summaryrefslogtreecommitdiffstats
path: root/dom/webgpu/tests/cts/checkout/src/webgpu/api/operation/resource_init/buffer.spec.ts
diff options
context:
space:
mode:
Diffstat (limited to 'dom/webgpu/tests/cts/checkout/src/webgpu/api/operation/resource_init/buffer.spec.ts')
-rw-r--r--dom/webgpu/tests/cts/checkout/src/webgpu/api/operation/resource_init/buffer.spec.ts899
1 files changed, 899 insertions, 0 deletions
diff --git a/dom/webgpu/tests/cts/checkout/src/webgpu/api/operation/resource_init/buffer.spec.ts b/dom/webgpu/tests/cts/checkout/src/webgpu/api/operation/resource_init/buffer.spec.ts
new file mode 100644
index 0000000000..73c50b8393
--- /dev/null
+++ b/dom/webgpu/tests/cts/checkout/src/webgpu/api/operation/resource_init/buffer.spec.ts
@@ -0,0 +1,899 @@
+import { makeTestGroup } from '../../../../common/framework/test_group.js';
+import { unreachable } from '../../../../common/util/util.js';
+import { GPUConst } from '../../../constants.js';
+import { GPUTest } from '../../../gpu_test.js';
+import { getTextureCopyLayout } from '../../../util/texture/layout.js';
+import { PerTexelComponent } from '../../../util/texture/texel_data.js';
+
+export const description = `
+Test uninitialized buffers are initialized to zero when read
+(or read-written, e.g. with depth write or atomics).
+
+Note that:
+- We don't need 'copy_buffer_to_buffer_copy_destination' here because there has already been an
+ operation test 'command_buffer.copyBufferToBuffer.single' that provides the same functionality.
+`;
+
+const kMapModeOptions = [GPUConst.MapMode.READ, GPUConst.MapMode.WRITE];
+const kBufferUsagesForMappedAtCreationTests = [
+ GPUConst.BufferUsage.COPY_DST | GPUConst.BufferUsage.MAP_READ,
+ GPUConst.BufferUsage.COPY_SRC | GPUConst.BufferUsage.MAP_WRITE,
+ GPUConst.BufferUsage.COPY_SRC,
+];
+
+class F extends GPUTest {
+ GetBufferUsageFromMapMode(mapMode: GPUMapModeFlags): number {
+ switch (mapMode) {
+ case GPUMapMode.READ:
+ return GPUBufferUsage.COPY_DST | GPUBufferUsage.MAP_READ;
+ case GPUMapMode.WRITE:
+ return GPUBufferUsage.COPY_SRC | GPUBufferUsage.MAP_WRITE;
+ default:
+ unreachable();
+ return 0;
+ }
+ }
+
+ async CheckGPUBufferContent(
+ buffer: GPUBuffer,
+ bufferUsage: GPUBufferUsageFlags,
+ expectedData: Uint8Array
+ ): Promise<void> {
+ const mappable = bufferUsage & GPUBufferUsage.MAP_READ;
+ this.expectGPUBufferValuesEqual(buffer, expectedData, 0, { method: mappable ? 'map' : 'copy' });
+ }
+
+ TestBufferZeroInitInBindGroup(
+ computeShaderModule: GPUShaderModule,
+ buffer: GPUBuffer,
+ bufferOffset: number,
+ boundBufferSize: number
+ ): void {
+ const computePipeline = this.device.createComputePipeline({
+ layout: 'auto',
+ compute: {
+ module: computeShaderModule,
+ entryPoint: 'main',
+ },
+ });
+ const outputTexture = this.device.createTexture({
+ format: 'rgba8unorm',
+ size: [1, 1, 1],
+ usage: GPUTextureUsage.COPY_SRC | GPUTextureUsage.STORAGE_BINDING,
+ });
+ this.trackForCleanup(outputTexture);
+ const bindGroup = this.device.createBindGroup({
+ layout: computePipeline.getBindGroupLayout(0),
+ entries: [
+ {
+ binding: 0,
+ resource: {
+ buffer,
+ offset: bufferOffset,
+ size: boundBufferSize,
+ },
+ },
+ {
+ binding: 1,
+ resource: outputTexture.createView(),
+ },
+ ],
+ });
+
+ const encoder = this.device.createCommandEncoder();
+ const computePass = encoder.beginComputePass();
+ computePass.setBindGroup(0, bindGroup);
+ computePass.setPipeline(computePipeline);
+ computePass.dispatchWorkgroups(1);
+ computePass.end();
+ this.queue.submit([encoder.finish()]);
+
+ this.CheckBufferAndOutputTexture(buffer, boundBufferSize + bufferOffset, outputTexture);
+ }
+
+ CreateRenderPipelineForTest(
