWna16MarlinGemm >	protectedstatic
Make(const Config &config, const Tensor tensor, Args &&... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Make(const Tensor tensor, Args &&... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Make(const Config &config, const std::vector< Tensor > tensors, Args &&... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Make(const std::vector< Tensor > tensors, Args &&... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Make(const Config &config, const Tensor tensor, Args &&... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Make(const Tensor tensor, Args &&... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Make(const Config &config, const std::vector< Tensor > tensors, Args &&... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
Make(const std::vector< Tensor > tensors, Args &&... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inlinestatic
MoeWna16MarlinGemm(const Tensor a, const Tensor b_q_weight, std::optional< Tensor > b_bias_or_none, const Tensor b_scales, std::optional< Tensor > a_scales, std::optional< Tensor > global_scale, std::optional< Tensor > b_zeros_or_none, std::optional< Tensor > g_idx_or_none, std::optional< Tensor > perm_or_none, const Tensor workspace, const Tensor sorted_token_ids, const Tensor expert_ids, const Tensor num_tokens_past_padded, const Tensor topk_weights, const int64_t moe_block_size, const int64_t top_k, const bool mul_topk_weights, const int64_t b_type_id, const int64_t size_m, const int64_t size_n, const int64_t size_k, const bool is_full_k, const bool use_atomic_add, const bool use_fp32_reduce, const bool is_zp_float, const int64_t thread_k, const int64_t thread_n, const int64_t blocks_per_sm, Tensor out)	infini::ops::MoeWna16MarlinGemm	inline
operator()(const Tensor a, const Tensor b_q_weight, std::optional< Tensor > b_bias_or_none, const Tensor b_scales, std::optional< Tensor > a_scales, std::optional< Tensor > global_scale, std::optional< Tensor > b_zeros_or_none, std::optional< Tensor > g_idx_or_none, std::optional< Tensor > perm_or_none, const Tensor workspace, const Tensor sorted_token_ids, const Tensor expert_ids, const Tensor num_tokens_past_padded, const Tensor topk_weights, const int64_t moe_block_size, const int64_t top_k, const bool mul_topk_weights, const int64_t b_type_id, const int64_t size_m, const int64_t size_n, const int64_t size_k, const bool is_full_k, const bool use_atomic_add, const bool use_fp32_reduce, const bool is_zp_float, const int64_t thread_k, const int64_t thread_n, const int64_t blocks_per_sm, Tensor out) const =0	infini::ops::MoeWna16MarlinGemm	pure virtual
Operator< MoeWna16MarlinGemm >::operator()(const Handle &handle, const Args &... args)	infini::ops::Operator< MoeWna16MarlinGemm >	inline
Operator< MoeWna16MarlinGemm >::operator()(const Args &... args) const	infini::ops::Operator< MoeWna16MarlinGemm >	inline
set_config(const Config &config)	infini::ops::OperatorBase	inline
set_config(const Config &config)	infini::ops::OperatorBase	inline
set_handle(const Handle &handle)	infini::ops::OperatorBase	inline
set_handle(const Handle &handle)	infini::ops::OperatorBase	inline
set_stream(void *stream)	infini::ops::OperatorBase	inline
set_stream(void *stream)	infini::ops::OperatorBase	inline
set_workspace(void *workspace)	infini::ops::OperatorBase	inline
set_workspace(void *workspace)	infini::ops::OperatorBase	inline
set_workspace_size_in_bytes(std::size_t workspace_size_in_bytes)	infini::ops::OperatorBase	inline
set_workspace_size_in_bytes(std::size_t workspace_size_in_bytes)	infini::ops::OperatorBase	inline
stream_	infini::ops::OperatorBase	protected
ValidateCallMetadata(const Tensor a, const Tensor b_q_weight, std::optional< Tensor > b_bias_or_none, const Tensor b_scales, std::optional< Tensor > a_scales, std::optional< Tensor > global_scale, std::optional< Tensor > b_zeros_or_none, std::optional< Tensor > g_idx_or_none, std::optional< Tensor > perm_or_none, const Tensor workspace, const Tensor sorted_token_ids, const Tensor expert_ids, const Tensor num_tokens_past_padded, const Tensor topk_weights, const int64_t moe_block_size, const int64_t top_k, const bool mul_topk_weights, const int64_t b_type_id, const int64_t size_m, const int64_t size_n, const int64_t size_k, const bool is_full_k, const bool use_atomic_add, const bool use_fp32_reduce, const bool is_zp_float, const int64_t thread_k, const int64_t thread_n, const int64_t blocks_per_sm, Tensor out) const	infini::ops::MoeWna16MarlinGemm	inlineprotected
workspace_	infini::ops::OperatorBase	protected
workspace_size_in_bytes() const	infini::ops::OperatorBase	inlinevirtual
workspace_size_in_bytes() const	infini::ops::OperatorBase	inlinevirtual
workspace_size_in_bytes_	infini::ops::OperatorBase	protected
~OperatorBase()=default	infini::ops::OperatorBase	virtual
~OperatorBase()=default	infini::ops::OperatorBase	virtual