For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Mojo function

causal_conv1d_channel_first_fwd_gpu_no_bias

def causal_conv1d_channel_first_fwd_gpu_no_bias[x_dtype: DType, weight_dtype: DType, output_dtype: DType, kNThreads: Int, kWidth: Int, kNElts: Int, x_LT: TensorLayout, weight_LT: TensorLayout, output_LT: TensorLayout](batch: Int, dim: Int, seqlen: Int, width: Int, x: TileTensor[x_dtype, x_LT, MutUntrackedOrigin], weight: TileTensor[weight_dtype, weight_LT, MutUntrackedOrigin], output: TileTensor[output_dtype, output_LT, MutUntrackedOrigin], x_batch_stride: UInt32, x_c_stride: UInt32, x_l_stride: UInt32, weight_c_stride: UInt32, weight_width_stride: UInt32, out_batch_stride: UInt32, out_c_stride: UInt32, out_l_stride: UInt32, silu_activation: Int8)

Optimized causal conv1d implementation for channel first data layout using SIMD operations (no bias).

Key optimizations:

SIMD vectorization for input/output operations
Efficient memory access patterns with coalesced loads
Vectorized weight loading and computation
Optimized activation function with SIMD operations
Better thread utilization and memory bandwidth usage

Grid: (ceildiv(seqlen, kNThreads * kNElts), dim, batch) Block: kNThreads

Parameters:

x_dtype (DType): Element type of the input tensor x.
weight_dtype (DType): Element type of the weight tensor weight.
output_dtype (DType): Element type of the output tensor output.
kNThreads (Int): Number of threads per block used to process the sequence dimension.
kWidth (Int): Compile-time convolution kernel width; must match the runtime width argument.
kNElts (Int): Number of sequence elements each thread processes, used for SIMD vectorization and ILP.
x_LT (TensorLayout): TensorLayout of the input tensor x.
weight_LT (TensorLayout): TensorLayout of the weight tensor weight.
output_LT (TensorLayout): TensorLayout of the output tensor output.

Args:

batch (Int): Batch size.
dim (Int): Number of channels.
seqlen (Int): Sequence length.
width (Int): Kernel width (must match kWidth compile-time parameter).
x (TileTensor[x_dtype, x_LT, MutUntrackedOrigin]): Input tensor of shape (B, C, L).
weight (TileTensor[weight_dtype, weight_LT, MutUntrackedOrigin]): Weight tensor of shape (C, W).
output (TileTensor[output_dtype, output_LT, MutUntrackedOrigin]): Output tensor of shape (B, C, L).
x_batch_stride (UInt32): Stride for the batch dimension of the input tensor.
x_c_stride (UInt32): Stride for the channel dimension of the input tensor.
x_l_stride (UInt32): Stride for the sequence length dimension of the input tensor.
weight_c_stride (UInt32): Stride for the channel dimension of the weight tensor.
weight_width_stride (UInt32): Stride for the width dimension of the weight tensor.
out_batch_stride (UInt32): Stride for the batch dimension of the output tensor.
out_c_stride (UInt32): Stride for the channel dimension of the output tensor.
out_l_stride (UInt32): Stride for the sequence length dimension of the output tensor.
silu_activation (Int8): Whether to apply SiLU activation (Int8: 0 or 1).