+ vertexShaderModule: GPUShaderModule,
+ testVertexBuffer: boolean
+ ): GPURenderPipeline {
+ const renderPipelineDescriptor: GPURenderPipelineDescriptor = {
+ layout: 'auto',
+ vertex: {
+ module: vertexShaderModule,
+ entryPoint: 'main',
+ },
+ fragment: {
+ module: this.device.createShaderModule({
+ code: `
+ @fragment
+ fn main(@location(0) i_color : vec4<f32>) -> @location(0) vec4<f32> {
+ return i_color;
+ }`,
+ }),
+ entryPoint: 'main',
+ targets: [{ format: 'rgba8unorm' }],
+ },
+ primitive: {
+ topology: 'point-list',
+ },
+ };
+ if (testVertexBuffer) {
+ renderPipelineDescriptor.vertex.buffers = [
+ {
+ arrayStride: 16,
+ attributes: [{ format: 'float32x4', offset: 0, shaderLocation: 0 }],
+ },
+ ];
+ }
+
+ return this.device.createRenderPipeline(renderPipelineDescriptor);
+ }
+
+ RecordInitializeTextureColor(
+ encoder: GPUCommandEncoder,
+ texture: GPUTexture,
+ color: GPUColor
+ ): void {
+ const renderPass = encoder.beginRenderPass({
+ colorAttachments: [
+ {
+ view: texture.createView(),
+ clearValue: color,
+ loadOp: 'clear',
+ storeOp: 'store',
+ },
+ ],
+ });
+ renderPass.end();
+ }
+
+ CheckBufferAndOutputTexture(
+ buffer: GPUBuffer,
+ bufferSize: number,
+ outputTexture: GPUTexture,
+ outputTextureSize: [number, number, number] = [1, 1, 1],
+ outputTextureColor: PerTexelComponent<number> = { R: 0.0, G: 1.0, B: 0.0, A: 1.0 }
+ ): void {
+ this.expectSingleColor(outputTexture, 'rgba8unorm', {
+ size: outputTextureSize,
+ exp: outputTextureColor,
+ });
+
+ const expectedBufferData = new Uint8Array(bufferSize);
+ this.expectGPUBufferValuesEqual(buffer, expectedBufferData);
+ }
+}
+
+export const g = makeTestGroup(F);
+
+g.test('partial_write_buffer')
+ .desc(
+ `Verify when we upload data to a part of a buffer with writeBuffer() just after the creation of
+the buffer, the remaining part of that buffer will be initialized to 0.`
+ )
+ .paramsSubcasesOnly(u => u.combine('offset', [0, 8, -12]))
+ .fn(async t => {
+ const { offset } = t.params;
+ const bufferSize = 32;
+ const appliedOffset = offset >= 0 ? offset : bufferSize + offset;
+
+ const buffer = t.device.createBuffer({
+ size: bufferSize,
+ usage: GPUBufferUsage.COPY_SRC | GPUBufferUsage.COPY_DST,
+ });
+ t.trackForCleanup(buffer);
+
+ const copySize = 12;
+ const writeData = new Uint8Array(copySize);
+ const expectedData = new Uint8Array(bufferSize);
+ for (let i = 0; i < copySize; ++i) {
+ expectedData[appliedOffset + i] = writeData[i] = i + 1;
+ }
+ t.queue.writeBuffer(buffer, appliedOffset, writeData, 0);
+
+ t.expectGPUBufferValuesEqual(buffer, expectedData);
+ });
+
+g.test('map_whole_buffer')
+ .desc(
+ `Verify when we map the whole range of a mappable GPUBuffer to a typed array buffer just after
+creating the GPUBuffer, the contents of both the typed array buffer and the GPUBuffer itself
+have already been initialized to 0.`
+ )
+ .params(u => u.combine('mapMode', kMapModeOptions))
+ .fn(async t => {
+ const { mapMode } = t.params;
+
+ const bufferSize = 32;
+ const bufferUsage = t.GetBufferUsageFromMapMode(mapMode);
+ const buffer = t.device.createBuffer({
+ size: bufferSize,
+ usage: bufferUsage,
+ });
+ t.trackForCleanup(buffer);
+
+ await buffer.mapAsync(mapMode);
+ const readData = new Uint8Array(buffer.getMappedRange());
+ for (let i = 0; i < bufferSize; ++i) {
+ t.expect(readData[i] === 0);
+ }
+ buffer.unmap();
+
+ const expectedData = new Uint8Array(bufferSize);
+ await t.CheckGPUBufferContent(buffer, bufferUsage, expectedData);
+ });
+
+g.test('map_partial_buffer')
+ .desc(
+ `Verify when we map a subrange of a mappable GPUBuffer to a typed array buffer just after the
+creation of the GPUBuffer, the contents of both the typed array buffer and the GPUBuffer have
