Skip to content
Volten Volten

v.readDebug()

Use v.readDebug(node) after running a debug-enabled node.

const node = v.pass(kernel, bindings, { debug: true });


v.run(node);


const result = await v.readDebug(node);
Section titled “Print Logs”

For quick inspection, call print().

result.print();

Console output:

[1,0,0] vector: 0, 1, 2, 3

The prefix is the invocation id that produced the log: [gid.x,gid.y,gid.z].

Read Logs Yourself

Section titled “Read Logs Yourself”

logs is an array of decoded records.

for (const log of result.logs) {
    console.log(log.gid, log.kind, log.message, log.value);
}

Each log has:

| Field | Meaning | | --- | --- | | gid | Invocation id as [x, y, z] | | kind | Logged value kind, such as f32 or vec4f | | message | Optional message from the shader | | value | Number or number array |

Buffer Size

Section titled “Buffer Size”

Debug records are stored in a hidden debug buffer. The default payload size is 16 KiB.

Increase it when you expect more logs.

const node = v.pass(kernel, bindings, {
    debug: { bufferSize: 64 * 1024 }
});

bufferSize is measured in bytes.

Dropped Logs

Section titled “Dropped Logs”

If the debug buffer fills up, Volten keeps the records that fit and reports how many were dropped.

if (result.dropped > 0 || result.truncated) {
    console.warn('Some debug logs were dropped.');
}

| Field | Meaning | | --- | --- | | dropped | Records the shader tried to write after the buffer was full | | usedWords | Number of 32-bit words used by written records | | truncated | The recorded cursor moved past the available payload | | bufferSize | Debug payload size in bytes |