Usage¶

CPU stream quickstart¶

import numpy as np
import pyshmem

writer = pyshmem.create("frames", shape=(480, 640), dtype=np.float32)
reader = pyshmem.open("frames")

writer.write(np.ones((480, 640), dtype=np.float32))
frame = reader.read()

GPU stream quickstart¶

import numpy as np
import pyshmem

writer = pyshmem.create(
    "activations",
    shape=(1024, 1024),
    dtype=np.float32,
    gpu_device="cuda:0",
)
reader = pyshmem.open("activations", gpu_device="cuda:0")

writer.write(np.ones((1024, 1024), dtype=np.float32))
activation = reader.read()

Reading modes¶

read(safe=True) returns a consistent snapshot of the latest completed write.

read(safe=False) exposes the live backing storage and therefore requires the caller to hold the stream lock first.

with reader.locked():
    raw = reader.read(safe=False)

Waiting for the next write¶

Use read_new when you want to block until a new payload arrives.

next_frame = reader.read_new(timeout=1.0)

Lifecycle¶

close() releases only the current handle
unlink() destroys the shared-memory stream
delete() is an alias for unlink()

Be conservative with unlink() in multi-process systems: any attached handle can destroy the stream.