+already been initialized to 0.`
+ )
+ .params(u => u.combine('mapMode', kMapModeOptions).beginSubcases().combine('offset', [0, 8, -16]))
+ .fn(async t => {
+ const { mapMode, offset } = t.params;
+ const bufferSize = 32;
+ const appliedOffset = offset >= 0 ? offset : bufferSize + offset;
+
+ const bufferUsage = t.GetBufferUsageFromMapMode(mapMode);
+ const buffer = t.device.createBuffer({
+ size: bufferSize,
+ usage: bufferUsage,
+ });
+ t.trackForCleanup(buffer);
+
+ const expectedData = new Uint8Array(bufferSize);
+ {
+ const mapSize = 16;
+ await buffer.mapAsync(mapMode, appliedOffset, mapSize);
+ const mappedData = new Uint8Array(buffer.getMappedRange(appliedOffset, mapSize));
+ for (let i = 0; i < mapSize; ++i) {
+ t.expect(mappedData[i] === 0);
+ if (mapMode === GPUMapMode.WRITE) {
+ mappedData[i] = expectedData[appliedOffset + i] = i + 1;
+ }
+ }
+ buffer.unmap();
+ }
+
+ await t.CheckGPUBufferContent(buffer, bufferUsage, expectedData);
+ });
+
+g.test('mapped_at_creation_whole_buffer')
+ .desc(
+ `Verify when we call getMappedRange() at the whole range of a GPUBuffer created with
+mappedAtCreation === true just after its creation, the contents of both the returned typed
+array buffer of getMappedRange() and the GPUBuffer itself have all been initialized to 0.`
+ )
+ .params(u => u.combine('bufferUsage', kBufferUsagesForMappedAtCreationTests))
+ .fn(async t => {
+ const { bufferUsage } = t.params;
+
+ const bufferSize = 32;
+ const buffer = t.device.createBuffer({
+ mappedAtCreation: true,
+ size: bufferSize,
+ usage: bufferUsage,
+ });
+ t.trackForCleanup(buffer);
+
+ const mapped = new Uint8Array(buffer.getMappedRange());
+ for (let i = 0; i < bufferSize; ++i) {
+ t.expect(mapped[i] === 0);
+ }
+ buffer.unmap();
+
+ const expectedData = new Uint8Array(bufferSize);
+ await t.CheckGPUBufferContent(buffer, bufferUsage, expectedData);
+ });
+
+g.test('mapped_at_creation_partial_buffer')
+ .desc(
+ `Verify when we call getMappedRange() at a subrange of a GPUBuffer created with
+mappedAtCreation === true just after its creation, the contents of both the returned typed
+array buffer of getMappedRange() and the GPUBuffer itself have all been initialized to 0.`
+ )
+ .params(u =>
+ u
+ .combine('bufferUsage', kBufferUsagesForMappedAtCreationTests)
+ .beginSubcases()
+ .combine('offset', [0, 8, -16])
+ )
+ .fn(async t => {
+ const { bufferUsage, offset } = t.params;
+ const bufferSize = 32;
+ const appliedOffset = offset >= 0 ? offset : bufferSize + offset;
+
+ const buffer = t.device.createBuffer({
+ mappedAtCreation: true,
+ size: bufferSize,
+ usage: bufferUsage,
+ });
+ t.trackForCleanup(buffer);
+
+ const expectedData = new Uint8Array(bufferSize);
+ {
+ const mappedSize = 12;
+ const mapped = new Uint8Array(buffer.getMappedRange(appliedOffset, mappedSize));
+ for (let i = 0; i < mappedSize; ++i) {
+ t.expect(mapped[i] === 0);
+ if (!(bufferUsage & GPUBufferUsage.MAP_READ)) {
+ mapped[i] = expectedData[appliedOffset + i] = i + 1;
+ }
+ }
+ buffer.unmap();
+ }
+
+ await t.CheckGPUBufferContent(buffer, bufferUsage, expectedData);
+ });
+
+g.test('copy_buffer_to_buffer_copy_source')
+ .desc(
+ `Verify when the first usage of a GPUBuffer is being used as the source buffer of
+CopyBufferToBuffer(), the contents of the GPUBuffer have already been initialized to 0.`
+ )
+ .fn(async t => {
+ const bufferSize = 32;
+ const bufferUsage = GPUBufferUsage.COPY_SRC;
+ const buffer = t.device.createBuffer({
+ size: bufferSize,
+ usage: bufferUsage,
+ });
+ t.trackForCleanup(buffer);
+
+ const expectedData = new Uint8Array(bufferSize);
+ // copyBufferToBuffer() is called inside t.CheckGPUBufferContent().
