@octomil/browser 1.2.0 → 1.3.0

package/dist/index.cjs

@@ -34702,17 +34702,14 @@ async function runLocalGeneration(request, config) {
   return extractGeneratedText(generation);
 }
 async function getGenerator(config) {
-  const key = JSON.stringify([
-    config.runtimeModel,
-    resolveDevice(config.device),
-    config.dtype
-  ]);
+  const device = await resolveDevice(config.device);
+  const key = JSON.stringify([config.runtimeModel, device, config.dtype]);
   let pending = pipelineCache.get(key);
   if (!pending) {
     pending = (async () => {
       const { pipeline: pipeline3 } = await importTransformers(config);
       return pipeline3("text-generation", config.runtimeModel, {
-        device: resolveDevice(config.device),
+        device,
         dtype: config.dtype
       });
     })();
@@ -34732,11 +34729,65 @@ async function importTransformers(config) {
   }
   return transformers;
 }
-function resolveDevice(device) {
-  if (device === "webgpu" || device === "wasm") {
-    return device;
+var resolvedDeviceCache = null;
+async function resolveDevice(device) {
+  if (device === "webgpu" || device === "wasm") return device;
+  if (resolvedDeviceCache) return resolvedDeviceCache;
+  const result = await probeWebGPU();
+  resolvedDeviceCache = result;
+  return result;
+}
+async function probeWebGPU() {
+  try {
+    if (typeof navigator === "undefined" || !("gpu" in navigator)) return "wasm";
+    const gpu = navigator.gpu;
+    const adapter = await gpu.requestAdapter();
+    if (!adapter) return "wasm";
+    const device = await adapter.requestDevice();
+    const module2 = device.createShaderModule({
+      code: `@group(0) @binding(0) var<storage, read_write> out: array<f32>;
+@compute @workgroup_size(1)
+fn main() { out[0] = 42.0; }`
+    });
+    const storageBuffer = device.createBuffer({
+      size: 4,
+      usage: GPUBufferUsage.STORAGE | GPUBufferUsage.COPY_SRC
+    });
+    const readBuffer = device.createBuffer({
+      size: 4,
+      usage: GPUBufferUsage.MAP_READ | GPUBufferUsage.COPY_DST
+    });
+    const bindGroupLayout = device.createBindGroupLayout({
+      entries: [{ binding: 0, visibility: GPUShaderStage.COMPUTE, buffer: { type: "storage" } }]
+    });
+    const pipelineLayout = device.createPipelineLayout({ bindGroupLayouts: [bindGroupLayout] });
+    const pipeline3 = device.createComputePipeline({
+      layout: pipelineLayout,
+      compute: { module: module2, entryPoint: "main" }
+    });
+    const bindGroup = device.createBindGroup({
+      layout: bindGroupLayout,
+      entries: [{ binding: 0, resource: { buffer: storageBuffer } }]
+    });
+    const encoder = device.createCommandEncoder();
+    const pass = encoder.beginComputePass();
+    pass.setPipeline(pipeline3);
+    pass.setBindGroup(0, bindGroup);
+    pass.dispatchWorkgroups(1);
+    pass.end();
+    encoder.copyBufferToBuffer(storageBuffer, 0, readBuffer, 0, 4);
+    device.queue.submit([encoder.finish()]);
+    await readBuffer.mapAsync(GPUMapMode.READ);
+    const data = new Float32Array(readBuffer.getMappedRange());
+    const value = data[0];
+    readBuffer.unmap();
+    storageBuffer.destroy();
+    readBuffer.destroy();
+    device.destroy();
+    return value === 42 ? "webgpu" : "wasm";
+  } catch {
+    return "wasm";
   }
-  return typeof navigator !== "undefined" && "gpu" in navigator ? "webgpu" : "wasm";
 }
 function renderGenerationInput(generator, messages) {
   const applyChatTemplate = generator.tokenizer?.apply_chat_template;