Skip to content

Causal depthwise conv1d in CUDA, with a PyTorch interface

License

Notifications You must be signed in to change notification settings

neuralmagic/causal-conv1d

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

64 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Causal depthwise conv1d in CUDA with a PyTorch interface

Features:

  • Support fp32, fp16, bf16.
  • Kernel size 2, 3, 4.

How to use

from causal_conv1d import causal_conv1d_fn
def causal_conv1d_fn(x, weight, bias=None, activation=None):
    """
    x: (batch, dim, seqlen)
    weight: (dim, width)
    bias: (dim,)
    activation: either None or "silu" or "swish"

    out: (batch, dim, seqlen)
    """

Equivalent to:

import torch.nn.functional as F

F.conv1d(x, weight.unsqueeze(1), bias, padding=width - 1, groups=dim)[..., :seqlen]

Additional Prerequisites for AMD cards

Patching ROCm

If you are on ROCm 6.0, run the following steps to avoid errors during compilation. This is not required for ROCm 6.1 onwards.

  1. Locate your ROCm installation directory. This is typically found at /opt/rocm/, but may vary depending on your installation.

  2. Apply the Patch. Run with sudo in case you encounter permission issues.

     patch /opt/rocm/include/hip/amd_detail/amd_hip_bf16.h < rocm_patch/rocm6_0.patch 

About

Causal depthwise conv1d in CUDA, with a PyTorch interface

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Cuda 48.4%
  • Python 30.7%
  • C++ 18.2%
  • C 2.7%