+ await t.CheckGPUBufferContent(buffer, bufferUsage, expectedData);
+ });
+
+g.test('copy_buffer_to_texture')
+ .desc(
+ `Verify when the first usage of a GPUBuffer is being used as the source buffer of
+CopyBufferToTexture(), the contents of the GPUBuffer have already been initialized to 0.`
+ )
+ .paramsSubcasesOnly(u => u.combine('bufferOffset', [0, 8]))
+ .fn(async t => {
+ const { bufferOffset } = t.params;
+ const textureSize: [number, number, number] = [8, 8, 1];
+ const dstTextureFormat = 'rgba8unorm';
+
+ const dstTexture = t.device.createTexture({
+ size: textureSize,
+ format: dstTextureFormat,
+ usage: GPUTextureUsage.COPY_SRC | GPUTextureUsage.COPY_DST,
+ });
+ t.trackForCleanup(dstTexture);
+ const layout = getTextureCopyLayout(dstTextureFormat, '2d', textureSize);
+ const srcBufferSize = layout.byteLength + bufferOffset;
+ const srcBufferUsage = GPUBufferUsage.COPY_SRC;
+ const srcBuffer = t.device.createBuffer({
+ size: srcBufferSize,
+ usage: srcBufferUsage,
+ });
+ t.trackForCleanup(srcBuffer);
+
+ const encoder = t.device.createCommandEncoder();
+ encoder.copyBufferToTexture(
+ {
+ buffer: srcBuffer,
+ offset: bufferOffset,
+ bytesPerRow: layout.bytesPerRow,
+ rowsPerImage: layout.rowsPerImage,
+ },
+ { texture: dstTexture },
+ textureSize
+ );
+ t.queue.submit([encoder.finish()]);
+
+ t.CheckBufferAndOutputTexture(srcBuffer, srcBufferSize, dstTexture, textureSize, {
+ R: 0.0,
+ G: 0.0,
+ B: 0.0,
+ A: 0.0,
+ });
+ });
+
+g.test('resolve_query_set_to_partial_buffer')
+ .desc(
+ `Verify when we resolve a query set into a GPUBuffer just after creating that GPUBuffer, the
+remaining part of it will be initialized to 0.`
+ )
+ .paramsSubcasesOnly(u => u.combine('bufferOffset', [0, 256]))
+ .fn(async t => {
+ const { bufferOffset } = t.params;
+ const bufferSize = bufferOffset + 8;
+ const bufferUsage = GPUBufferUsage.COPY_SRC | GPUBufferUsage.QUERY_RESOLVE;
+ const dstBuffer = t.device.createBuffer({
+ size: bufferSize,
+ usage: bufferUsage,
+ });
+ t.trackForCleanup(dstBuffer);
+
+ const querySet = t.device.createQuerySet({ type: 'occlusion', count: 1 });
+ const encoder = t.device.createCommandEncoder();
+ encoder.resolveQuerySet(querySet, 0, 1, dstBuffer, bufferOffset);
+ t.queue.submit([encoder.finish()]);
+
+ const expectedBufferData = new Uint8Array(bufferSize);
+ await t.CheckGPUBufferContent(dstBuffer, bufferUsage, expectedBufferData);
+ });
+
+g.test('copy_texture_to_partial_buffer')
+ .desc(
+ `Verify when we copy from a GPUTexture into a GPUBuffer just after creating that GPUBuffer, the
+remaining part of it will be initialized to 0.`
+ )
+ .paramsSubcasesOnly(u =>
+ u
+ .combine('bufferOffset', [0, 8, -16])
+ .combine('arrayLayerCount', [1, 3])
+ .combine('copyMipLevel', [0, 2])
+ .combine('rowsPerImage', [16, 20])
+ .filter(t => {
+ // We don't need to test the copies that will cover the whole GPUBuffer.
