19854902 319854902 319854902 319854902 webgpu交Q流群

我们找个例子看看别人的renderer three.js v152
首先init函数
看见中间有个对象 WebGPURenderPipelines
跟进去看看

这个构造函数里可以看见有这么些数据

get( renderObject ) {

		const device = this.device;
		const cache = this._getCache( renderObject );

		let currentPipeline = cache.currentPipeline;

		if ( this._needsUpdate( renderObject ) ) {

			// release previous cache

			this._releasePipeline( renderObject );

			// get shader

			const nodeBuilder = this.nodes.get( renderObject );

			// programmable stages

			let stageVertex = this.stages.vertex.get( nodeBuilder.vertexShader );

			if ( stageVertex === undefined ) {

				stageVertex = new WebGPUProgrammableStage( device, nodeBuilder.vertexShader, 'vertex' );
				this.stages.vertex.set( nodeBuilder.vertexShader, stageVertex );

			}

			let stageFragment = this.stages.fragment.get( nodeBuilder.fragmentShader );

			if ( stageFragment === undefined ) {

				stageFragment = new WebGPUProgrammableStage( device, nodeBuilder.fragmentShader, 'fragment' );
				this.stages.fragment.set( nodeBuilder.fragmentShader, stageFragment );

			}

			// determine render pipeline

			currentPipeline = this._acquirePipeline( stageVertex, stageFragment, renderObject );
			cache.currentPipeline = currentPipeline;

			// keep track of all used times

			currentPipeline.usedTimes ++;
			stageVertex.usedTimes ++;
			stageFragment.usedTimes ++;

		}

		return currentPipeline;

	}

首先会从cache里找找current pipeline
然后看看是不是需要update，然后需要更新就把原来的pipeline删了，重新构建一个，放进cache
结束
remove dispose比较简单
_acquirePipeline就是把shader啥的作为参数构建新的pipeline
_computeCacheKey就是计算key
大概就这些关键信息
继续看renderer
很多函数名和Webgl一样的， 比如getClearAlpha啥的，这些行为都和以前一样的
接下来 render函数

先把几个特别的东西update一下

然后这个是把通过裁剪判断的物体收集一下，再排序一下

东西收集好了，要开始了，但是有点长，下次继续

19854902 319854902 319854902 319854902 webgpu交Q流群