|
MayaFlux 1.0.0
Digital-First Multimedia Processing Framework
|
The stages that run the tensor kernels. More...
#include "MayaFlux/Yantra/Executors/GpuDispatchCore.hpp"#include "Layers.hpp"#include "Layouts.hpp"Go to the source code of this file.
Namespaces | |
| namespace | MayaFlux |
| Main namespace for the Maya Flux audio engine. | |
| namespace | MayaFlux::Yantra |
| namespace | MayaFlux::Yantra::TensorKernels |
| namespace | MayaFlux::Yantra::TensorKernels::detail |
Functions | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::add_stage (uint64_t a, uint64_t b, uint64_t dst, uint32_t count) |
| One nn_add dispatch: the element-wise sum of two sequences of count floats. | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::attention_mix_stage (uint64_t weights, uint64_t v, uint64_t dst, uint32_t heads, uint32_t dim, uint32_t time) |
| One nn_attention_mix dispatch: the values of every position mixed by each query's attention weights. | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::attention_scores_stage (uint64_t q, uint64_t k, uint64_t scores, uint32_t heads, uint32_t dim, uint32_t time, float scale) |
| One nn_attention_scores dispatch: scale times the dot product of every query with every key, per head. | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::conv1d_stage (const Conv1d &layer, uint64_t src, uint64_t weights, uint64_t bias, uint64_t dst, uint32_t batch, uint32_t t_in) |
| One nn_conv1d dispatch: [batch, c_in, t_in] to [batch, c_out, conv1d_length]. | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::frame_audio_stage (uint64_t samples, uint64_t frames, uint32_t chunks, uint32_t chunk_size, uint32_t context, uint32_t pad) |
| One nn_frame_audio dispatch: cut a sample stream into overlapping frames, reflect padded on the right. | |
| std::array< uint32_t, 3 > | MayaFlux::Yantra::TensorKernels::detail::group_grid (size_t invocations, uint32_t local) |
| Workgroup counts covering the invocations, along x up to max_groups_per_axis and then along y. | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::grouped_conv1d_stage (const GroupedConv1d &layer, uint64_t src, uint64_t weights, uint64_t bias, uint64_t dst, uint32_t batch, uint32_t t_in) |
| One nn_conv1d_grouped dispatch: [batch, c_in, t_in] to [batch, c_out, grouped_conv1d_length]. | |
| uint32_t | MayaFlux::Yantra::TensorKernels::detail::groups_for (size_t invocations, uint32_t local) |
| DependencyStage | MayaFlux::Yantra::TensorKernels::lstm_sequence_stage (uint64_t x, uint64_t w_ih, uint64_t w_hh, uint64_t b_ih, uint64_t b_hh, uint64_t state, uint64_t hs, uint32_t steps, uint32_t hidden) |
| One nn_lstm_sequence dispatch: an LSTM cell over a whole sequence in one workgroup. | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::magnitude_stage (uint64_t src, uint64_t dst, uint32_t batch, uint32_t channels, uint32_t time) |
| One nn_magnitude dispatch: sqrt(re^2 + im^2) of a complex signal stored as two channel halves. | |
| template<typename PC> | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::detail::make_grid_stage (const char *shader, const PC &pc, size_t invocations) |
| template<typename PC> | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::detail::make_stage (const char *shader, uint32_t local, const PC &pc, uint32_t groups) |
| DependencyStage | MayaFlux::Yantra::TensorKernels::normalize_apply_stage (uint64_t src, uint64_t stats, uint64_t weight, uint64_t bias, uint64_t dst, uint32_t outer, uint32_t extent, uint32_t inner, NormAffine affine, Activation activation=Activation::None) |
| The nn_normalize dispatch that writes (x - mean) * scale from the measures of normalize_stats_stage. | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::normalize_stats_stage (uint64_t src, uint64_t stats, uint32_t outer, uint32_t extent, uint32_t inner, float epsilon) |
| The nn_normalize dispatch that measures every row of a tensor [outer, extent, inner] along extent. | |
| DependencyStage | MayaFlux::Yantra::TensorKernels::softmax_stage (uint64_t data, uint32_t outer, uint32_t extent, uint32_t inner) |
| One nn_softmax dispatch: softmax of a tensor [outer, extent, inner] along extent, in place. | |
The stages that run the tensor kernels.
Each function returns a DependencyStage for record_stages, so a network is a list of stages recorded into one submission. Addresses are device addresses of float32 buffers, from a context's shared buffers (shared_buffer_address).
Definition in file Stages.hpp.