+ return !(t.bufferOffset === 0 && t.rowsPerImage === 16);
+ })
+ )
+ .fn(async t => {
+ const { bufferOffset, arrayLayerCount, copyMipLevel, rowsPerImage } = t.params;
+ const srcTextureFormat = 'r8uint';
+ const textureSize = [32, 16, arrayLayerCount] as const;
+
+ const srcTexture = t.device.createTexture({
+ format: srcTextureFormat,
+ usage: GPUTextureUsage.COPY_SRC | GPUTextureUsage.RENDER_ATTACHMENT,
+ size: textureSize,
+ mipLevelCount: copyMipLevel + 1,
+ });
+ t.trackForCleanup(srcTexture);
+
+ const bytesPerRow = 256;
+ const layout = getTextureCopyLayout(srcTextureFormat, '2d', textureSize, {
+ mipLevel: copyMipLevel,
+ bytesPerRow,
+ rowsPerImage,
+ });
+
+ const dstBufferSize = layout.byteLength + Math.abs(bufferOffset);
+ const dstBuffer = t.device.createBuffer({
+ size: dstBufferSize,
+ usage: GPUBufferUsage.COPY_SRC | GPUBufferUsage.COPY_DST,
+ });
+ t.trackForCleanup(dstBuffer);
+
+ const encoder = t.device.createCommandEncoder();
+
+ // Initialize srcTexture
+ for (let layer = 0; layer < arrayLayerCount; ++layer) {
+ const renderPass = encoder.beginRenderPass({
+ colorAttachments: [
+ {
+ view: srcTexture.createView({
+ baseArrayLayer: layer,
+ arrayLayerCount: 1,
+ baseMipLevel: copyMipLevel,
+ }),
+ clearValue: { r: layer + 1, g: 0, b: 0, a: 0 },
+ loadOp: 'clear',
+ storeOp: 'store',
+ },
+ ],
+ });
+ renderPass.end();
+ }
+
+ // Do texture-to-buffer copy
+ const appliedOffset = Math.max(bufferOffset, 0);
+ encoder.copyTextureToBuffer(
+ { texture: srcTexture, mipLevel: copyMipLevel },
+ { buffer: dstBuffer, offset: appliedOffset, bytesPerRow, rowsPerImage },
+ layout.mipSize
+ );
+ t.queue.submit([encoder.finish()]);
+
+ // Check if the contents of the destination buffer are what we expect.
+ const expectedData = new Uint8Array(dstBufferSize);
+ for (let layer = 0; layer < arrayLayerCount; ++layer) {
+ for (let y = 0; y < layout.mipSize[1]; ++y) {
+ for (let x = 0; x < layout.mipSize[0]; ++x) {
+ expectedData[appliedOffset + layer * bytesPerRow * rowsPerImage + y * bytesPerRow + x] =
+ layer + 1;
+ }
+ }
+ }
+ t.expectGPUBufferValuesEqual(dstBuffer, expectedData);
+ });
+
+g.test('uniform_buffer')
+ .desc(
+ `Verify when we use a GPUBuffer as a uniform buffer just after the creation of that GPUBuffer,
+ all the contents in that GPUBuffer have been initialized to 0.`
+ )
+ .paramsSubcasesOnly(u => u.combine('bufferOffset', [0, 256]))
+ .fn(async t => {
+ const { bufferOffset } = t.params;
+
+ const boundBufferSize = 16;
+ const buffer = t.device.createBuffer({
+ size: bufferOffset + boundBufferSize,
+ usage: GPUBufferUsage.COPY_SRC | GPUBufferUsage.UNIFORM,
+ });
+ t.trackForCleanup(buffer);
+
+ const computeShaderModule = t.device.createShaderModule({
+ code: `
+ struct UBO {
+ value : vec4<u32>
+ };
+ @group(0) @binding(0) var<uniform> ubo : UBO;
+ @group(0) @binding(1) var outImage : texture_storage_2d<rgba8unorm, write>;
+
+ @compute @workgroup_size(1) fn main() {
+ if (all(ubo.value == vec4<u32>(0u, 0u, 0u, 0u))) {
+ textureStore(outImage, vec2<i32>(0, 0), vec4<f32>(0.0, 1.0, 0.0, 1.0));
+ } else {
+ textureStore(outImage, vec2<i32>(0, 0), vec4<f32>(1.0, 0.0, 0.0, 1.0));
+ }
+ }`,
+ });
+
+ // Verify the whole range of the buffer has been initialized to 0 in a compute shader.
+ t.TestBufferZeroInitInBindGroup(computeShaderModule, buffer, bufferOffset, boundBufferSize);
+ });
+
+g.test('readonly_storage_buffer')
+ .desc(
+ `Verify when we use a GPUBuffer as a read-only storage buffer just after the creation of that
+ GPUBuffer, all the contents in that GPUBuffer have been initialized to 0.`
+ )
+ .paramsSubcasesOnly(u => u.combine('bufferOffset', [0, 256]))
+ .fn(async t => {
+ const { bufferOffset } = t.params;
+ const boundBufferSize = 16;
+ const buffer = t.device.createBuffer({
+ size: bufferOffset + boundBufferSize,
+ usage: GPUBufferUsage.COPY_SRC | GPUBufferUsage.STORAGE,
+ });
+ t.trackForCleanup(buffer);
+
+ const computeShaderModule = t.device.createShaderModule({
+ code: `
+ struct SSBO {
+ value : vec4<u32>
+ };
+ @group(0) @binding(0) var<storage, read> ssbo : SSBO;
+ @group(0) @binding(1) var outImage : texture_storage_2d<rgba8unorm, write>;
+
+ @compute @workgroup_size(1) fn main() {
+ if (all(ssbo.value == vec4<u32>(0u, 0u, 0u, 0u))) {
+ textureStore(outImage, vec2<i32>(0, 0), vec4<f32>(0.0, 1.0, 0.0, 1.0));
+ } else {
+ textureStore(outImage, vec2<i32>(0, 0), vec4<f32>(1.0, 0.0, 0.0, 1.0));
+ }
+ }`,
+ });
+
+ // Verify the whole range of the buffer has been initialized to 0 in a compute shader.
