blob: 69c2005a22e279d3150a80b0bc92666ccd1a8c4c [file] [log] [blame] [edit]
// Copyright 2025 The Dawn & Tint Authors
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are met:
//
// 1. Redistributions of source code must retain the above copyright notice, this
// list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other materials provided with the distribution.
//
// 3. Neither the name of the copyright holder nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
// DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
// FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
// DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
// SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
// CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
// OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#include <vector>
#include "dawn/common/GPUInfo.h"
#include "dawn/tests/DawnTest.h"
// The purpose of these tests is to prevent regressions of features and limits on architectures that
// are known to have these supported capabilities.
//
// Why? Because of the nature of feature/limit support in WebGPU we can easily accidentally drop
// (regress) optional capabilities for specific devices and dawn/CTS testing machinery and even
// external webgpu clients will respond gracefully. We (dawn) never want to silently regress
// features on critical platforms. If we are regressing it should be a deliberate decision which may
// involve changes to the tests below.
//
// How? Currently it is difficult to know our features/limit support for all hardware (could even be
// driver version dependent). So here we conservatively assert that specific architectures have
// specific features/limits.
//
// When? This file may need to be updated with new conservative tests if the CQ changes hardware or
// if we have deliberately chosen to regress specific features/limits on some devices.
//
// Not an official dawn/webgpu resource but this page can provide insights into features/limits:
// https://web3dsurvey.com/webgpu
namespace dawn {
namespace {
using FeatureArchInfoTestBase = DawnTestWithParams<>;
class FeatureArchInfoTest_MaxLimits : public FeatureArchInfoTestBase {
protected:
void GetRequiredLimits(const dawn::utils::ComboLimits& supported,
dawn::utils::ComboLimits& required) override {
supported.UnlinkedCopyTo(&required);
}
};
TEST_P(FeatureArchInfoTest_MaxLimits, SubgroupsSupported) {
const bool subgroupSupportExpected =
// All Apple Silicon
gpu_info::IsApple(GetParam().adapterProperties.vendorID) ||
// All Intel >Gen9 on Metal/Vulkan, and D3D12+DXC
((IsMetal() || IsVulkan() || IsDXC()) &&
gpu_info::IsIntel(GetParam().adapterProperties.vendorID) &&
!gpu_info::IsIntelGen9(GetParam().adapterProperties.vendorID,
GetParam().adapterProperties.deviceID));
DAWN_TEST_UNSUPPORTED_IF(!subgroupSupportExpected);
EXPECT_TRUE(this->SupportsFeatures({wgpu::FeatureName::Subgroups}));
}
TEST_P(FeatureArchInfoTest_MaxLimits, ShaderF16Supported) {
// All apple silicon devices support f16
const bool shaderF16SupportExpected = gpu_info::IsApple(GetParam().adapterProperties.vendorID);
DAWN_TEST_UNSUPPORTED_IF(!shaderF16SupportExpected);
EXPECT_TRUE(this->SupportsFeatures({wgpu::FeatureName::ShaderF16}));
}
TEST_P(FeatureArchInfoTest_MaxLimits, WorkgroupSizeMin1024Expected) {
// All apple silicon devices are known to have 1k workgroup size limit.
// This code will need to be changed if a new apple silicon device is released with a different
// limit. Nvidia also support >= 1k workgroups.
const bool isWorkgroupSizeGT1k = gpu_info::IsApple(GetParam().adapterProperties.vendorID) ||
gpu_info::IsNvidia(GetParam().adapterProperties.vendorID);
DAWN_TEST_UNSUPPORTED_IF(!isWorkgroupSizeGT1k);
EXPECT_GE(GetSupportedLimits().maxComputeInvocationsPerWorkgroup, 1024u);
}
class FeatureArchInfoTest_TieredMaxLimits : public FeatureArchInfoTestBase {
protected:
bool GetRequireUseTieredLimits() override { return true; }
void GetRequiredLimits(const dawn::utils::ComboLimits& supported,
dawn::utils::ComboLimits& required) override {
supported.UnlinkedCopyTo(&required);
}
};
TEST_P(FeatureArchInfoTest_TieredMaxLimits, D3DHighMaxVertexAttributes) {
const bool isWindowsHighEnd =
gpu_info::IsNvidia(GetParam().adapterProperties.vendorID) && (IsD3D11() || IsD3D12());
DAWN_TEST_UNSUPPORTED_IF(!isWindowsHighEnd);
// High-end windows desktop GPU should report at least 30 even when tiered is enabled
// See crbug.com/430371785
EXPECT_GE(GetSupportedLimits().maxVertexAttributes, 30u);
}
TEST_P(FeatureArchInfoTest_TieredMaxLimits, AppleMaxTextureDimension2D) {
const bool isAppleSilicon = gpu_info::IsApple(GetParam().adapterProperties.vendorID);
DAWN_TEST_UNSUPPORTED_IF(!isAppleSilicon);
// Apple silicon should report > 16k. Most modern devices should support > 16k. Update this when
// appropriate.
EXPECT_GE(GetSupportedLimits().maxTextureDimension2D, 16384u);
}
TEST_P(FeatureArchInfoTest_TieredMaxLimits, MaxUniformBufferBindingSize) {
// Swiftshader will return a lower limit than any modern device on CQ.
DAWN_TEST_UNSUPPORTED_IF(IsSwiftshader() || IsANGLESwiftShader());
EXPECT_EQ(GetSupportedLimits().maxUniformBufferBindingSize, 65536u);
}
TEST_P(FeatureArchInfoTest_MaxLimits, MinUniformBufferOffsetAlignment) {
// All devices on CQ report 256. A value of 32 is commonplace on iOS (note the less than).
EXPECT_LE(GetSupportedLimits().minUniformBufferOffsetAlignment, 256u);
}
TEST_P(FeatureArchInfoTest_TieredMaxLimits, D3DHighMaxDynamicUniformBuffersPerPipelineLayout) {
const bool isWindowsHighEnd =
gpu_info::IsNvidia(GetParam().adapterProperties.vendorID) && (IsD3D11() || IsD3D12());
DAWN_TEST_UNSUPPORTED_IF(!isWindowsHighEnd);
// High-end windows desktop GPU should report at least 10 even when tiered is enabled
// See crbug.com/440381283
EXPECT_GE(GetSupportedLimits().maxDynamicUniformBuffersPerPipelineLayout, 10u);
}
DAWN_INSTANTIATE_TEST(FeatureArchInfoTest_MaxLimits,
D3D11Backend(),
D3D12Backend(),
MetalBackend(),
OpenGLBackend(),
OpenGLESBackend(),
VulkanBackend(),
WebGPUBackend());
DAWN_INSTANTIATE_TEST(FeatureArchInfoTest_TieredMaxLimits,
D3D11Backend(),
D3D12Backend(),
MetalBackend(),
OpenGLBackend(),
OpenGLESBackend(),
VulkanBackend(),
WebGPUBackend());
} // anonymous namespace
} // namespace dawn