GpuDiagnosticSink
Delegates · AiDotNet.Configuration
Delegate that receives GPU backend diagnostic messages in lieu of
WriteLine. Applications register a sink
to route GPU diagnostics through their logging framework of choice
(Spectre.Console, Serilog, Microsoft.Extensions.Logging, structured
logs, etc.).
Inheritance: Delegate → MulticastDelegate → GpuDiagnosticSink
How It Works
Addresses Option C of github.com/ooples/AiDotNet#1122: “Use ILogger instead of Console.WriteLine — Then applications control output through their logging framework.”
Forward-compatibility: the sink is captured in
Sink immediately. When the underlying
AiDotNet.Tensors package supports sink routing (v0.39+), the diagnostic
messages are delivered to the sink WITH level tagging. On current
Tensors v0.38.0, the sink is stored but the Console.WriteLine calls
in the Tensors layer still go to Console directly; the bool-level
gate (Verbose) still suppresses
them when Silent or
Minimal.
For Microsoft.Extensions.Logging integration, see
ToSink
which wraps an ILogger instance as a sink.
Constructors
| Constructor | Summary |
|---|---|
GpuDiagnosticSink(object object, IntPtr method) |
Methods
| Method | Summary |
|---|---|
IAsyncResult BeginInvoke(GpuDiagnosticLevel level, string message, AsyncCallback callback, object object) |
|
void EndInvoke(IAsyncResult result) |
|
void Invoke(GpuDiagnosticLevel level, string message) |