+ t.TestBufferZeroInitInBindGroup(computeShaderModule, buffer, bufferOffset, boundBufferSize);
+ });
+
+g.test('storage_buffer')
+ .desc(
+ `Verify when we use a GPUBuffer as a storage buffer just after the creation of that
+ GPUBuffer, all the contents in that GPUBuffer have been initialized to 0.`
+ )
+ .paramsSubcasesOnly(u => u.combine('bufferOffset', [0, 256]))
+ .fn(async t => {
+ const { bufferOffset } = t.params;
+ const boundBufferSize = 16;
+ const buffer = t.device.createBuffer({
+ size: bufferOffset + boundBufferSize,
+ usage: GPUBufferUsage.COPY_SRC | GPUBufferUsage.STORAGE,
+ });
+ t.trackForCleanup(buffer);
+
+ const computeShaderModule = t.device.createShaderModule({
+ code: `
+ struct SSBO {
+ value : vec4<u32>
+ };
+ @group(0) @binding(0) var<storage, read_write> ssbo : SSBO;
+ @group(0) @binding(1) var outImage : texture_storage_2d<rgba8unorm, write>;
+
+ @compute @workgroup_size(1) fn main() {
+ if (all(ssbo.value == vec4<u32>(0u, 0u, 0u, 0u))) {
+ textureStore(outImage, vec2<i32>(0, 0), vec4<f32>(0.0, 1.0, 0.0, 1.0));
+ } else {
+ textureStore(outImage, vec2<i32>(0, 0), vec4<f32>(1.0, 0.0, 0.0, 1.0));
+ }
+ }`,
+ });
+
+ // Verify the whole range of the buffer has been initialized to 0 in a compute shader.
+ t.TestBufferZeroInitInBindGroup(computeShaderModule, buffer, bufferOffset, boundBufferSize);
+ });
+
+g.test('vertex_buffer')
+ .desc(
+ `Verify when we use a GPUBuffer as a vertex buffer just after the creation of that
+ GPUBuffer, all the contents in that GPUBuffer have been initialized to 0.`
+ )
+ .paramsSubcasesOnly(u => u.combine('bufferOffset', [0, 16]))
+ .fn(async t => {
+ const { bufferOffset } = t.params;
+
+ const renderPipeline = t.CreateRenderPipelineForTest(
+ t.device.createShaderModule({
+ code: `
+ struct VertexOut {
+ @location(0) color : vec4<f32>,
+ @builtin(position) position : vec4<f32>,
+ };
+
+ @vertex fn main(@location(0) pos : vec4<f32>) -> VertexOut {
+ var output : VertexOut;
+ if (all(pos == vec4<f32>(0.0, 0.0, 0.0, 0.0))) {
+ output.color = vec4<f32>(0.0, 1.0, 0.0, 1.0);
+ } else {
+ output.color = vec4<f32>(1.0, 0.0, 0.0, 1.0);
+ }
+ output.position = vec4<f32>(0.0, 0.0, 0.0, 1.0);
+ return output;
+ }`,
+ }),
+ true
+ );
+
+ const bufferSize = 16 + bufferOffset;
+ const vertexBuffer = t.device.createBuffer({
+ size: bufferSize,
+ usage: GPUBufferUsage.VERTEX | GPUBufferUsage.COPY_SRC,
+ });
+ t.trackForCleanup(vertexBuffer);
+
+ const outputTexture = t.device.createTexture({
+ format: 'rgba8unorm',
+ size: [1, 1, 1],
+ usage: GPUTextureUsage.COPY_SRC | GPUTextureUsage.RENDER_ATTACHMENT,
+ });
+ t.trackForCleanup(outputTexture);
+
+ const encoder = t.device.createCommandEncoder();
+ const renderPass = encoder.beginRenderPass({
+ colorAttachments: [
+ {
+ view: outputTexture.createView(),
+ clearValue: { r: 0.0, g: 0.0, b: 0.0, a: 0.0 },
+ loadOp: 'clear',
+ storeOp: 'store',
+ },
+ ],
+ });
+ renderPass.setVertexBuffer(0, vertexBuffer, bufferOffset);
+ renderPass.setPipeline(renderPipeline);
+ renderPass.draw(1);
+ renderPass.end();
+ t.queue.submit([encoder.finish()]);
+
+ t.CheckBufferAndOutputTexture(vertexBuffer, bufferSize, outputTexture);
+ });
+
+g.test('index_buffer')
+ .desc(
+ `Verify when we use a GPUBuffer as an index buffer just after the creation of that
+GPUBuffer, all the contents in that GPUBuffer have been initialized to 0.`
+ )
+ .paramsSubcasesOnly(u => u.combine('bufferOffset', [0, 16]))
+ .fn(async t => {
+ const { bufferOffset } = t.params;
+
+ const renderPipeline = t.CreateRenderPipelineForTest(
+ t.device.createShaderModule({
+ code: `
+ struct VertexOut {
+ @location(0) color : vec4<f32>,
+ @builtin(position) position : vec4<f32>,
+ };
+
+ @vertex
+ fn main(@builtin(vertex_index) VertexIndex : u32) -> VertexOut {
+ var output : VertexOut;
+ if (VertexIndex == 0u) {
+ output.color = vec4<f32>(0.0, 1.0, 0.0, 1.0);
+ } else {
+ output.color = vec4<f32>(1.0, 0.0, 0.0, 1.0);
+ }
+ output.position = vec4<f32>(0.0, 0.0, 0.0, 1.0);
+ return output;
+ }`,
+ }),
+ false
+ );
+
+ // The size of GPUBuffer must be at least 4.
+ const bufferSize = 4 + bufferOffset;
+ const indexBuffer = t.device.createBuffer({
+ size: bufferSize,
+ usage: GPUBufferUsage.INDEX | GPUBufferUsage.COPY_SRC,
+ });
+ t.trackForCleanup(indexBuffer);
+
+ const outputTexture = t.device.createTexture({
+ format: 'rgba8unorm',
+ size: [1, 1, 1],
+ usage: GPUTextureUsage.COPY_SRC | GPUTextureUsage.RENDER_ATTACHMENT,
+ });
+ t.trackForCleanup(outputTexture);
+
+ const encoder = t.device.createCommandEncoder();
+ const renderPass = encoder.beginRenderPass({
+ colorAttachments: [
+ {
+ view: outputTexture.createView(),
+ clearValue: { r: 0.0, g: 0.0, b: 0.0, a: 0.0 },
+ loadOp: 'clear',
+ storeOp: 'store',
+ },
+ ],
+ });
+ renderPass.setPipeline(renderPipeline);
+ renderPass.setIndexBuffer(indexBuffer, 'uint16', bufferOffset, 4);
+ renderPass.drawIndexed(1);
+ renderPass.end();
+ t.queue.submit([encoder.finish()]);
+
+ t.CheckBufferAndOutputTexture(indexBuffer, bufferSize, outputTexture);
+ });
+
+g.test('indirect_buffer_for_draw_indirect')
+ .desc(
+ `Verify when we use a GPUBuffer as an indirect buffer for drawIndirect() or
+drawIndexedIndirect() just after the creation of that GPUBuffer, all the contents in that GPUBuffer
+have been initialized to 0.`
+ )
+ .params(u =>
+ u.combine('test_indexed_draw', [true, false]).beginSubcases().combine('bufferOffset', [0, 16])
+ )
+ .fn(async t => {
+ const { test_indexed_draw, bufferOffset } = t.params;
+
+ const renderPipeline = t.CreateRenderPipelineForTest(
+ t.device.createShaderModule({
+ code: `
+ struct VertexOut {
+ @location(0) color : vec4<f32>,
+ @builtin(position) position : vec4<f32>,
+ };
+
+ @vertex fn main() -> VertexOut {
+ var output : VertexOut;
+ output.color = vec4<f32>(1.0, 0.0, 0.0, 1.0);
+ output.position = vec4<f32>(0.0, 0.0, 0.0, 1.0);
+ return output;
+ }`,
+ }),
+ false
+ );
+
+ const kDrawIndirectParametersSize = 16;
+ const kDrawIndexedIndirectParametersSize = 20;
+ const bufferSize =
+ Math.max(kDrawIndirectParametersSize, kDrawIndexedIndirectParametersSize) + bufferOffset;
+ const indirectBuffer = t.device.createBuffer({
+ size: bufferSize,
+ usage: GPUBufferUsage.COPY_SRC | GPUBufferUsage.INDIRECT,
+ });
+ t.trackForCleanup(indirectBuffer);
+
+ const outputTexture = t.device.createTexture({
+ format: 'rgba8unorm',
+ size: [1, 1, 1],
+ usage: GPUTextureUsage.COPY_SRC | GPUTextureUsage.RENDER_ATTACHMENT,
+ });
+ t.trackForCleanup(outputTexture);
+
+ // Initialize outputTexture to green.
+ const encoder = t.device.createCommandEncoder();
+ t.RecordInitializeTextureColor(encoder, outputTexture, { r: 0.0, g: 1.0, b: 0.0, a: 1.0 });
+
+ const renderPass = encoder.beginRenderPass({
+ colorAttachments: [
+ {
+ view: outputTexture.createView(),
+ loadOp: 'load',
+ storeOp: 'store',
+ },
+ ],
+ });
+ renderPass.setPipeline(renderPipeline);
+
+ let indexBuffer = undefined;
+ if (test_indexed_draw) {
+ indexBuffer = t.device.createBuffer({
+ size: 4,
+ usage: GPUBufferUsage.INDEX,
+ });
+ t.trackForCleanup(indexBuffer);
+ renderPass.setIndexBuffer(indexBuffer, 'uint16');
+ renderPass.drawIndexedIndirect(indirectBuffer, bufferOffset);
+ } else {
+ renderPass.drawIndirect(indirectBuffer, bufferOffset);
+ }
+
+ renderPass.end();
+ t.queue.submit([encoder.finish()]);
+
+ // The indirect buffer should be lazily cleared to 0, so we actually draw nothing and the color
+ // attachment will keep its original color (green) after we end the render pass.
+ t.CheckBufferAndOutputTexture(indirectBuffer, bufferSize, outputTexture);
+ });
+
+g.test('indirect_buffer_for_dispatch_indirect')
+ .desc(
+ `Verify when we use a GPUBuffer as an indirect buffer for dispatchWorkgroupsIndirect() just
+ after the creation of that GPUBuffer, all the contents in that GPUBuffer have been initialized
+ to 0.`
+ )
+ .paramsSubcasesOnly(u => u.combine('bufferOffset', [0, 16]))
+ .fn(async t => {
+ const { bufferOffset } = t.params;
+
+ const computePipeline = t.device.createComputePipeline({
+ layout: 'auto',
+ compute: {
+ module: t.device.createShaderModule({
+ code: `
+ @group(0) @binding(0) var outImage : texture_storage_2d<rgba8unorm, write>;
+
+ @compute @workgroup_size(1) fn main() {
+ textureStore(outImage, vec2<i32>(0, 0), vec4<f32>(1.0, 0.0, 0.0, 1.0));
+ }`,
+ }),
+ entryPoint: 'main',
+ },
+ });
+
+ const kDispatchIndirectParametersSize = 12;
+ const bufferSize = kDispatchIndirectParametersSize + bufferOffset;
+ const indirectBuffer = t.device.createBuffer({
+ size: bufferSize,
+ usage: GPUBufferUsage.COPY_SRC | GPUBufferUsage.INDIRECT,
+ });
+ t.trackForCleanup(indirectBuffer);
+
+ const outputTexture = t.device.createTexture({
+ format: 'rgba8unorm',
+ size: [1, 1, 1],
+ usage:
+ GPUTextureUsage.COPY_SRC |
+ GPUTextureUsage.RENDER_ATTACHMENT |
+ GPUTextureUsage.STORAGE_BINDING,
+ });
+ t.trackForCleanup(outputTexture);
+
+ // Initialize outputTexture to green.
+ const encoder = t.device.createCommandEncoder();
+ t.RecordInitializeTextureColor(encoder, outputTexture, { r: 0.0, g: 1.0, b: 0.0, a: 1.0 });
+
+ const bindGroup = t.device.createBindGroup({
+ layout: computePipeline.getBindGroupLayout(0),
+ entries: [
+ {
+ binding: 0,
+ resource: outputTexture.createView(),
+ },
+ ],
+ });
+
+ // The indirect buffer should be lazily cleared to 0, so we actually don't execute the compute
+ // shader and the output texture should keep its original color (green).
+ const computePass = encoder.beginComputePass();
+ computePass.setBindGroup(0, bindGroup);
+ computePass.setPipeline(computePipeline);
+ computePass.dispatchWorkgroupsIndirect(indirectBuffer, bufferOffset);
+ computePass.end();
+ t.queue.submit([encoder.finish()]);
+
+ // The indirect buffer should be lazily cleared to 0, so we actually draw nothing and the color
+ // attachment will keep its original color (green) after we end the compute pass.
+ t.CheckBufferAndOutputTexture(indirectBuffer, bufferSize, outputTexture);
+ });